Constantino Carlos Reyes-Aldasoro

Constantino Carlos Reyes-Aldasoro

Senior Lecturer in Biomedical Image Analysis at City, University of London

Sitemap

A list of all the posts and pages found on the site. For you robots out there is an XML version available for digesting as well.

Pages

About me

Posts

Blog Post number 4

less than 1 minute read

This is a sample blog post. Lorem ipsum I can’t remember the rest of lorem ipsum and don’t have an internet connection right now. Testing testing testing this blog post. Blog posts are cool.

Published:

Blog Post number 3

less than 1 minute read

This is a sample blog post. Lorem ipsum I can’t remember the rest of lorem ipsum and don’t have an internet connection right now. Testing testing testing this blog post. Blog posts are cool.

Published:

Blog Post number 2

less than 1 minute read

This is a sample blog post. Lorem ipsum I can’t remember the rest of lorem ipsum and don’t have an internet connection right now. Testing testing testing this blog post. Blog posts are cool.

Published:

Blog Post number 1

less than 1 minute read

This is a sample blog post. Lorem ipsum I can’t remember the rest of lorem ipsum and don’t have an internet connection right now. Testing testing testing this blog post. Blog posts are cool.

Published:

bookchapters

collaborations

contact

gallery

mainoutreach

maintech

news

outreach

preprints

projects

Structure of Biomimetic Materials

A large number of natural and synthetic hydrogels are currently used for tissue engineering and regenerative medicine. Over the last decade, there has been an increasing awareness of the role of material properties of the substrates in guiding cellular behaviour. This has inspired chemists to create a new generation of materials with mechanical properties closed to that of natural occurring biopolymer networks. Recently, the groups of Prof. Alan Rowan (Queens University, Australia) and Prof. Paul Kouwer (Radboud University of Nijmegen, The Netherlands) were able to develop a fully synthetic material that mimics in all aspects the gels prepared from cellular filaments. These synthetics gels are prepared from polyisocyanopeptides (PICs) grafted with oligo(ethylene glycol) chains and share structural features of biopolymers: their helical structure renders the polymer molecules relatively stiff while the interaction between the side chains enable the formation of bundles or fibrils of defined dimensions. The triethylene glycol side chains attached to the polymer backbone render the material thermo-responsive (it will gel upon heating beyond 20 °C and become liquid again upon cooling). Despite being characterized extensively in bulk, the fundamental dynamics and the relation between the macroscopic properties and the microscopic structure at cellular length scales of PIC-based hydrogels remains obscure.

Structure of the PIC polymer/monomer unit and cartoon of the polymer structure showing the helical structure.

Classically, structural characterization of materials is performed with electron microscopy or scanning probe microscopy. Despite the high spatial resolution achievable with these techniques, they are unable to measure dynamics ‘in situ’ and sample preparation can be a laborious process. In contrast, optical microscopy has the potential to unravel the dynamics in complex heterogeneous systems but has been limited to a spatial resolution of ca. 200 nm. In the past 10 years fluorescence imaging has been revolutionized by the successful development of sub-diffraction (super-resolution) microscopy modalities which can achieve resolutions down to tens of nanometers (see Molecular Organization at the Nanoscale).The various possibilities of fluorescence microscopy to probe dynamics and heterogeneities, with molecular resolution, for a wide range of time scales makes it an ideal tool to address many topics of polymer science. In this project we are using STED to image the polymer network at the nanometer scale.

STED image of PIC network.


For more information on PIC-based hydrogels:

  • Kouwer P.H.J., et al. (2013) Responsive biomimetic networks from polyisocyanopeptide hydrogels, Nature, 493, pages 651–655 (article can be found here)
  • Jasper M., et al. (2014) Ultra-responsive soft matter from strain-stiffening hydrogels, Nature Communications, 5, 5808 (article can be found here)
  • Jasper M., et al. (2016) Bundle Formation in Biomimetic Hydrogels, Macromolecules, 17(8), pages 2642–2649 (article can be found here)

Polymer reptation in 3D

Our current theoretical understanding of entangled polymer chain dynamics is based on the reptation model. First proposed by Doi and Edwards, and further expanded by de Gennes, the reptation model assumes that a polymer chain is confined by the surrounding matrix and is therefore forced to move inside an imaginary tube defined by the transient network of entangled neighboring chains. Intuitively this motion resembles that of a snake or worm. The reptation model predicts five dynamical regimes for segment diffusion, summarized in the figure below. These regimes are as follows: (0) sub-segmental processes (“glassy dynamics”) at very short times (microseconds), (I) small motion subject only to chain connectivity, (II) “local reptation”: short-distance motion within the constraints imposed by the surrounding chains (“tube”), (III) “reptation”: diffusive motion along the curvilinear tube over distances larger than the polymer size, and (IV) free diffusion.

Rheology at the micrometer scale

Due to the crucial role of physical cues in regulating cell behaviour, the mechanical properties of hydrogels are a key design parameter in tissue engineering applications. The shear elastic properties of viscoelastic materials are commonly measured by mechanical rheometers. Storage and loss moduli of a material can be measured by application of strain while measuring stress or vice versa. In contrast, recently developed optical micro-rheology techniques use nanometer- or micrometer-sized particles embedded in the material to obtain the viscoelastic response parameters. Thermal or passive micro-rheology for viscoelastic materials is based on an extension of the concepts of Brownian motion of particles in simple liquids. The movement of the embedded particles can be monitored using particle tracking. Initially developed to investigate the rheological properties of uniform complex fluids, particle tracking micro-rheology (PTM) is becoming a popular technique to analyze polymer blends and gels, as well as the deformability and elasticity within cells. However, if the beads locally modify the structure of the gel or are contained in a pore in an inhomogeneous matrix, the bulk rheological properties will not be retrieved. A solution is to use the cross-correlated thermal fluctuation of pairs of tracer particles, ‘two-point micro-rheology’. This method provides a better agreement between micro and macro-rheology, even in complex micro-structured fluids. However, technical constrains limit the wide application of this technique. One of the major limitations of two-point micro-rheology is the reduced number of trajectories that can be used for analysis. During particle tracking micro-rheology, the length of the calculated trajectories is limited by the time spent by the tracers in the field of view (x,y) and depth of focus (z). Consequently, mechanical characterization of complex polymer matrixes at the micrometer scale would benefit greatly of a new method for (fast) tracking in 3D. We are developing a new method for fast tracking of (fluorescent) beads in 3D using a multi-plane wide field microscope. This will allow a better mechanical characterization of soft materials, at the microscale.

Cellular adhesion in 3D matrices

Cells sense physical forces and the mechanical properties of the microenvironment via several distinct mechanisms and cellular components. The first step of cellular adhesion to the ECM occurs via transmembrane heterodimers of the integrin family. Once integrin molecules adhere to the ECM, they are activated and form clusters. As the number of bound molecules increases, some of the focal complexes evolve from small (0.5-1µm in diameter) transient ‘dot-like’ contacts to elongated structures (3-10µm) which couple with actin and associated proteins. The mechanical coupling between the ECM and the cell cytoskeleton is controlled by the dynamics of the focal adhesion complexes (assembly, disassembly and turnover).

Protein-Protein Interactions

Protein-protein interactions (PPIs) are intrinsic to all cellular processes, driving both metabolic and regulatory pathways. Despite the numerous techniques available, detection of transient short-lived PPIs remains challenging4. The main fluorescence microscopic techniques developed for visualizing PPIs in a cellular context are based on Föster resonance energy transfer (FRET) or bimolecular fluorescence complementation (BiFC)5. Both techniques detect the interaction between a pair of labeled molecules. Although highly informative, they require fine positioning of the labels and in the majority of the applications the spatial resolution achieved is limited by the diffraction of light to about 200 nm. More information concerning the use of FRET to detect PPIs can be found at Cellular Signalling).

We have use a single molecule localization based super-resolution technique to detect and map PPIs at the cell membrane. This new variant of PAINT that enables mapping of short-lived transient interactions between cytosolic and membrane-bound proteins inside living mammalian cells, at the nanometer scale. In this method the protein of interest is labeled with a light-controllable fluorescent protein and imaged under TIRF illumination, which leads to the selective activation and subsequent detection of molecules in close proximity with the plasma membrane. Interacting molecules are discriminated using a stringent fitting of the fluorescence signal recorded for every single molecule.

Organoids: models for cell communication

Nowadays, human organoids are becoming a highly promising tool to model organ development, function and especially human diseases in vitro. In general, organoids are miniature, simplified organs that can easily propagate in vitro originating from one or a few cells, typically stem cells.

Single cell manipulation by endoscopy

Nanowire-based endoscopy has attracted interest due to its ability to manipulate cells at the single-cell level with minimal cellular perturbation. High-density, vertically aligned nanowire arrays have been used as an efficient gene delivery system. Despite the high transfection rates, culturing the cells on nanowire arrays might have other influences on the cellular behaviour. For example, stem cells cultured on silicon nanowires show significantly different adhesion, proliferation and differentiation, compared with flat silicon or other control substrates. Furthermore, such arrays are not location-specific and require optimization of the nanowire density and dimension for the different the cell types. In collaboration with the group of Prof. Hiroshi Uji-i we are developing a method to delivery genetic material using a single nanowire. In contrast to the existing methods, this approach can be applied to any cell type and is extremely specific: it can target a single cell and it can deliver the genetic material exactly at the desired position, such as inside of the nucleus, with no damage to the cell. Since gene editing is a stochastic event occurring in only a fraction of the cells, the transfer of genetic material (or proteins) is of crucial importance in genome editing methods, where the nucleases must be efficiently delivered. The duration and magnitude of the nuclease expression are critical parameters for the level of both on-target and off-target nuclease activity. Additionally, the dose of donor template DNA is important to ensure efficient homologous recombination. The proposed method offers the possibility to deliver different molecules at different times, in synchronization with the cell cycle. The lab of Prof. Uji-i is one of the first (and few) groups worldwide to have developed and optimized a novel nanoscopic technique using 1D nanowires, with a diameter of less than 100 nm, for SERS endoscopic studies. It has been already proven by us that the thin diameter and 1D structure of the NW greatly reduces the damage induced to a live cell during probe insertion. Although designed for a different purpose, this nanoprobe is ideal as a starting point to develop a new NW-based gene delivery system.

Principle of nanowire-based gene delivery system.

New Drug delivery systems

In this decade, the pharmacology field has been intensively exploring different approaches to deliver multiple drugs with a single drug nano-carrier, such as liposomes, polymer nanoparticles, and inorganic nanoparticles. The advantage of nanoparticle based drug delivery is the ability to unify pharmacokinetics by simultaneous delivery of multiple drugs to specific target cells.

Ever since first reported in 2001, mesoporous silica nanoparticles (MSNPs) have manifested themselves as highly potential candidates for targeted drug delivery. They owe their popularity to their high drug load capacity, chemical stability, biocompatibility and easy functionalization. Since the diameter of the nanoparticles (100 to 200 nm) is tunable, one can obtain a size suitable for passive targeting through the hyperpermeable tumor vasculature, thereby promoting accumulation of the nanoparticles in tumor tissue due to the enhanced permeability and retention effect (EPR). Additionally, functionalization of the nanoparticles with ligands which have a high affinity for tumor cell specific surface receptors promotes more specific internalization in cancer cells. For example, hyaluronic acid (HA) has been extensively used as a targeting ligand due to its affinity for CD44, a transmembrane glycoprotein receptor that plays a critical role in malignant cell activities and, most importantly, it is overexpressed in many solid tumor cells, in metastasis and cancer stem cells.

Correlative AFM and Fluorescence Microscopy

Biological processes are often carried out in the context of macromolecular assemblies. In addition, arrangements of these complexes can be dynamic, resulting in a heterogeneous ensemble. Single molecule techniques can resolve distinct populations in heterogeneous systems, in contrast to bulk experiments where heterogeneity is averaged out. In turn, mechanistic details of bio-macromolecular interactions can be uncovered. Atomic force microscopy (AFM) is a technique that can generate 3D reconstructions of individual biomolecules and complexes thereof in a label-free fashion, and with ~ nm resolution. To this end a very sharp tip, mounted on a flexible cantilever, scans a sample surface in a raster pattern using a piezo-scanner, while keeping the interaction force between sample and tip constant. In every pixel (x,y) of the scanned area, the z-position is recorded. Consequently, a 3D representation of the surface topography can be reconstructed. An alternative way to study single molecules is by fluorescence microscopy. The molecule of interest is labeled with a fluorescent tag providing high contrast. Emission of the tag after excitation, is detected through an optical system. Due to the wave character of light, the emitted light is spread out on the detector described by the point spread function (PSF) of the optical system. This effect limits the resolution achieved with optical microscopy, referred to as the diffraction limit. However, when the signal of a single molecule is detected, the position of this molecule can be determined by fitting of the recorded fluorescence signal with a mathematical approximation of the PSF such as a two-dimensional Gaussian function. This principle underlies single molecule localization microscopy (SMLM). AFM and SMLM are highly complementary technologies: AFM can provide insight in topographic features at a nanometer resolution while SMLM is sensitive towards specifically labelled molecules in complex samples. Integrated setups combining both technologies can therefore provide orthogonal information at the single-molecule level.

Cell signalling: probes and methods

Cell signaling involves the sensing of an extracellular signal by a cell surface receptor, which then transduces this signal to an intracellular response. Despite the numerous studies performed on signaling pathways and mechanisms, little is known about the initial steps occurring at the plasma membrane: receptor pre-assembly at the molecular level and potential reorganization after ligand activation. Traditionally crystallography is used to investigate receptor multimerization. However, the crystallized state might not represent the biochemically active form due to the harsh preparation conditions and the absence of the cellular environment. Other approaches include macroscopic biochemical or biophysical methods, such as chemical cross-linking, ion-channel gating, immunoprecipitation or binding assays. Nowadays, established fluorescence imaging and spectroscopic techniques offer a versatile toolbox to study membrane receptor organization in (living) cells.

In the lab we are using fluorescence fluctuation spectroscopy to quantify physicochemical processes (mobility, binding affinity, stoichiometry, absolute concentration) occurring on a micro-to-millisecond time scale. Fluorescence experiments down to picoseconds are also commonly possible with methods such as time-correlated single photon counting (TCSPC), that allow, e.g., measuring fluorescence lifetimes and molecular tumbling. Additionally, spatially resolved microscopy with high temporal resolution also has clear benefits. For example, combined with confocal laser scanning microscopy (LSM), TCSPC allows protein-protein interactions (PPIs) to be imaged via Förster resonance energy transfer (FRET) based fluorescence lifetime imaging microscopy (FLIM). Imaging based FCS methods such as raster (RICS), number and brightness analysis (N&B) or (spatio-) temporal image correlation spectroscopy [(S)TICS] combine the quantitative analytical power of fluctuation methods with spatial information to map, among many other things, mobility and stoichiometry inside living systems. Simultaneous dual-color fluorescence imaging is possible when fast alternating excitation (alias pulsed interleaved excitation, PIE) is employed. PIE renders analysis of dual-color point FCS experiments considerably more straightforward. The combination of PIE with fluctuation imaging (PIE-FI) allows extracting the maximum amount of molecular information (mobility, stoichiometry, interactions…) from each species present in dual-color LSM images.

PIE (a), PIE-FI (b) and subsequent analyses, based on spatial/temporal auto-/cross-correlation or fluorescence lifetimes, which allow to extract the maximum amount of information of the molecules present in the imaged structure.

For more information on these methods:

  • Hendrix J., Lamb D.C. (2014) Implementation and Application of Pulsed Interleaved Excitation for Dual-Color FCS and RICS. In: Engelborghs Y., Visser A. (eds) Fluorescence Spectroscopy and Microscopy. Methods in Molecular Biology (Methods and Protocols), vol 1076. Humana Press, Totowa, NJ (chapter can be found here)
  • Hendrix J., Schrimpf W., Höller M., Lamb D.C. (2013) Pulsed Interleaved Excitation Fluctuation Imaging, Biophysical Journal, 105(4), 848-861 (article can be found here)

Imaging single HIV virions

Viruses are simple agents exhibiting complex reproductive mechanisms. Decades of research have provided crucial basic insights, antiviral medication and moderately successful gene therapy trials. The most infectious viral particle is, however, not always the most abundant one in a population, questioning the utility of classic ensemble-averaging virology. Indeed, viral replication is often not particularly efficient, prone to errors or containing parallel routes. In collaboration with Prof. Zeger Debeyser (KU Leuven) and Prof Hendrix (UHasselt) we have applied different single-molecule sensitive fluorescence methods to investigate viruses, one-by-one. While this collaboration is still ongoing, there is already several publications that show-case the potential of imaging single virions.

publications

Telecommunications and Internet in the future society: myths and realities

Abstract


The Internet boom has led to the image of a global village, a totally communicated planet where national borders have no meaning and where the Internet and the information highway have become the path to solve any problem. Technological and scientific research has been conducted in order to exchange information through the telephone lines, which provide a solid platform for the new services related to the Internet such as E-mail, electronic commerce and Internet telephony. Non-technical issues should be taken into account in order to notice that not all the people an Earth are aware of the Internet and the new technological society. Many persons are not educated for those new technologies that are out of their reach. Indeed, even if they should be able to access a specific communications network, the new services would not be affordable for them economically and/or intellectually. This article presents indicators that magnify the problem: the gap between the rich and the poor, the educated and non-educated, the Internet-literate and illiterate. Reflections on several important issues are presented, and ultimately, the global village is questioned as a reality or just a new myth.

Published in IEEE International Symposium on Computers and Communications, 1999

Download here

Volumetric texture description and discriminant feature selection for MRI

Abstract


This paper considers the problem of classification of Magnetic Resonance Images using 2D and 3D texture measures. Joint statistics such as co-occurrence matrices are common for analysing texture in 2D since they are simple and effective to implement. However, the computational complexity can be prohibitive especially in 3D. In this work, we develop a texture classification strategy by a sub-band filtering technique that can be extended to 3D. We further propose a feature selection technique based on the Bhattacharyya distance measure that reduces the number of features required for the classification by selecting a set of discriminant features conditioned on a set training texture samples. We describe and illustrate the methodology by quantitatively analysing a series of images: 2D synthetic phantom, 2D natural textures, and MRI of human knees.”

Published in Information Processing in Medical Imaging, 2003

Volumetric texture segmentation by discriminant feature selection and multiresolution classification

Abstract


In this paper, a multiresolution volumetric texture segmentation (M-VTS) algorithm is presented. The method extracts textural measurements from the Fourier domain of the data via subband filtering using an orientation pyramid (Wilson and Spann, 1988). A novel Bhattacharyya space, based on the Bhattacharyya distance, is proposed for selecting the most discriminant measurements and producing a compact feature space. An oct tree is built of the multivariate features space and a chosen level at a lower spatial resolution is first classified. The classified voxel labels are then projected to lower levels of the tree where a boundary refinement procedure is performed with a three-dimensional (3-D) equivalent of butterfly filters. The algorithm was tested with 3-D artificial data and three magnetic resonance imaging sets of human knees with encouraging results. The regions segmented from the knees correspond to anatomical structures that can be used as a starting point for other measurements such as cartilage extraction.”

Published in IEEE Transactions on Medical Imaging, 2007

Estimation of apparent tumor vascular permeability from multiphoton fluorescence microscopic images of P22 rat sarcomas in vivo

Abstract


OBJECTIVE: To develop an image processing-based method to quantify the rate of extravasation of fluorescent contrast agents from tumor microvessels, and to investigate the effect of the tumor vascular disrupting agent combretastatin A-4-P (CA-4-P) on apparent tumor vascular permeability to 40 kDa fluorescein isothiocyanate (FITC) labeled dextra”

Published in Microcirculation, 2008

Measuring cellular migration with image processing

Abstract


An image-processing algorithm for analysis of migration of vascular endothelial cells in culture is presented. The algorithm correctly detected the cellular regions on either side of an artificial ‘wound’ made by dragging a sterile pipette tip across the monolayer of cells (scratch wound assay). Frequency filtering and mathematical morphology were used to approximate the boundaries of cellular regions. This allowed measurement of the distance between the regions, and therefore the migration rates, regardless of the orientation of the wound and even in cases where the cells were sparse and not tightly packed.”

Published in Electronics Letters, 2008

Download here

Retrospective shading correction algorithm based on signal envelope estimation

Abstract


An additive retrospective non-parametric algorithm for the correction of the inhomogeneous intensity background of images, commonly known as shading, is presented. The algorithm assumed that an original unbiased image was corrupted by slowly-varying shading that could be estimated from the signal envelope in a process analogous to amplitude modulation detection. Unlike other filtering algorithms, the algorithm did not require pre-processing, parameter setting, user interaction, computationally intensive optimisation algorithms nor a restriction in size of the objects of interest relative to the scale of background variations. The algorithm provided satisfactory results for artificial and microscopical images.”

Published in Electronics Letters, 2009

Download here

Application of intravital microscopy in studies of tumor microcirculation

Abstract


To grow and progress, solid tumors develop a vascular network through co-option and angiogenesis that is characterized by multiple structural and functional abnormalities, which negatively influence therapeutic outcome through direct and indirect mechanisms. As such, the morphology and function of tumor blood vessels, plus their response to different treatments, are a vital and active area of biological research. Intravital microscopy (IVM) has played a key role in studies of tumor angiogenesis, and ongoing developments in molecular probes, imaging techniques, and postimage analysis methods have ensured its continued and widespread use. In this review we discuss some of the primary advantages and disadvantages of IVM approaches and describe recent technological advances in optical microscopy (e.g., confocal microscopy, multiphoton microscopy, hyperspectral imaging, and optical coherence tomography) with examples of their application to studies of tumor angiogenesis.”

Published in Journal of Biomedical Optics, 2010

Download here

An automatic algorithm for the segmentation and morphological analysis of microvessels in immunostained histological tumour sections

“<h2> Abstract </h2>” “A fully automatic segmentation and morphological analysis algorithm for the analysis of microvessels from CD31 immunostained histological tumour sections is presented. Development of the algorithm exploited the distinctive hues of stained vascular endothelial cells, cell nuclei and background, to provide the seeds for a ‘region-growing’ method for object segmentation in the 3D hue, saturation, value (HSV) colour model. The segmented objects, identified as microvessels by CD31 immunostaining, were post-processed with three morphological tasks: joining separate objects that were likely to belong to a single vessel, closing objects that had a narrow gap around their periphery, and splitting objects with multiple lumina into individual vessels. The automatic segmentation was validated against a hand-segmented set of 44 images from three different SW1222 human colorectal carcinomas xenografted into mice. 96.3 ± 0.9\% of pixels were found to be correctly classified. Automated segmentation was carried out on a further 53 images from three histologically distinct mouse fibrosarcomas (MFs) for morphological comparison with the SW1222 tumours. Four morphometric measurements were calculated for each segmented vessel: vascular area (VA), ratio of lumen area to vascular area (lu/VA), eccentricity (e), and roundness (ro). In addition, the total vascular area relative to tumour tissue area (rVA) was calculated. lu/VA, e and ro were found to be significantly smaller in MF tumours than in SW1222 tumours (p {\textless} 0.05; unpaired t-test). The algorithm is available through the website http://www.caiman.org.uk where images can be uploaded, processed and sent back to users. The output from CAIMAN consists of the original image with boundaries of segmented vessels overlaid, the calculated parameters and a Matlab file, which contains the segmentation that the user can use to derive further results.”

Published in Journal of Microscopy, 2011

Download here

Activation of hypoxia-inducible factor-1α (Hif-1α) delays inflammation resolution by reducing neutrophil apoptosis and reverse migration in a zebrafish inflammation model

Abstract


The oxygen-sensing transcription factor hypoxia-inducible factor-1α (HIF-1α) plays a critical role in the regulation of myeloid cell function. The mechanisms of regulation are not well understood, nor are the phenotypic consequences of HIF modulation in the context of neutrophilic inflammation. Species conservation across higher metazoans enables the use of the genetically tractable and transparent zebrafish (Danio rerio) embryo to study in vivo resolution of the inflammatory response. Using both a pharmacologic approach known to lead to stabilization of HIF-1α, and selective genetic manipulation of zebrafish HIF-1α homologs, we sought to determine the roles of HIF-1α in inflammation resolution. Both approaches reveal that activated Hif-1α delays resolution of inflammation after tail transection in zebrafish larvae. This delay can be replicated by neutrophil-specific Hif activation and is a consequence of both reduced neutrophil apoptosis and increased retention of neutrophils at the site of tissue injury. Hif-activated neutrophils continue to patrol the injury site during the resolution phase, when neutrophils would normally migrate away. Site-directed mutagenesis of Hif in vivo reveals that hydroxylation of Hif-1α by prolyl hydroxylases critically regulates the Hif pathway in zebrafish neutrophils. Our data demonstrate that Hif-1α regulates neutrophil function in complex ways during inflammation resolution in vivo.”

Published in Blood, 2011

CAIMAN an online algorithm repository for Cancer Image Analysis

Abstract


CAIMAN (CAncer IMage ANalysis: http://www.caiman.org.uk) is an online algorithm repository that provides specifically designed algorithms to analyse the images produced by experiments relevant to Cancer Research and Life Sciences, especially vascular biology. CAIMAN is accessed through a user-friendly website where researchers can upload their images and the results are returned by email. CAIMAN does not intend to replace more sophisticated software solutions such as ImageJ, Matlab, or commercial packages, but it will provide a first stop where any researcher can upload images and can obtain quantitative results without having to do any programming at all.”

Published in Computer Methods and Programs in Biomedicine, 2011

Microflow of fluorescently labelled red blood cells in tumours expressing single isoforms of VEGF and their response to vascular targeting agents

Abstract


In this work we studied the functional differences between the microcirculation of murine tumours that express only single isoforms of vascular endothelial growth factor-A (VEGF), namely VEGF120 and VEGF188, and the effect of VEGF receptor tyrosine kinase (VEGF-R TK) inhibition on their functional response to the vascular disrupting agent, combretastatin A-4 phosphate (CA-4-P), using measurement of red blood cell (RBC) velocity by a ‘keyhole’ tracking algorithm. RBC velocities in VEGF188 tumours were unaffected by chronic treatment with a VEGF-R tyrosine kinase inhibitor, SU5416, whereas RBC velocities in VEGF120 tumours were significantly increased compared to control VEGF120 tumours. This effect was accompanied by a reduced tumour vascularisation. Pre-treatment of VEGF120 tumours with SU5416 made them much more resistant to CA-4-P treatment, with a RBC velocity response that was very similar to that of the more mature vasculature of the VEGF188 tumours. This study shows that vascular normalisation following anti-angiogenic treatment with a VEGF-R tyrosine kinase inhibitor reduced the response of a previously sensitive tumour line to CA-4-P.

Published in Medical Engineering & Physics, 2011

Vascular effects dominate solid tumor response to treatment with combretastatin A-4-phosphate

Abstract


Vascular-targeted therapeutics are increasingly used in the clinic. However, less is known about the direct response of tumor cells to these agents. We have developed a combretastatin-A-4-phosphate (CA4P) resistant variant of SW1222 human colorectal carcinoma cells to examine the relative importance of vascular versus tumor cell targeting in the ultimate treatment response. SW1222(Res) cells were generated through exposure of wild-type cells (SW1222(WT) ) to increasing CA4P concentrations in vitro. Increased resistance was confirmed through analyses of cell viability, apoptosis and multidrug-resistance (MDR) protein expression. In vivo, comparative studies examined tumor cell necrosis, apoptosis, vessel morphology and functional vascular end-points following treatment with CA4P (single 100 mg/kg dose). Tumor response to repeated CA4P dosing (50 mg/kg/day, 5 days/week for 2 weeks) was examined through growth measurement, and ultimate tumor cell survival was studied by ex vivo clonogenic assay. In vitro, SW1222(Res) cells showed reduced CA4P sensitivity, enhanced MDR protein expression and a reduced apoptotic index. In vivo, CA4P induced significantly lower apoptotic cell death in SW1222(Res) versus SW1222(WT) tumors indicating maintenance of resistance characteristics. However, CA4P-induced tumor necrosis was equivalent in both lines. Similarly, rapid CA4P-mediated vessel disruption and blood flow shut-down were observed in both lines. Cell surviving fraction was comparable in the two tumor types following single dose CA4P and SW1222(Res) tumors were at least as sensitive as SW1222(WT) tumors to repeated dosing. Despite tumor cell resistance to CA4P, SW1222(Res) response in vivo was not impaired, strongly supporting the view that vascular damage dominates the therapeutic response to this agent.”

Published in International Journal of Cancer, 2011

Download here

The neutrophil’s eye-view inference and visualisation of the chemoattractant field driving cell chemotaxis in vivo

Abstract


As we begin to understand the signals that drive chemotaxis in vivo, it is becoming clear that there is a complex interplay of chemotactic factors, which changes over time as the inflammatory response evolves. New animal models such as transgenic lines of zebrafish, which are near transparent and where the neutrophils express a green fluorescent protein, have the potential to greatly increase our understanding of the chemotactic process under conditions of wounding and infection from video microscopy data. Measurement of the chemoattractants over space (and their evolution over time) is a key objective for understanding the signals driving neutrophil chemotaxis. However, it is not possible to measure and visualise the most important contributors to in vivo chemotaxis, and in fact the understanding of the main contributors at any particular time is incomplete. The key insight that we make in this investigation is that the neutrophils themselves are sensing the underlying field that is driving their action and we can use the observations of neutrophil movement to infer the hidden net chemoattractant field by use of a novel computational framework. We apply the methodology to multiple in vivo neutrophil recruitment data sets to demonstrate this new technique and find that the method provides consistent estimates of the chemoattractant field across the majority of experiments. The framework that we derive represents an important new methodology for cell biologists investigating the signalling processes driving cell chemotaxis, which we label the neutrophils eye-view of the chemoattractant field.”

Published in PLoS ONE, 2012

Download here

Local affine texture tracking for serial registration of zebrafish images

Abstract


The aim of this work is to register serial in-vivo confocal microscopy images of zebrafish to enable accurate cell tracking on corresponding fluorescence images. The following problem arises during acquisition; the zebrafish tail may undergo a series of movement and non-linear deformations, which if not corrected, adds to the motion of leukocytes being tracked. This makes it difficult to accurately assess their motion. We developed a correlation based, local affine image matching method, which is well suited to the textured DIC images of the anatomy of the zebrafish and enables accurate and efficient tracking of image regions over successive frames. Experimental results of the serial registration and tracking demonstrate its accuracy in estimating local affine motions in zebrafish sequences.

Published in ISBI, 2012

Download here

Neutrophil-delivered myeloperoxidase dampens the hydrogen peroxide burst after tissue wounding in zebrafish

Abstract


Prompt neutrophil arrival is critical for host defense immediately after injury [1-3]. Following wounding, a hydrogen peroxide (H(2)O(2)) burst generated in injured tissues is the earliest known leukocyte chemoattractant [4]. Generating this tissue-scale H(2)O(2) gradient uses dual oxidase [4] and neutrophils sense H(2)O(2) by a mechanism involving the LYN Src-family kinase [5], but the molecular mechanisms responsible for H(2)O(2) clearance are unknown [6]. Neutrophils carry abundant amounts of myeloperoxidase, an enzyme catalyzing an H(2)O(2)-consuming reaction [7, 8]. We hypothesized that this neutrophil-delivered myeloperoxidase downregulates the high tissue H(2)O(2) concentrations that follow wounding. This was tested in zebrafish using simultaneous fluorophore-based imaging of H(2)O(2) concentrations and leukocytes [4, 9-11] and a new neutrophil-replete but myeloperoxidase-deficient mutant (durif). Leukocyte-depleted zebrafish had an abnormally sustained wound H(2)O(2) burst, indicating that leukocytes themselves were required for H(2)O(2) downregulation. Myeloperoxidase-deficient zebrafish also had abnormally sustained high wound H(2)O(2) concentrations despite similar numbers of arriving neutrophils. A local H(2)O(2)/myeloperoxidase interaction within wound-recruited neutrophils was demonstrated. These data demonstrate that leukocyte-delivered myeloperoxidase cell-autonomously downregulates tissue-generated wound H(2)O(2) gradients in vivo, defining a new requirement for myeloperoxidase during inflammation. Durif provides a new animal model of myeloperoxidase deficiency closely phenocopying the prevalent human disorder [7, 12, 13], offering unique possibilities for investigating its clinical consequences.”

Published in Current Biology, 2012

Download here

Online chromatic and scale-space microvessel-tracing analysis for transmitted light optical images

Abstract


Limited contrast in transmitted light optical images from intravital microscopy is problematic for analysing tumour vascular morphology. Moreover, in some cases, changes in vasculature are visible to a human observer but are not easy to quantify. In this paper two online algorithms are presented: scale-space vessel tracing and chromatic decomposition for analysis of the vasculature of SW1222 human colorectal carcinoma xenografts growing in dorsal skin-fold ““window”” chambers in mice. Transmitted light optical images of tumours were obtained from mice treated with the tumour vascular disrupting agent, combretastatin-A-4-phosphate (CA4P), or saline. The tracing algorithm was validated against hand-traced vessels with accurate results. The measurements extracted with the algorithms confirmed the known effects of CA4P on tumour vascular topology. Furthermore, changes in the chromaticity suggest a deoxygenation of the blood with a recovery to initial levels in CA4P-treated tumours relative to the controls. The algorithms can be freely applied to other studies through the CAIMAN website (CAncer IMage ANalysis: http://www.caiman.org.uk).”

Published in Microvascular Research, 2012

Repelled from the wound, or randomly dispersed? Reverse migration behaviour of neutrophils characterized by dynamic modelling

Abstract


Following neutralization of infectious threats, neutrophils must be removed from inflammatory sites for normal tissue function to be restored. Recently, a new paradigm has emerged, in which viable neutrophils migrate away from inflammatory sites by a process best described as reverse migration. It has generally been assumed that this process is the mirror image of chemotaxis, where neutrophils are drawn into the areas of infection or tissue damage by gradients of chemotactic cues. Indeed, efforts are underway to identify cues that drive neutrophils away by the reverse process, fugetaxis. By using photoconvertible pigments expressed in neutrophils in transparent zebrafish larvae, we were able to image the position of each neutrophil during inflammation resolution in vivo. These neutrophil coordinates were analysed within a dynamic modelling framework, using different forms of the drift-diffusion equation with model selection and parameter estimation based on approximate Bayesian computation. This analysis revealed the experimental data were best fitted by a model incorporating a diffusion term but no drift term-where the presence of drift would indicate fugetaxis. This result, for the first time, provides rigorous data-driven evidence that reverse migration of neutrophils in vivo is not a form of fugetaxis, but rather a stochastic redistribution.”

Published in Journal of the Royal Society Interface, 2012

Download here

PhagoSight an open-source MATLAB® package for the analysis of fluorescent neutrophil and macrophage migration in a zebrafish model

Abstract


Neutrophil migration in zebrafish larvae is increasingly used as a model to study the response of these leukocytes to different determinants of the cellular inflammatory response. However, it remains challenging to extract comprehensive information describing the behaviour of neutrophils from the multi-dimensional data sets acquired with widefield or confocal microscopes. Here, we describe PhagoSight, an open-source software package for the segmentation, tracking and visualisation of migrating phagocytes in three dimensions. The algorithms in PhagoSight extract a large number of measurements that summarise the behaviour of neutrophils, but that could potentially be applied to any moving fluorescent cells. To derive a useful panel of variables quantifying aspects of neutrophil migratory behaviour, and to demonstrate the utility of PhagoSight, we evaluated changes in the volume of migrating neutrophils. Cell volume increased as neutrophils migrated towards the wound region of injured zebrafish. PhagoSight is openly available as MATLAB® m-files under the GNU General Public License. Synthetic data sets and a comprehensive user manual are available from http://www.phagosight.org.”

Published in PLoS ONE, 2013

Download here

Cxcl8 (IL-8) mediates neutrophil recruitment and behavior in the zebrafish inflammatory response

Abstract


Neutrophils play a pivotal role in the innate immune response. The small cytokine CXCL8 (also known as IL-8) is known to be one of the most potent chemoattractant molecules that, among several other functions, is responsible for guiding neutrophils through the tissue matrix until they reach sites of injury. Unlike mice and rats that lack a CXCL8 homolog, zebrafish has two distinct CXCL8 homologs: Cxcl8-l1 and Cxcl8-l2. Cxcl8-l1 is known to be upregulated under inflammatory conditions caused by bacterial or chemical insult but until now the role of Cxcl8s in neutrophil recruitment has not been studied. In this study we show that both Cxcl8 genes are upregulated in response to an acute inflammatory stimulus, and that both are crucial for normal neutrophil recruitment to the wound and normal resolution of inflammation. Additionally, we have analyzed neutrophil migratory behavior through tissues to the site of injury in vivo, using open-access phagocyte tracking software PhagoSight. Surprisingly, we observed that in the absence of these chemokines, the speed of the neutrophils migrating to the wound was significantly increased in comparison with control neutrophils, although the directionality was not affected. Our analysis suggests that zebrafish may possess a subpopulation of neutrophils whose recruitment to inflamed areas occurs independently of Cxcl8 chemokines. Moreover, we report that Cxcl8-l2 signaled through Cxcr2 for inducing neutrophil recruitment. Our study, therefore, confirms the zebrafish as an excellent in vivo model to shed light on the roles of CXCL8 in neutrophil biology.”

Published in Journal of Immunology, 2013

Download here

Cancer Image Analysis

Abstract


Welcome to the first in a series of articles that will highlight the role of image analysis in oncology. Cancer Image Analysis (CIA) is concerned with the extraction and manipulation of useful information from oncological images, and therefore it is closely related to Cancer Imaging per se and can be seen as complementary step in the process towards diagnosis, screening, drug testing or assessing treatments. CIA is a very wide field of research, not only due to the wide range of cancer-related images, from MRI to histology to optical images, but also because image analysis has inherited many techniques from the fields of Statistics, Pattern Recognition and Computer Vision. This series will show the potential of image analysis as applied to cancer with the ultimate objective of fostering an interdisciplinary cross-fertilization that will bring benefits to clinicians, scientists and ultimately the patient.

Published in Oncology News, 2013

Download here

Influence of soluble or matrix-bound isoforms of vascular endothelial growth factor-A on tumor response to vascular-targeted strategies

Abstract


Antiangiogenic therapy based on blocking the actions of vascular endothelial growth factor-A (VEGF) can lead to ““normalization”” of blood vessels in both animal and human tumors. Differential expression of VEGF isoforms affects tumor vascular maturity, which could influence the normalization process and response to subsequent treatment. Fibrosarcoma cells expressing only VEGF120 or VEGF188 isoforms were implanted either subcutaneously (s.c.) or in dorsal skin-fold ““window”” chambers in SCID mice. VEGF120 was associated with vascular fragility and hemorrhage. Tumor-bearing mice were treated with repeat doses of SU5416, an indolinone receptor tyrosine kinase inhibitor with activity against VEGFR-2 and proven preclinical ability to induce tumor vascular normalization. SU5416 reduced vascularization in s.c. implants of both VEGF120 and VEGF188 tumors. However, in the window chamber, SU5416 treatment increased red cell velocity in VEGF120 (representing vascular normalization) but not VEGF188 tumors. SU5416 treatment had no effect on growth or necrosis levels in either tumor type but tended to counteract the increase in interstitial fluid pressure seen with growth of VEGF120 tumors. SU5416 pretreatment resulted in the normally fragile blood vessels in VEGF120-expressing tumors becoming resistant to the vascular damaging effects of the tubulin-binding vascular disrupting agent (VDA), combretastatin A4 3-O-phosphate (CA4P). Thus, vascular normalization induced by antiangiogenic treatment can reduce the efficacy of subsequent VDA treatment. Expression of VEGF120 made tumors particularly susceptible to vascular normalization by SU5416, which in turn made them resistant to CA4P. Therefore, VEGF isoform expression may be useful for predicting response to both antiangiogenic and vascular-disrupting therapy.”

Published in International Journal of Cancer, 2013

Tumour cells expressing single VEGF isoforms display distinct growth, survival and migration characteristics

Abstract


Vascular endothelial growth factor-A (VEGF) is produced by most cancer cells as multiple isoforms, which display distinct biological activities. VEGF plays an undisputed role in tumour growth, vascularisation and metastasis; nevertheless the functions of individual isoforms in these processes remain poorly understood. We investigated the effects of three main murine isoforms (VEGF188, 164 and 120) on tumour cell behaviour, using a panel of fibrosarcoma cells we developed that express them individually under endogenous promoter control. Fibrosarcomas expressing only VEGF188 (fs188) or wild type controls (fswt) were typically mesenchymal, formed ruffles and displayed strong matrix-binding activity. VEGF164- and VEGF120-producing cells (fs164 and fs120 respectively) were less typically mesenchymal, lacked ruffles but formed abundant cell-cell contacts. On 3D collagen, fs188 cells remained mesenchymal while fs164 and fs120 cells adopted rounded/amoeboid and a mix of rounded and elongated morphologies respectively. Consistent with their mesenchymal characteristics, fs188 cells migrated significantly faster than fs164 or fs120 cells on 2D surfaces while contractility inhibitors accelerated fs164 and fs120 cell migration. VEGF164/VEGF120 expression correlated with faster proliferation rates and lower levels of spontaneous apoptosis than VEGF188 expression. Nevertheless, VEGF188 was associated with constitutively active/phosphorylated AKT, ERK1/2 and Stat3 proteins. Differences in proliferation rates and apoptosis could be explained by defective signalling downstream of pAKT to FOXO and GSK3 in fs188 and fswt cells, which also correlated with p27/p21 cyclin-dependent kinase inhibitor over-expression. All cells expressed tyrosine kinase VEGF receptors, but these were not active/activatable suggesting that inherent differences between the cell lines are governed by endogenous VEGF isoform expression through complex interactions that are independent of tyrosine kinase receptor activation. VEGF isoforms are emerging as potential biomarkers for anti-VEGF therapies. Our results reveal novel roles of individual isoforms associated with cancer growth and metastasis and highlight the importance of understanding their diverse actions.”

Published in PLoS ONE, 2014

Download here

Image-based investigation into the primary fabric of stress-transmitting particles in sand

Abstract


Understanding the stress distribution amongst the constituent grains is fundamental to predict the response of soil and advance science-based, rather than purely empirical, constitutive models. Photoelastic experiments and discrete element method simulations have provided evidence that, upon loading, discrete force chains form in granular materials. These force chains are made up of particles transmitting relatively large stresses and they are aligned in the direction of the major principal stress. A few qualitative studies have identified the presence of these force chains in sands but direct measurements of force chains have not been previously documented and tracking stress transmission in assemblies of real soil grains remains a challenging task. The present study makes use of three dimensional micro CT images to investigate the evolution of the internal topology of a sand subjected to triaxial compression loading. The analysis of the contact normal and branch vector orientations has shown the realignment of the contact normals in the direction of the major principal stress as a clear indication of the formation of force chains in the post-peak regime. Here the extent of the non-colinearity of the branch and contact normal vectors is explored. Using the micro CT data contact force networks within and outside of shear bands are compared.

Published in Geomechanics from Micro to Macro, 2014

Download here

Automatic segmentation of focal adhesions from mouse embryonic fibroblasts

Abstract


This work describes an automatic algorithm for the segmentation and quantification of focal adhesions from mouse embryonic fibroblasts. The main challenges solved by this algorithm are: the variability of the intensity of the focal adhesions, the detection of an outer ring, which distinguishes the cell periphery responsible for the cell migration, and the quantification of the characteristics of the focal adhesions. The algorithm detects maximal regions through gradients and uses a region-growing algorithm limited by intensity-based edges. The outer ring is calculated based on the average radial intensity from an extended centroid of the cell. Finally, traditional morphological characteristics are obtained to distinguish between two groups of cells. Two of the measurements employed showed statistical difference between two groups of cells.

Published in ISBI, 2015

Download here

Regional variation in arterial stiffening and dysfunction in Western diet-induced obesity

Abstract


Increased central vascular stiffening, assessed in vivo by determination of pulse wave velocity (PWV), is an independent predictor of cardiovascular event risk. Recent evidence demonstrates that accelerated aortic stiffening occurs in obesity; however, little is known regarding stiffening of other disease-relevant arteries or whether regional variation in arterial stiffening occurs in this setting. We addressed this gap in knowledge by assessing femoral PWV in vivo in conjunction with ex vivo analyses of femoral and coronary structure and function in a mouse model of Western diet (WD; high-fat/high-sugar)-induced obesity and insulin resistance. WD feeding resulted in increased femoral PWV in vivo. Ex vivo analysis of femoral arteries revealed a leftward shift in the strain-stress relationship, increased modulus of elasticity, and decreased compliance indicative of increased stiffness following WD feeding. Confocal and multiphoton fluorescence microscopy revealed increased femoral stiffness involving decreased elastin/collagen ratio in conjunction with increased femoral transforming growth factor-β (TGF-β) content in WD-fed mice. Further analysis of the femoral internal elastic lamina (IEL) revealed a significant reduction in the number and size of fenestrae with WD feeding. Coronary artery stiffness and structure was unchanged by WD feeding. Functionally, femoral, but not coronary, arteries exhibited endothelial dysfunction, whereas coronary arteries exhibited increased vasoconstrictor responsiveness not present in femoral arteries. Taken together, our data highlight important regional variations in the development of arterial stiffness and dysfunction associated with WD feeding. Furthermore, our results suggest TGF-β signaling and IEL fenestrae remodeling as potential contributors to femoral artery stiffening in obesity.”

Published in American Journal of Physiology. Heart and Circulatory Physiology, 2015

Continuous representation of tumor microvessel density and detection of angiogenic hotspots in histological whole-slide images

“<h2> Abstract </h2>” “Blood vessels in solid tumors are not randomly distributed, but are clustered in angiogenic hotspots. Tumor microvessel density (MVD) within these hotspots correlates with patient survival and is widely used both in diagnostic routine and in clinical trials. Still, these hotspots are usually subjectively defined. There is no unbiased, continuous and explicit representation of tumor vessel distribution in histological whole slide images. This shortcoming distorts angiogenesis measurements and may account for ambiguous results in the literature. In the present study, we describe and evaluate a new method that eliminates this bias and makes angiogenesis quantification more objective and more efficient. Our approach involves automatic slide scanning, automatic image analysis and spatial statistical analysis. By comparing a continuous MVD function of the actual sample to random point patterns, we introduce an objective criterion for hotspot detection: An angiogenic hotspot is defined as a clustering of blood vessels that is very unlikely to occur randomly. We evaluate the proposed method in N=11 images of human colorectal carcinoma samples and compare the results to a blinded human observer. For the first time, we demonstrate the existence of statistically significant hotspots in tumor images and provide a tool to accurately detect these hotspots.”

Published in Oncotarget, 2015

Download here

Whole cell tracking through the optimal control of geometric evolution laws

Abstract


Cell tracking algorithms which automate and systematise the analysis of time lapse image data sets of cells are an indispensable tool in the modelling and understanding of cellular phenomena. In this study we present a theoretical framework and an algorithm for whole cell tracking. Within this work we consider that “tracking” is equivalent to a dynamic reconstruction of the whole cell data (morphologies) from static image data sets. The novelty of our work is that the tracking algorithm is driven by a model for the motion of the cell. This model may be regarded as a simplification of a recently developed physically meaningful model for cell motility. The resulting problem is the optimal control of a geometric evolution law and we discuss the formulation and numerical approximation of the optimal control problem. The overall goal of this work is to design a framework for cell tracking within which the recovered data reflects the physics of the forward model. A number of numerical simulations are presented that illustrate the applicability of our approach.”

Published in Journal of Computational Physics, 2015

Download here

Arterial Stiffening in Western Diet-Fed Mice Is Associated with Increased Vascular Elastin, Transforming Growth Factor-β, and Plasma Neuraminidase

Abstract


Consumption of excess fat and carbohydrate (Western diet, WD) is associated with alterations in the structural characteristics of blood vessels. This vascular remodeling contributes to the development of cardiovascular disease, particularly as it affects conduit and resistance arteries. Vascular remodeling is often associated with changes in the elastin-rich internal elastic lamina (IEL) and the activation of transforming growth factor (TGF)-β. In addition, obesity and type II diabetes have been associated with increased serum neuraminidase, an enzyme known to increase TGF-β cellular output. Therefore, we hypothesized that WD-feeding would induce structural modifications to the IEL of mesenteric resistance arteries in mice, and that these changes would be associated with increased levels of circulating neuraminidase and the up-regulation of elastin and TGF-β in the arterial wall. To test this hypothesis, a WD, high in fat and sugar, was used to induce obesity in mice, and the effect of this diet on the structure of mesenteric resistance arteries was investigated. 4-week old, Post-weaning mice were fed either a normal diet (ND) or WD for 16 weeks. Mechanically, arteries from WD-fed mice were stiffer and less distensible, with marginally increased wall stress for a given strain, and a significantly increased Young’s modulus of elasticity. Structurally, the wall cross-sectional area and the number of fenestrae found in the internal elastic lamina (IEL) of mesenteric arteries from mice fed a WD were significantly smaller than those of arteries from the ND-fed mice. There was also a significant increase in the volume of elastin, but not collagen in arteries from the WD cohort. Plasma levels of neuraminidase and the amount of TGF-β in mesenteric arteries were elevated in mice fed a WD, while ex vivo, cultured vascular smooth muscle cells exposed to neuraminidase secreted greater amounts of tropoelastin and TGF-β than those exposed to vehicle. These data suggest that consumption of a diet high in fat and sugar causes stiffening of the vascular wall in resistance arteries through a process that may involve increased neuraminidase and TGF-β activity, elevated production of elastin, and a reduction in the size and number of fenestrae in the arterial IEL.”

Published in Frontiers in Physiology, 2016

Download here

Mapping tumour tissue: quantitative maps of histological whole slide images

Abstract


Immunohistochemistry (IHC) is the standard method to assess tumour tissue on a micro-scopic scale. IHC selectively highlights microscopic structures in the tissue and yields quantitative information that can be used to answer questions like:“How many immune cells are present in a given tumour?”,“How many tumour cells are actively proliferating?”, or “How many blood vessels are present in the tumour?”. These questions are addressed by histopathologists who visually observe regions of immunostained slides of tumour tissue and count structures of interest, for instance, cells or blood vessels. In the clinic, this quantitative information can be then used to estimate the prognosis of a patient. For example, the number of blood vessels in tumour tissue is a prognostic factor for colorectal cancer patients [1]. Pathologists combine the excellent human vision and pattern recognition skills of the brain with an extensive training in tissue observation. Traditionally, pathologists use only a microscope to identify and assess structures of interest manually. However, the limitations of manual procedures are evident; besides the possibility of human error, the dimensions of tissue slides in high magnification are huge and it is not feasible to view the whole slide nor manually visualise or count any objects of interest. Therefore, microscopic structures such as blood vessels are only quantified in a small fraction of the entire tumour image [2]. However, tumour tissue is highly heterogeneous and adjacent tumour tissue areas may have very different properties [3]. This reflects a problem in traditional histopathology: if we only quantify objects in a small part of the whole.

Published in Oncology News, 2016

Download here

Maternal Hyperleptinemia Is Associated with Male Offspring’s Altered Vascular Function and Structure in Mice

Abstract


Children of mothers with gestational diabetes have greater risk of developing hypertension but little is known about the mechanisms by which this occurs. The objective of this study was to test the hypothesis that high maternal concentrations of leptin during pregnancy, which are present in mothers with gestational diabetes and/or obesity, alter blood pressure, vascular structure and vascular function in offspring. Wildtype (WT) offspring of hyperleptinemic, normoglycemic, Leprdb/+ dams were compared to genotype matched offspring of WT-control dams. Vascular function was assessed in male offspring at 6, and at 31 weeks of age after half the offspring had been fed a high fat, high sucrose diet (HFD) for 6 weeks. Blood pressure was increased by HFD but not affected by maternal hyperleptinemia. On a standard diet, offspring of hyperleptinemic dams had outwardly remodeled mesenteric arteries and an enhanced vasodilatory response to insulin. In offspring of WT but not Leprdb/+ dams, HFD induced vessel hypertrophy and enhanced vasodilatory responses to acetylcholine, while HFD reduced insulin responsiveness in offspring of hyperleptinemic dams. Offspring of hyperleptinemic dams had stiffer arteries regardless of diet. Therefore, while maternal hyperleptinemia was largely beneficial to offspring vascular health under a standard diet, it had detrimental effects in offspring fed HFD. These results suggest that circulating maternal leptin concentrations may interact with other factors in the pre- and post -natal environments to contribute to altered vascular function in offspring of diabetic pregnancies.”

Published in PLoS ONE, 2016

Download here

Behavior Analysis of Ants from Video Sequences

Abstract


The movement of small animals in well-defined environments is increasingly studied in many areas; including ecotoxicology, learning, and behavioural ecology. Here we de- scribe an algorithm designed to analyze individual foraging ants from a colony of Lasius niger. The inputs to the algorithm were images from a video sequence. The algorithm performed a series of pre-processing steps to identify the ants from the pixels, measurements were extracted and individual ants were tracked in time. The location of the ants in position and time were recorded as heat maps denoting the favorite locations of the ants. The ants were videoed in a foraging experiment on a T-maze a single trail bifurcation.

Published in ICPR2016, 2016

Download here

Three-dimensional quantification of the morphology and intragranular void ratio of a shelly carbonate sand

Abstract


Shelly carbonate sands represent an extreme soil type in terms of their mechanical behavior which derives from the bioclastic nature of the constituent grains. In their uncemented form, these deposits exhibit very high compressibility, which has posed a number of geotechnical engineering problems; in most cases related to the reduction in the bearing capacities of both shallow and deep foundations. Remarkable features of these carbonate sands include the complex shape and the structural weakness of the grains and the high inter and intra granular porosity. Previous studies, have quoted the interlocking of the angular shelly particles to be at the origin of their high friction angles and high initial void ratio, however, up until now, no scientific micro-scale examination has been carried out. This paper presents a non-invasive image based investigation into the grain morphology of a carbonate sand from the Persian Gulf. This sand has a median grain size of 570μm and a high CaCO3 content in the form of aragonite and calcite. Three-dimensional images from x-ray computed tomography (3DXRCT) with a size of 6μm were used. The presence of various skeletal bodies such as shells of small organisms with distinct densities and composition poses real challenges for an accurate segmentation. Image processing algorithms were developed in order to identify the individual sand grains and quantify their properties. Earlier work on silica sands has highlighted the importance of 3D non-invasive techniques in providing an accurate distribution of the grain sizes when compared to more traditional techniques such as sieving analysis and 2D microscopy.

Published in Deformation Characteristics of Geomaterials, 2016

Download here

Image-based investigation into the primary fabric of stress-transmitting particles in sand

Abstract


This paper uses three-dimensional images of a natural silica sand to analyse the mechanisms of stress transmission under triaxial compression. As discussed in Fonseca, J., O’Sullivan, C., Coop, M., Lee, P.D., (2012), the irregular morphology and locked fabric that can be found in natural sands lead to the formation of contacts with extended surface areas. However, most of our current understanding of stress-transmission phenomena comes from DEM simulations and photo-elastic experiments using idealised grain shapes and contact topologies. The direct measurement of stress transmission in assemblies of real soil grains is a challenging task. The present study postulates that important insight can be obtained by following the evolution of intergranular contacts as the grains rearrange and by considering how these rearrangements enhance the stability of the material. The methodology consists of measuring the geometrical data of the individual grains and their associated contacts obtained at successive load stages in the post-peak regime (after shear band formation). A statistical analysis of the vectors normal to the contacts reveals a realignment of these vectors in the direction of the major principal stress; this is a clear indication of the formation of force chains. A subsequent analysis shows that these columnar structures of stress-transmitting grains are associated with larger contact surfaces and have distinct patterns in the regions affected by the formation of a shear band. An algorithm based on stability and load-transmission criteria is developed to contribute new insight into the characterisation of load-bearing sand particles.”

Published in Soils and Foundations, 2016

Download here

The proportion of cancer-related entries in PubMed has increased considerably; is cancer truly ‘The Emperor of All Maladies’?

Abstract


In this work, the public database of biomedical literature PubMed was mined using queries with combinations of keywords and year restrictions. It was found that the proportion of Cancer-related entries per year in PubMed has risen from around 6\% in 1950 to more than 16\% in 2016. This increase is not shared by other conditions such as AIDS, Malaria, Tuberculosis, Diabetes, Cardiovascular, Stroke and Infection some of which have, on the contrary, decreased as a proportion of the total entries per year. Organ-related queries were performed to analyse the variation of some specific cancers. A series of queries related to incidence, funding, and relationship with DNA, Computing and Mathematics, were performed to test correlation between the keywords, with the hope of elucidating the cause behind the rise of Cancer in PubMed. Interestingly, the proportion of Cancer-related entries that contain ““DNA””, ““Computational”” or ““Mathematical”” have increased, which suggests that the impact of these scientific advances on Cancer has been stronger than in other conditions. It is important to highlight that the results obtained with the data mining approach here presented are limited to the presence or absence of the keywords on a single, yet extensive, database. Therefore, results should be observed with caution. All the data used for this work is publicly available through PubMed and the UK’s Office for National Statistics. All queries and figures were generated with the software platform Matlab and the files are available as supplementary material.”

Published in PLoS ONE, 2017

A hybrid energy model for region based curve evolution - Application to CTA coronary segmentation

Abstract


BACKGROUND AND OBJECTIVE: State-of-the-art medical imaging techniques have enabled non-invasive imaging of the internal organs. However, high volumes of imaging data make manual interpretation and delineation of abnormalities cumbersome for clinicians. These challenges have driven intensive research into efficient medical image segmentation. In this work, we propose a hybrid region-based energy formulation for effective segmentation in computed tomography angiography (CTA) imager”

Published in Computer Methods and Programs in Biomedicine, 2017

Download here

Framework for detection and localization of coronary non-calcified plaques in cardiac CTA using mean radial profiles

Abstract


BACKGROUND AND OBJECTIVE: The high mortality rate associated with coronary heart disease (CHD) has driven intensive research in cardiac imaging and image analysis. The advent of computed tomography angiography (CTA) has turned non-invasive diagnosis of cardiovascular anomalies into reality as calcified coronary plaques can be easily identified due to their high intensity values. However, the detection of non-calcified plaques in CTA is still a challenging problem because of lower intensity values, which are often similar to the nearby blood and muscle tissues. In this work, we propose the use of mean radial profiles for the detection of non-calcified plaques in CTA imager”

Published in Computers in Biology and Medicine, 2017

Download here

Segmentation and Shape Analysis of Macrophages Using Anglegram Analysis

Abstract


Cell migration is crucial in many processes of development and maintenance of multicellular organisms and it can also be related to disease, e.g., Cancer metastasis, when cells migrate to organs different to where they originate. A precise analysis of the cell shapes in biological studies could lead to insights about migration. However, in some cases, the interaction and overlap of cells can complicate the detection and interpretation of their shapes. This paper describes an algorithm to segment and analyse the shape of macrophages in fluorescent microscopy image sequences, and compares the segmentation of overlapping cells through different algorithms. A novel 2D matrix with multiscale angle variation, called the anglegram, based on the angles between points of the boundary of an object, is used for this purpose. The anglegram is used to find junctions of cells and applied in two different applications: (i) segmentation of overlapping cells and for non-overlapping cells; (ii) detection of the “corners” or pointy edges in the shapes. The functionalities of the anglegram were tested and validated with synthetic data and on fluorescently labelled macrophages observed on embryos of Drosophila melanogaster. The information that can be extracted from the anglegram shows a good promise for shape determination and analysis, whether this involves overlapping or non-overlapping objects.”

Published in Journal of Imaging, 2017

Download here

Automated Quantification of Non-Calcified Coronary Plaques in Cardiac CT Angiographic Imagery

Abstract

The high mortality rate associated with coronary heart disease (CHD) has driven intensive research in cardiac image analysis. The advent of computed tomography angiogra-phy (CTA) has turned non-invasive diagnosis of cardiovascular anomalies into reality as calcified coronary plaques can be easily identified due to high intensity values. However, detection and quantification of the non-calcified plaques in CTA is still a challenging problem because of their lower intensity values, which are often similar to the nearby blood and muscle tissues. In this work, we propose Bayesian posterior based model for precise quantification of the non-calcified plaques in CTA imagery. The only indicator of non-calcified plaques in CTA is relatively lower intensity. Hence, we exploited intensity variations to discriminate voxels into lumen and plaque classes. Based on the normal coronary segments, we computed the vessel-wall thickness in first step. In the subsequent step, we removed vessel wall from the seg-mented tree and employed Gaussian Mixture Model to compute optimal distribution parameters. In the final step, distribution parameters were employed in Bayesian posterior model to classify voxels into lumen or plaque. A total of 18 CTA volumes were analyzed in this work using two different approaches. According to the experimental results, mean Jaccard overlap is around 88\% with respect to the manual expert. In terms of sensitivity, specificity and accuracy, the proposed method achieves 84.13\% ,79.15\% and 82.02\% success, respectively. Conclusion: According to the experimental results, it is shown that the proposed plaque quantification method achieves accuracy equivalent to human experts.”

Published in International Journal of Advanced Computer Science and Applications (IJACSA), 2018

Download here

Large-scale database mining reveals hidden trends and future directions for cancer immunotherapy

Abstract


Cancer immunotherapy has fundamentally changed the landscape of oncology in recent years and significant resources are invested into immunotherapy research. It is in the interests of researchers and clinicians to identify promising and less promising trends in this field in order to rationally allocate resources. This requires a quantitative large-scale analysis of cancer immunotherapy related databases. We developed a novel tool for text mining, statistical analysis and data visualization of scientific literature data. We used this tool to analyze 72002 cancer immunotherapy publications and 1469 clinical trials from public databases. All source codes are available under an open access license. The contribution of specific topics within the cancer immunotherapy field has markedly shifted over the years. We show that the focus is moving from cell-based therapy and vaccination towards checkpoint inhibitors, with these trends reaching statistical significance. Rapidly growing subfields include the combination of chemotherapy with checkpoint blockade. Translational studies have shifted from hematological and skin neoplasms to gastrointestinal and lung cancer and from tumor antigens and angiogenesis to tumor stroma and apoptosis. This work highlights the importance of unbiased large-scale database mining to assess trends in cancer research and cancer immunotherapy in particular. Researchers, clinicians and funding agencies should be aware of quantitative trends in the immunotherapy field, allocate resources to the most promising areas and find new approaches for currently immature topics.”

Published in Oncoimmunology, 2018

Download here

Quantitative MRI Brain Studies in Mild Cognitive Impairment and Alzheimer’s Disease A Methodological Review

Abstract


Classifying and predicting Alzheimer’s disease (AD) in individuals with memory disorders through clinical and psychometric assessment is challenging, especially in mild cognitive impairment (MCI) subjects. Quantitative structural magnetic resonance imaging acquisition methods in combination with computer-aided diagnosis are currently being used for the assessment of AD. These acquisitions methods include voxel-based morphometry, volumetric measurements in specific regions of interest (ROIs), cortical thickness measurements, shape analysis, and texture analysis. This review evaluates the aforementioned methods in the classification of cases into one of the following three groups: normal controls, MCI, and AD subjects. Furthermore, the performance of the methods is assessed on the prediction of conversion from MCI to AD. In parallel, it is also assessed which ROIs are preferred in both classification and prognosis through the different states of the disease. Structural changes in the early stages of the disease are more pronounced in the medial temporal lobe, especially in the entorhinal cortex, whereas with disease progression, both entorhinal cortex and hippocampus offer similar discriminative power. However, for the conversion from MCI subjects to AD, entorhinal cortex provides better predictive accuracies rather than other structures, such as the hippocampus.”

Published in IEEE reviews in biomedical engineering, 2018

Download here

Shape analysis and tracking of migrating macrophages

Abstract


Cell migration is important in many human processes of development and disease. In Cancer, migration can be related to metastasis or cell defects. A precise analysis of the cell shapes in biological studies could lead to insights about migration. Therefore, this paper describes an algorithm to iteratively segment, track and analyse the shape of macrophages from fluorescent microscopy image sequences. This process allows observation of shape variations as the cells migrate. The algorithm identifies and separates overlapping and non-overlapping cells, then for the non-overlapping cases analyses the shape and extracts a series of measurements, including the number of “corner” or pointy edges through a multiscale angle variation matrix, anglegram. The shape evolution algorithm was tested on fluorescently labelled macrophages observed on embryos of Drosophila melanogaster.

Published in ISBI, 2018

Download here

Morphological Analysis of Fenestrae in Arteries

Abstract


Cardiovascular disease (CVD) is worldwide cause of death. The morphological structure of one of the regions of arteries called the internal elastic lamina (IEL) is associated with the stiffness of arteries, especially the presence and characteristics of small holes called . Structural analysis of the IEL as observed with multiphoton or confocal fluorescent microscopy is complicated, primarily due to is three-dimensional distribution along a series of z-stack slices. In addition, whilst the top slices of an artery cross long lamina sections, the bottom slices only cross short sections of the lamina and would be better observed in a different plane than that of the acquired image.

Published in Emerging Technologies in Computing. First International Conference, iCETiC 2018, 2018

Download here

Generating Magnetic Resonance Spectroscopy Imaging Data of Brain Tumours from Linear, Non-linear and Deep Learning Models

Abstract


Magnetic Resonance Spectroscopy (MRS) provides valuable information to help with the identification and understanding of brain tumors, yet MRS is not a widely available medical imaging modality. Aiming to counter this issue, this research draws on the advancements in machine learning techniques in other fields for the generation of artificial data. The generated methods were tested through the evaluation of their output against that of a real-world labelled MRS brain tumor data-set. Furthermore the resultant output from the generative techniques were each used to train separate traditional classifiers which were tested on a subset of the real MRS brain tumor dataset. The results suggest that there exist methods capable of producing accurate, ground truth based MRS voxels. These findings indicate that through generative techniques, large datasets can be made available for training deep, learning models for the use in brain tumor diagnosis.

Published in MICCAI Sashimi, 2018

Download here

Texture Segmentation An Objective Comparison between Five Traditional Algorithms and a Deep-Learning U-Net Architecture

Abstract


This paper compares a series of traditional and deep learning methodologies for the segmentation of textures. Six well-known texture composites first published by Randen and Hus\øy were used to compare traditional segmentation techniques (co-occurrence, filtering, local binary patterns, watershed, multiresolution sub-band filtering) against a deep-learning approach based on the U-Net architecture. For the latter, the effects of depth of the network, number of epochs and different optimisation algorithms were investigated. Overall, the best results were provided by the deep-learning approach. However, the best results were distributed within the parameters, and many configurations provided results well below the traditional techniques.”

Published in Applied Sciences, 2019

Download here

Digital Pathology: 15th European Congress, ECDP 2019, Warwick, UK, April 10–13, 2019, Proceedings

Abstract


This book constitutes the refereed proceedings of the 15th European Congress on Digital Pathology, ECDP 2019, held in Warwick, UK in April 2019. The 21 full papers presented in this volume were carefully reviewed and selected from 30 submissions. The congress theme will be Accelerating Clinical Deployment, with a focus on computational pathology and leveraging the power of big data and artificial intelligence to bridge the gaps between research, development, and clinical uptake.

Published in ECDP2019, 2019

Download here

Macrosight A Novel Framework to Analyze the Shape and Movement of Interacting Macrophages Using Matlab®

Abstract


This paper presents a novel software framework, called macrosight, which incorporates routines to detect, track, and analyze the shape and movement of objects, with special emphasis on macrophages. The key feature presented in macrosight consists of an algorithm to assess the changes of direction derived from cell\–cell contact, where an interaction is assumed to occur. The main biological motivation is the determination of certain cell interactions influencing cell migration. Thus, the main objective of this work is to provide insights into the notion that interactions between cell structures cause a change in orientation. Macrosight analyzes the change of direction of cells before and after they come in contact with another cell. Interactions are determined when the cells overlap and form clumps of two or more cells. The framework integrates a segmentation technique capable of detecting overlapping cells and a tracking framework into a tool for the analysis of the trajectories of cells before and after they overlap. Preliminary results show promise into the analysis and the hypothesis proposed, and lays the groundwork for further developments. The extensive experimentation and data analysis show, with statistical significance, that under certain conditions, the movement changes before and after an interaction are different from movement in controlled cases.”

Published in Journal of Imaging, 2019

Download here

Predicting survival from colorectal cancer histology slides using deep learning A retrospective multicenter study

Abstract


Background For virtually every patient with colorectal cancer (CRC), hematoxylin–eosin (HE)–stained tissue slides are available. These images contain quantitative information, which is not routinely used to objectively extract prognostic biomarkers. In the present study, we investigated whether deep convolutional neural networks (CNNs) can extract prognosticators directly from these widely available images. Methods and findings We hand-delineated single-tissue regions in 86 CRC tissue slides, yielding more than 100,000 HE image patches, and used these to train a CNN by transfer learning, reaching a nine-class accuracy of 94% in an independent data set of 7,180 images from 25 CRC patients. With this tool, we performed automated tissue decomposition of representative multitissue HE images from 862 HE slides in 500 stage I–IV CRC patients in the The Cancer Genome Atlas (TCGA) cohort, a large international multicenter collection of CRC tissue. Based on the output neuron activations in the CNN, we calculated a “deep stroma score,” which was an independent prognostic factor for overall survival (OS) in a multivariable Cox proportional hazard model (hazard ratio [HR] with 95% confidence interval [CI]: 1.99 [1.27–3.12], p = 0.0028), while in the same cohort, manual quantification of stromal areas and a gene expression signature of cancer-associated fibroblasts (CAFs) were only prognostic in specific tumor stages. We validated these findings in an independent cohort of 409 stage I–IV CRC patients from the “Darmkrebs: Chancen der Verhütung durch Screening” (DACHS) study who were recruited between 2003 and 2007 in multiple institutions in Germany. Again, the score was an independent prognostic factor for OS (HR 1.63 [1.14–2.33], p = 0.008), CRC-specific OS (HR 2.29 [1.5–3.48], p = 0.0004), and relapse-free survival (RFS; HR 1.92 [1.34–2.76], p = 0.0004). A prospective validation is required before this biomarker can be implemented in clinical workflows. Conclusions In our retrospective study, we show that a CNN can assess the human tumor microenvironment and predict prognosis directly from histopathological images.

Published in PLoS Medicine, 2019

Download here

Asymmetries in tongue-palate contact during speech

Abstract

Research has shown that speech articulation tends to be asymmetrical in the transverse plane of the vocal tract. A recent meta-study of previously published electropalatograms revealed that 83% of these images show asymmetrical tongue-palate contact [1].

Published in Proceedings of the 19th International Congress of Phonetic Sciences., 2019

Download here

Segmentation and Modelling of the Nuclear Envelope of HeLa Cells Imaged with Serial Block Face Scanning Electron Microscopy

Abstract


This paper describes an unsupervised algorithm, which segments the nuclear envelope of HeLa cells imaged by Serial Block Face Scanning Electron Microscopy. The algorithm exploits the variations of pixel intensity in different cellular regions by calculating edges, which are then used to generate superpixels. The superpixels are morphologically processed and those that correspond to the nuclear region are selected through the analysis of size, position, and correspondence with regions detected in neighbouring slices. The nuclear envelope is segmented from the nuclear region. The three-dimensional segmented nuclear envelope is then modelled against a spheroid to create a two-dimensional (2D) surface. The 2D surface summarises the complex 3D shape of the nuclear envelope and allows the extraction of metrics that may be relevant to characterise the nature of cells. The algorithm was developed and validated on a single cell and tested in six separate cells, each with 300 slices of 2000 \× 2000 pixels. Ground truth was available for two of these cells, i.e., 600 hand-segmented slices. The accuracy of the algorithm was evaluated with two similarity metrics: Jaccard Similarity Index and Mean Hausdorff distance. Jaccard values of the first/second segmentation were 93\%/90\% for the whole cell, and 98\%/94\% between slices 75 and 225, as the central slices of the nucleus are more regular than those on the extremes. Mean Hausdorff distances were 9/17 pixels for the whole cells and 4/13 pixels for central slices. One slice was processed in approximately 8 s and a whole cell in 40 min. The algorithm outperformed active contours in both accuracy and time.”

Published in Journal of Imaging, 2019

Download here

β-glucan-dependent shuttling of conidia from neutrophils to macrophages occurs during fungal infection establishment

Abstract


The initial host response to fungal pathogen invasion is critical to infection establishment and outcome. However, the diversity of leukocyte-pathogen interactions is only recently being appreciated. We describe a new form of interleukocyte conidial exchange called ““shuttling.”” In Talaromyces marneffei and Aspergillus fumigatus zebrafish in vivo infections, live imaging demonstrated conidia initially phagocytosed by neutrophils were transferred to macrophages. Shuttling is unidirectional, not a chance event, and involves alterations of phagocyte mobility, intercellular tethering, and phagosome transfer. Shuttling kinetics were fungal-species-specific, implicating a fungal determinant. β-glucan serves as a fungal-derived signal sufficient for shuttling. Murine phagocytes also shuttled in vitro. The impact of shuttling for microbiological outcomes of in vivo infections is difficult to specifically assess experimentally, but for these two pathogens, shuttling augments initial conidial redistribution away from fungicidal neutrophils into the favorable macrophage intracellular niche. Shuttling is a frequent host-pathogen interaction contributing to fungal infection establishment patterns.”

Published in PLoS Biology, 2019

Download here

Effect of Viscosity and Speed on Oil Cavitation Development in a Single Piston-Ring Lubricant Assembly

“<h2> Abstract </h2>” “A high-speed camera has been used to produce unique time-resolved images of high quality to describe the dynamics of the lubricant flow and cavitation characteristics in a sliding optical liner over a fixed single piston-ring lubricant assembly for three lubricants with different viscosities to establish their impact on cavitation formation and development. The images were obtained at two cranking speeds (or liner sliding velocity) of 300 rpm (0–0.36 m/s) and 600 rpm (0–0.72 m/s), at a lubricant temperature of 70 °C and a supply lubricant rate of 0.05 L/min. A special MATLAB programme has been developed to analyse the cavitation characteristics quantitatively. The dynamic process of cavities initiation was demonstrated by time-resolved images from fern cavity formation to fissure cavities and then their development to the sheet and strings cavities at a liner sliding velocity of around 0.17 m/s. The results for both up- and down-stroke motions showed that the cavities reach their fully developed state downstream of the contact point when the liner velocity reaches its highest velocity and that they start to collapse around TDC and BDC when the liner comes to rest. Within the measured range, viscosity had a great influence on length of cavities so that a decrease in viscosity (from Lubricant A to C) caused a reduction in length of cavities of up to 35\% for Lubricant C. On the other hand, an increase in speed, from 300 rpm to 600 rpm, have increased the number of string cavities and also increased the length of cavities due to thicker oil film thickness with the higher speed. Overall, the agreement between the processed data by MATLAB and visualisation measurements were good, but further thresholds refinement is required to improve the accuracy.”

Published in Lubricants, 2019

Download here

Detection of Pitt-Hopkins Syndrome based on morphological facial features

Abstract


This work describes an automatic methodology to discriminate between individuals with the genetic disorder Pitt-Hopkins syndrome (PTHS), and healthy individuals. As input data, the methodology accepts unconstrained frontal facial photographs, from which faces are located with Histograms of Oriented Gradients features descriptors. Pre-processing steps of the methodology consist of colour normalisation, scaling down, rotation, and cropping in order to produce a series of images of faces with consistent dimensions. Sixty eight facial landmarks are automatically located on each face through a cascade of regression functions learnt via gradient boosting to estimate the shape from an initial approximation. The intensities of a sparse set of pixels indexed relative to this initial estimate are used to determine the landmarks. A set of carefully selected geometric features, for example, relative width of the mouth, or angle of the nose, are extracted from the landmarks. The features are used to investigate the statistical differences between the two populations of PTHS and healthy controls. The methodology was tested on 71 individuals with PTHS and 55 healthy controls. Two geometric features related to the nose and mouth showed statistical difference between the two populations.

Published in arXiv, 2020

Download here

The relationship between Fully Connected Layers and number of classes for the analysis of retinal images

Abstract


This paper experiments with the number of fully-connected layers in a deep convolutional neural network as applied to the classification of fundus retinal images. The images analysed corresponded to the ODIR 2019 (Peking University International Competition on Ocular Disease Intelligent Recognition) [9], which included images of various eye diseases (cataract, glaucoma, myopia, diabetic retinopathy, age-related macular degeneration (AMD), hypertension) as well as normal cases. This work focused on the classification of Normal, Cataract, AMD and Myopia. The feature extraction (convolutional) part of the neural network is kept the same while the feature mapping (linear) part of the network is changed. Different data sets are also explored on these neural nets. Each data set differs from another by the number of classes it has. This paper hence aims to find the relationship between number of classes and number of fully-connected layers. It was found out that the effect of increasing the number of fully-connected layers of a neural networks depends on the type of data set being used. For simple, linearly separable data sets, addition of fully-connected layer is something that should be explored and that could result in better training accuracy, but a direct correlation was not found. However as complexity of the data set goes up(more overlapping classes), increasing the number of fully-connected layers causes the neural network to stop learning. This phenomenon happens quicker the more complex the data set is.

Published in arXiv, 2020

Download here

Geometric differences between nuclear envelopes of Wild-type and Chlamydia trachomatis-infected HeLa cells

Abstract


{\textless}h3{\textgreater}Abstract{\textless}/h3{\textgreater} {\textless}p{\textgreater}In this work, the geometrical characteristics of two different types of cells observed with Electron Microscopy were analysed. The nuclear envelope of Wild-type HeLa cells and Chlamydia trachomatis-infected HeLa cells were automatically segmented and then modelled against a spheroid and converted to a two-dimensional surface. Geometric measurements from this surface and the volumetric nuclear envelope were extracted to compare the two types of cells. The measurements included the nuclear volume, the sphericity of the nucleus, its flatness or spikiness. In total 13 different cells were segmented: 7 Wild-type and 6 Chlamydia trachomatis-infected. The cells were statistically different in the following measurements. Wild-type HeLa cells have greater volumes than that of Chlamydia trachomatis-infected HeLa cells and they are more spherical as Jaccard index suggests. Standard deviation (\textit{σ}), and range of values for the nuclear envelope, which shows the distance of the highest peaks and deepest valleys from the spheroid, were also extracted from the modelling against a spheroid and these metrics were used to compare two different data sets in order to draw conclusions.{\textless}/p{\textgreater}”

Published in bioRxiv, 2020

Download here

Geometric semi-automatic analysis of radiographs of Colles’ fractures

Abstract


Fractures of the wrist are common in Emergency Departments, where some patients are treated with a procedure called Manipulation under Anaesthesia. In some cases, this procedure is unsuccessful and patients need to revisit the hospital where they undergo surgery to treat the fracture. This work describes a geometric semi-automatic image analysis algorithm to analyse and compare the x-rays of healthy controls and patients with dorsally displaced wrist fractures (Colles’ fractures) who were treated with Manipulation under Anaesthesia. A series of 161 posterior-anterior radiographs from healthy controls and patients with Colles’ fractures were acquired and analysed. The patients’ group was further subdivided according to the outcome of the procedure (successful/unsuccessful) and pre- or post-intervention creating five groups in total (healthy, pre-successful, pre-unsuccessful, post-successful, post-unsuccessful). The semi-automatic analysis consisted of manual location of three landmarks (finger, lunate and radial styloid) and automatic processing to generate 32 geometric and texture measurements, which may be related to conditions such as osteoporosis and swelling of the wrist. Statistical differences were found between patients and controls, as well as between pre- and post-intervention, but not between the procedures. The most distinct measurements were those of texture. Although the study includes a relatively low number of cases and measurements, the statistical differences are encouraging.”

Published in PLoS ONE, 2020

Download here

Semantic segmentation of HeLa cells An objective comparison between one traditional algorithm and four deep-learning architectures

Abstract


The quantitative study of cell morphology is of great importance as the structure and condition of cells and their structures can be related to conditions of health or disease. The first step towards that, is the accurate segmentation of cell structures. In this work, we compare five approaches, one traditional and four deep-learning, for the semantic segmentation of the nuclear envelope of cervical cancer cells commonly known as HeLa cells. Images of a HeLa cancer cell were semantically segmented with one traditional image-processing algorithm and four three deep learning architectures: VGG16, ResNet18, Inception-ResNet-v2, and U-Net. Three hundred slices, each 2000 × 2000 pixels, of a HeLa Cell were acquired with Serial Block Face Scanning Electron Microscopy. The first three deep learning architectures were pre-trained with ImageNet and then fine-tuned with transfer learning. The U-Net architecture was trained from scratch with 36, 000 training images and labels of size 128 × 128. The image-processing algorithm followed a pipeline of several traditional steps like edge detection, dilation and morphological operators. The algorithms were compared by measuring pixel-based segmentation accuracy and Jaccard index against a labelled ground truth. The results indicated a superior performance of the traditional algorithm (Accuracy = 99\%, Jaccard = 93\%) over the deep learning architectures: VGG16 (93\%, 90\%), ResNet18 (94\%, 88\%), Inception-ResNet-v2 (94\%, 89\%), and U-Net (92\%, 56\%).”

Published in PLoS ONE, 2020

Download here

Morphological Estimation of Cellularity on Neo-Adjuvant Treated Breast Cancer Histological Images

Abstract


This paper describes a methodology that extracts key morphological features from histological breast cancer images in order to automatically assess Tumour Cellularity (TC) in Neo-Adjuvant treatment (NAT) patients. The response to NAT gives information on therapy efficacy and it is measured by the residual cancer burden index, which is composed of two metrics: TC and the assessment of lymph nodes. The data consist of whole slide images (WSIs) of breast tissue stained with Hematoxylin and Eosin (H\&E) released in the 2019 SPIE Breast Challenge. The methodology proposed is based on traditional computer vision methods (K-means, watershed segmentation, Otsu\’s binarisation, and morphological operations), implementing colour separation, segmentation, and feature extraction. Correlation between morphological features and the residual TC after a NAT treatment was examined. Linear regression and statistical methods were used and twenty-two key morphological parameters from the nuclei, epithelial region, and the full image were extracted. Subsequently, an automated TC assessment that was based on Machine Learning (ML) algorithms was implemented and trained with only selected key parameters. The methodology was validated with the score assigned by two pathologists through the intra-class correlation coefficient (ICC). The selection of key morphological parameters improved the results reported over other ML methodologies and it was very close to deep learning methodologies. These results are encouraging, as a traditionally-trained ML algorithm can be useful when limited training data are available preventing the use of deep learning approaches.”

Published in Journal of Imaging, 2020

Download here

Radiography Classification: A Comparison between Eleven Convolutional Neural Networks

Abstract


This paper investigates the classification of normal and abnormal radiographic images. Eleven convolutional neural network architectures (GoogleNet, Vgg-19, AlexNet, SqueezeNet, ResNet-18, Inception-v3, ResNet-50, Vgg-16, ResNet-101, DenseNet-201 and Inception-ResNet-v2) were used to classify a series of x-ray images from Stanford Musculoskeletal Radiographs (MURA) dataset corresponding to the wrist images of the data base. For each architecture, the results were compared against the known labels (normal / abnormal) and then the following metrics were calculated: accuracy (labels correctly classified) and Cohen’s kappa (a measure of agreement) following MURA guidelines. Numerous experiments were conducted by changing classifiers (Adam, Sgdm, RmsProp), the number of epochs, with/without data augmentation. The best results were provided by InceptionResnet-v2 (Mean accuracy = 0.723, Mean Kappa = 0.506). Interestingly, these results lower than those reported in the Leaderboard of MURA. We speculate that to improve the results from basic CNN architectures several options could be tested, for instance: pre-processing, post-processing or domain knowledge, and ensembles.

Published in Fourth International Conference on Multimedia Computing, Networking and Applications (MCNA), 2020

Download here

Experimental Assessment of Color Deconvolution and Color Normalization for Automated Classification of Histology Images Stained with Hematoxylin and Eosin

Abstract


Histological evaluation plays a major role in cancer diagnosis and treatment. The appearance of H\&E-stained images can vary significantly as a consequence of differences in several factors, such as reagents, staining conditions, preparation procedure and image acquisition system. Such potential sources of noise can all have negative effects on computer-assisted classification. To minimize such artefacts and their potentially negative effects several color pre-processing methods have been proposed in the literature-for instance, color augmentation, color constancy, color deconvolution and color transfer. Still, little work has been done to investigate the efficacy of these methods on a quantitative basis. In this paper, we evaluated the effects of color constancy, deconvolution and transfer on automated classification of H\&E-stained images representing different types of cancers-specifically breast, prostate, colorectal cancer and malignant lymphoma. Our results indicate that in most cases color pre-processing does not improve the classification accuracy, especially when coupled with color-based image descriptors. Some pre-processing methods, however, can be beneficial when used with some texture-based methods like Gabor filters and Local Binary Patterns.”

Published in Cancers, 2020

Volumetric Semantic Instance Segmentation of the Plasma Membrane of HeLa Cells

Abstract



In this work, an unsupervised volumetric semantic instance segmentation of the plasma membrane of HeLa cells as observed with serial block face scanning electron microscopy is described. The resin background of the images was segmented at different slices of a 3D stack of 518 slices with 8192 × 8192 pixels each. The background was used to create a distance map, which helped identify and rank the cells by their size at each slice. The centroids of the cells detected at different slices were linked to identify them as a single cell that spanned a number of slices. A subset of these cells, i.e., the largest ones and those not close to the edges were selected for further processing. The selected cells were then automatically cropped to smaller regions of interest of 2000 × 2000 × 300 voxels that were treated as cell instances. Then, for each of these volumes, the nucleus was segmented, and the cell was separated from any neighbouring cells through a series of traditional image processing steps that followed the plasma membrane. The segmentation process was repeated for all the regions of interest previously selected. For one cell for which the ground truth was available, the algorithm provided excellent results in Accuracy (AC) and the Jaccard similarity Index (JI): nucleus: JI =0.9665, AC =0.9975, cell including nucleus JI =0.8711, AC =0.9655, cell excluding nucleus JI =0.8094, AC =0.9629. A limitation of the algorithm for the plasma membrane segmentation was the presence of background. In samples with tightly packed cells, this may not be available. When tested for these conditions, the segmentation of the nuclear envelope was still possible. All the code and data were released openly through GitHub, Zenodo and EMPIAR.

Published in Journal of Imaging, 2021

Download here

Special Issue on Computational Pathology: An Overview

Abstract



The field of Digital Pathology has grown considerably in the past few years. The early years focused primarily on the scanning and remote viewing of histopathological images. However, as the use of tele-pathology has become widespread, the emphasis on digital pathology has increasingly moved towards the application of image analysis and artificial intelligence techniques to maximize the meaningful information that can be extracted from tissue sample with the final aim of aiding the diagnosis of the pathologist. Interestingly, with advancements in computational algorithms for digital pathology, the types of problems that researchers have embarked on have evolved from focusing solely on decision support for the pathologist to also to begin to prognosticate disease outcome, predict tumor biology and also therapeutic response.

Published in Medical Image Analysis, 2021

Download here

Classification and Visualisation of Normal and Abnormal Radiographs; A Comparison between Eleven Convolutional Neural Network Architectures

Abstract


This paper investigates the classification of radiographic images with eleven convolutional neural network (CNN) architectures (GoogleNet, VGG-19, AlexNet, SqueezeNet, ResNet-18, Inception-v3, ResNet-50, VGG-16, ResNet-101, DenseNet-201 and Inception-ResNet-v2). The CNNs were used to classify a series of wrist radiographs from the Stanford Musculoskeletal Radiographs (MURA) dataset into two classes—normal and abnormal. The architectures were compared for different hyper-parameters against accuracy and Cohen’s kappa coefficient. The best two results were then explored with data augmentation. Without the use of augmentation, the best results were provided by Inception-ResNet-v2 (Mean accuracy = 0.723, Mean kappa = 0.506). These were significantly improved with augmentation to Inception-ResNet-v2 (Mean accuracy = 0.857, Mean kappa = 0.703). Finally, Class Activation Mapping was applied to interpret activation of the network against the location of an anomaly in the radiographs

Published in Sensors (Basel), 2021

Download here

Contrastive Counterfactual Visual Explanations With Overdetermination

Abstract


A novel explainable AI method called CLEAR Image is introduced in this paper. CLEAR Image is based on the view that a satisfactory explanation should be contrastive, counterfactual and measurable. CLEAR Image explains an image’s classification probability by contrasting the image with a corresponding image generated automatically via adversarial learning. This enables both salient segmentation and perturbations that faithfully determine each segment’s importance. CLEAR Image was successfully applied to a medical imaging case study where it outperformed methods such as Grad-CAM and LIME by an average of 27% using a novel pointing game metric. CLEAR Image excels in identifying cases of “causal overdetermination” where there are multiple patches in an image, any one of which is sufficient by itself to cause the classification probability to be close to one.

Published in arXiv, 2021

Download here

Detection of Pitt–Hopkins Syndrome Based on Morphological Facial Features

Abstract


This work describes an automatic methodology to discriminate between individuals with the genetic disorder Pitt-Hopkins syndrome (PTHS), and healthy individuals. As input data, the methodology accepts unconstrained frontal facial photographs, from which faces are located with Histograms of Oriented Gradients features descriptors. Pre-processing steps of the methodology consist of colour normalisation, scaling down, rotation, and cropping in order to produce a series of images of faces with consistent dimensions. Sixty eight facial landmarks are automatically located on each face through a cascade of regression functions learnt via gradient boosting to estimate the shape from an initial approximation. The intensities of a sparse set of pixels indexed relative to this initial estimate are used to determine the landmarks. A set of carefully selected geometric features, for example, relative width of the mouth, or angle of the nose, are extracted from the landmarks. The features are used to investigate the statistical differences between the two populations of PTHS and healthy controls. The methodology was tested on 71 individuals with PTHS and 55 healthy controls. Two geometric features related to the nose and mouth showed statistical difference between the two populations.

Published in Applied Sciences, 2021

Download here

Automatic Gemstone Classification Using Computer Vision

Abstract


This paper presents a computer-vision-based methodology for automatic image-based classification of 2042 training images and 284 unseen (test) images divided into 68 categories of gemstones. A series of feature extraction techniques (33 including colour histograms in the RGB, HSV and CIELAB space, local binary pattern, Haralick texture and grey-level co-occurrence matrix properties) were used in combination with different machine-learning algorithms (Logistic Regression, Linear Discriminant Analysis, K-Nearest Neighbour, Decision Tree, Random Forest, Naive Bayes and Support Vector Machine). Deep-learning classification with ResNet-18 and ResNet-50 was also investigated. The optimal combination was provided by a Random Forest algorithm with the RGB eight-bin colour histogram and local binary pattern features, with an accuracy of 69.4% on unseen images; the algorithms required 0.0165 s to process the 284 test images. These results were compared against three expert gemmologists with at least 5 years of experience in gemstone identification, who obtained accuracies between 42.6% and 66.9% and took 42–175 min to classify the test images. As expected, the human experts took much longer than the computer vision algorithms, which in addition provided, albeit marginal, higher accuracy. Although these experiments included a relatively low number of images, the superiority of computer vision over humans is in line with what has been reported in other areas of study, and it is encouraging to further explore the application in gemmology and related areas.

Published in Minerals, 2021

Download here

Computational Image Analysis Techniques, Programming Languages and Software Platforms Used in Cancer Research: A Scoping Review

Abstract



Background: Cancer-related research, as indicated by the number of entries in Medline, the National Library of Medicine of the USA, has dominated the medical literature. An important component of this research is based on the use of computational techniques to analyse the data produced by the many acquisition modalities. This paper presents a review of the computational image analysis techniques that have been applied to cancer. The review was performed through automated mining of Medline/PubMed entries with a combination of keywords. In addition, the programming languages and software platforms through which these techniques are applied were also reviewed.

Published in Lecture Notes in Computer Science, 2022

Download here

Volumetric Semantic Instance Segmentation of the Plasma Membrane of HeLa Cells

Abstract



In this work, the performance of five deep learning architectures in classifying COVID-19 in a multi-class set-up is evaluated. The classifiers were built on pretrained ResNet-50, ResNet-50r (with kernel size 5×5 in the first convolutional layer), DenseNet-121, MobileNet-v3 and the state-of-the-art CaiT-24-XXS-224 (CaiT) transformer. The cross entropy and weighted cross entropy were minimised with Adam and AdamW. In total, 20 experiments were conducted with 10 repetitions and obtained the following metrics: accuracy (Acc), balanced accuracy (BA), F1 and F2 from the general Fβ macro score, Matthew’s Correlation Coefficient (MCC), sensitivity (Sens) and specificity (Spec) followed by bootstrapping. The performance of the classifiers was compared by using the Friedman–Nemenyi test. The results show that less complex architectures such as ResNet-50, ResNet-50r and DenseNet-121 were able to achieve better generalization with rankings of 1.53, 1.71 and 3.05 for the Matthew Correlation Coefficient, respectively, while MobileNet-v3 and CaiT obtained rankings of 3.72 and 5.0, respectively

Published in Journal of Imaging, 2022

Download here

FatNet: High Resolution Kernels for Classification Using Fully Convolutional Optical Neural Networks

Abstract


This paper describes the transformation of a traditional in-silico classification network into an optical fully convolutional neural network with high-resolution feature maps and kernels. When using the free-space 4f system to accelerate the inference speed of neural networks, higher resolutions of feature maps and kernels can be used without the loss in frame rate. We present FatNet for the classification of images, which is more compatible with free-space acceleration than standard convolutional classifiers. It neglects the standard combination of convolutional feature extraction and classifier dense layers by performing both in one fully convolutional network. This approach takes full advantage of the parallelism in the 4f free-space system and performs fewer conversions between electronics and optics by reducing the number of channels and increasing the resolution, making the network faster in optics than off-the-shelf networks. To demonstrate the capabilities of FatNet, it trained with the CIFAR100 dataset on GPU and the simulator of the 4f system, then compared the results against ResNet-18. The results show 8.2 times fewer convolution operations at the cost of only 6% lower accuracy compared to the original network. These are promising results for the approach of training deep learning with high-resolution kernels in the direction towards the upcoming optics era.

Published in arXiv, 2022

FatNet: High-Resolution Kernels for Classification Using Fully Convolutional Optical Neural Networks

Abstract



This paper describes the transformation of a traditional in silico classification network into an optical fully convolutional neural network with high-resolution feature maps and kernels. When using the free-space 4f system to accelerate the inference speed of neural networks, higher resolutions of feature maps and kernels can be used without the loss in frame rate. We present FatNet for the classification of images, which is more compatible with free-space acceleration than standard convolutional classifiers. It neglects the standard combination of convolutional feature extraction and classifier dense layers by performing both in one fully convolutional network. This approach takes full advantage of the parallelism in the 4f free-space system and performs fewer conversions between electronics and optics by reducing the number of channels and increasing the resolution, making this network faster in optics than off-the-shelf networks. To demonstrate the capabilities of FatNet, it was trained with the CIFAR100 dataset on GPU and the simulator of the 4f system. A comparison of the results against ResNet-18 shows 8.2 times fewer convolution operations at the cost of only 6% lower accuracy. This demonstrates that the optical implementation of FatNet results in significantly faster inference than the optical implementation of the original ResNet-18. These are promising results for the approach of training deep learning with high-resolution kernels in the direction toward the upcoming optics era.

Published in AI, 2023

Download here

Impact of Training Data, Ground Truth and Shape Variability in the Deep Learning-Based Semantic Segmentation of HeLa Cells Observed with Electron Microscopy

Abstract



This paper investigates the impact of the amount of training data and the shape variability on the segmentation provided by the deep learning architecture U-Net. Further, the correctness of ground truth (GT) was also evaluated. The input data consisted of a three-dimensional set of images of HeLa cells observed with an electron microscope with dimensions 8192×8192×517. From there, a smaller region of interest (ROI) of 2000×2000×300 was cropped and manually delineated to obtain the ground truth necessary for a quantitative evaluation. A qualitative evaluation was performed on the 8192×8192 slices due to the lack of ground truth. Pairs of patches of data and labels for the classes nucleus, nuclear envelope, cell and background were generated to train U-Net architectures from scratch. Several training strategies were followed, and the results were compared against a traditional image processing algorithm. The correctness of GT, that is, the inclusion of one or more nuclei within the region of interest was also evaluated. The impact of the extent of training data was evaluated by comparing results from 36,000 pairs of data and label patches extracted from the odd slices in the central region, to 135,000 patches obtained from every other slice in the set. Then, 135,000 patches from several cells from the 8192×8192 slices were generated automatically using the image processing algorithm. Finally, the two sets of 135,000 pairs were combined to train once more with 270,000 pairs. As would be expected, the accuracy and Jaccard similarity index improved as the number of pairs increased for the ROI. This was also observed qualitatively for the 8192×8192 slices. When the 8192×8192 slices were segmented with U-Nets trained with 135,000 pairs, the architecture trained with automatically generated pairs provided better results than the architecture trained with the pairs from the manually segmented ground truths. This suggests that the pairs that were extracted automatically from many cells provided a better representation of the four classes of the various cells in the 8192×8192 slice than those pairs that were manually segmented from a single cell. Finally, the two sets of 135,000 pairs were combined, and the U-Net trained with these provided the best results.

Published in Journal of Imaging, 2023

Download here

Contrastive counterfactual visual explanations with overdetermination

Abstract


A novel explainable AI method called CLEAR Image is introduced in this paper. CLEAR Image is based on the view that a satisfactory explanation should be contrastive, counterfactual and measurable. CLEAR Image seeks to explain an image’s classification probability by contrasting the image with a representative contrast image, such as an auto-generated image obtained via adversarial learning. This produces a salient segmentation and a way of using image perturbations to calculate each segment’s importance. CLEAR Image then uses regression to determine a causal equation describing a classifier’s local input–output behaviour. Counterfactuals are also identified that are supported by the causal equation. Finally, CLEAR Image measures the fidelity of its explanation against the classifier. CLEAR Image was successfully applied to a medical imaging case study where it outperformed methods such as Grad-CAM and LIME by an average of 27% using a novel pointing game metric. CLEAR Image also identifies cases of causal overdetermination, where there are multiple segments in an image that are sufficient individually to cause the classification probability to be close to one.

Published in Machine Learning, 2023

Download here

Dynamic analysis of the effects of vehicle movement over bridges observed with CCTV images

Abstract


This work describes a combination of image analysis techniques used to identify vehicles travelling on a bridge with a vectorised modal dynamic analysis that can handle efficiently a large number of wheel loads on the deck at each analysis step-time. In the absence of weight-in-motion data, a randomisation of the traffic flow is proposed to define the weight of the vehicles as a function of their identified size. The methodology is applied to real CCTV recording on a conventional road bridge with a large width-to-span ratio in which the deck is modelled with shell elements. The latter is found to be important to capture the significant contribution of local slab modes to the vibrations along the sidewalks. The dynamic analysis of a large number of traffic records indicate that code-based load cases with long truck convoys lead to unrealistically large contributions of high-order modes and to vibrations that are categorised as uncomfortable, whilst the more realistic traffic flows obtained from image analysis satisfy the comfort criteria based on root-mean-square accelerations.

Published in Engrxiv, 2024

Download here

Fat-U-Net: Non-Contracting U-Net for Free-Space Optical Neural Networks

Abstract



This paper describes the advantages and disadvantages of adapting the U-Net architecture from a traditional GPU to a 4f free-space optical environment. The implementation is based on an optical-based acceleration called FatNet and thus this adaption is called Fat-U-Net. Fat-U-Net neglects the pooling operations in UNet, but maintains a similar number of weights and pixels per layer as U-Net. Our results demonstrate that the conversion to Fat-U-Net offers significant improvement in speed for segmentation tasks, with Fat-U-Net achieving a remarkable ×538 acceleration in inference compared to U-Net when both are run on optical devices and x37 acceleration in inference compared to the results provided by U-Net on GPU. The performance loss after conversion remains minimal in two datasets, with reductions of 4.24% in IoU for the Oxford IIIt pet dataset and 1.76% in IoU of HeLa cells nucleus segmentation.

Published in AI and Optical Data Sciences V, 2024

Download here

Editorial: Quantitative imaging methods and analysis in Alzheimer’s disease assessment

Abstract



Alzheimer’s disease (AD) is a complex condition gradually eroding memory and cognitive abilities. While cognitive tests aid diagnosis, their limitations include an inability to detect subtle early-stage changes, subjective interpretation influenced by various factors, and inconsistent predictive accuracy. Imaging techniques play a pivotal role in AD detection by revealing structural and functional brain changes associated with the disease. Quantitative MRI highlights brain atrophy, ruling out other causes, while PET scans visualize hallmark proteins like beta-amyloid and tau. Cerebrospinal fluid (CSF) analysis assesses biomarkers linked to AD, notably Aβ42 and tau proteins. The APOE gene variants, specifically ε4, influence AD susceptibility. EEG, measuring brain electrical activity, offers supplementary information but isn’t a primary diagnostic tool.

Published in Frontiers in Aging Neuroscience, 2024

Download here

Pulses of Class I PI3kinase activity identify the release and recapture of prey from neutrophil phagosomes

Abstract



Class I PI3kinases coordinate the delivery of microbicidal effectors to the phagosome by forming the phosphoinositide lipid second messenger, phosphatidylinositol (3, 4, 5)-trisphosphate (PIP3). However, the dynamics of PIP3 in neutrophils during a bacterial infection are unknown. We have therefore developed an in vivo, live zebrafish infection model that enables visualisation of dynamic changes in Class 1 PI3kinases (PI3K) signalling on neutrophil phagosomes in real-time. We have identified that on approximately 12% of neutrophil phagosomes PHAkt-eGFP, a reporter for Class 1 PI3K signalling, re-recruits in pulsatile bursts. This phenomenon occurred on phagosomes containing structurally and morphologically distinct prey, including Staphylococcus aureus and Mycobacterium abscessus, and was dependent on the activity of the Class 1 PI3K isoform, PI3kinase Gamma. Detailed imaging suggested that pulsing phagosomes represent neutrophils transiently reopening and reclosing phagosomes. This finding challenges the concept that phagosomes remain closed after prey engulfment and we propose that neutrophils occasionally use this alternative pathway of phagosome maturation to release phagosome contents and/or to restart phagosome maturation if digestion has stalled.

Published in bioRxiv, 2024

Download here

DRD-UNet, a UNet-like architecture for multi-class breast cancer semantic segmentation

Abstract


Staining of histological slides with Hematoxylin and Eosin is widely used in clinical and laboratory settings as these dyes reveal nuclear structures as well as cytoplasm and collagen. For cancer diagnosis, these slides are used to recognize tissues and morphological changes. Tissue semantic segmentation is therefore important and at the same time a challenging and time-consuming task. This paper describes a UNet-like deep learning architecture called DRD-UNet , which adds a novel processing block called DRD (Dilation, Residual, and Dense block) to a UNet architecture. DRD is formed by the combination of dilated convolutions (D), residual connections (R), and dense layers (D). DRD-UNet was applied to the multi-class (tumor, stroma, inflammatory, necrosis, and other) semantic segmentation of histological images from breast cancer samples stained with Hematoxylin and Eosin. The histological images were released through the Breast Cancer Semantic Segmentation (BCSS) Challenge. DRD-UNet outperformed the original UNet architecture and 15 other UNet-based architectures on the segmentation of 12,930 image patches extracted from regions of interest that ranged in size between 1036 × 1222 to 6813 × 7360 pixels. DRD-UNet obtained the best performance as measured with Jaccard similarity index, Dice coefficient, in a per-class comparison and accuracy for overall segmentation.

Published in IEEE Access, 2024

Download here

International Classification of Diseases Prediction from MIMIIC-III Clinical Text Using Pre-Trained ClinicalBERT and NLP Deep Learning Models Achieving State of the Art

Abstract


The International Classification of Diseases (ICD) serves as a widely employed framework for assigning diagnosis codes to electronic health records of patients. These codes facilitate the encapsulation of diagnoses and procedures conducted during a patient’s hospitalisation. This study aims to devise a predictive model for ICD codes based on the MIMIC-III clinical text dataset. Leveraging natural language processing techniques and deep learning architectures, we constructed a pipeline to distill pertinent information from the MIMIC-III dataset: the Medical Information Mart for Intensive Care III (MIMIC-III), a sizable, de-identified, and publicly accessible repository of medical records. Our method entails predicting diagnosis codes from unstructured data, such as discharge summaries and notes encompassing symptoms. We used state-of-the-art deep learning algorithms, such as recurrent neural networks (RNNs), long short-term memory (LSTM) networks, bidirectional LSTM (BiLSTM) and BERT models after tokenizing the clinical test with Bio-ClinicalBERT, a pre-trained model from Hugging Face. To evaluate the efficacy of our approach, we conducted experiments utilizing the discharge dataset within MIMIC-III. Employing the BERT model, our methodology exhibited commendable accuracy in predicting the top 10 and top 50 diagnosis codes within the MIMIC-III dataset, achieving average accuracies of 88% and 80%, respectively. In comparison to recent studies by Biseda and Kerang, as well as Gangavarapu, which reported F1 scores of 0.72 in predicting the top 10 ICD-10 codes, our model demonstrated better performance, with an F1 score of 0.87. Similarly, in predicting the top 50 ICD-10 codes, previous research achieved an F1 score of 0.75, whereas our method attained an F1 score of 0.81. These results underscore the better performance of deep learning models over conventional machine learning approaches in this domain, thus validating our findings. The ability to predict diagnoses early from clinical notes holds promise in assisting doctors or physicians in determining effective treatments, thereby reshaping the conventional paradigm of diagnosis-then-treatment care. Our code is available online.

Published in Big Data and Cognitive Computing, 2024

Download here

Insights from EEG analysis of evoked memory recalls using deep learning for emotion charting

Abstract



Affect recognition in a real-world, less constrained environment is the principal prerequisite of the industrial-level usefulness of this technology. Monitoring the psychological profile using smart, wearable electroencephalogram (EEG) sensors during daily activities without external stimuli, such as memory-induced emotions, is a challenging research gap in emotion recognition. This paper proposed a deep learning framework for improved memory-induced emotion recognition leveraging a combination of 1D-CNN and LSTM as feature extractors integrated with an Extreme Learning Machine (ELM) classifier. The proposed deep learning architecture, combined with the EEG preprocessing, such as the removal of the average baseline signal from each sample and extraction of EEG rhythms (delta, theta, alpha, beta, and gamma), aims to capture repetitive and continuous patterns for memory-induced emotion recognition, underexplored with deep learning techniques. This work has analyzed EEG signals using a wearable, ultra-mobile sports cap while recalling autobiographical emotional memories evoked by affect-denoting words, with self-annotation on the scale of valence and arousal. With extensive experimentation using the same dataset, the proposed framework empirically outperforms existing techniques for the emerging area of memory-induced emotion recognition with an accuracy of 65.6%. The EEG rhythms analysis, such as delta, theta, alpha, beta, and gamma, achieved 65.5%, 52.1%, 65.1%, 64.6%, and 65.0% accuracies for classification with four quadrants of valence and arousal. These results underscore the significant advancement achieved by our proposed method for the real-world environment of memory-induced emotion recognition.

Published in Scientific Reports, 2024

Download here

Lateral asymmetry in the articulation of British English speech sounds: An electropalatographic study

Abstract



This study investigated lateral asymmetry in the linguopalatal speech sounds of British English by means of electropalatography. This instrumental technique visualizes tongue–palate contact during speech production and allows for the quantification of contact patterns. The first and main objective of the study was to establish a method of measuring asymmetry that would be more sensitive than the approach used previously and would facilitate statistical analysis. The method employed a modified index of asymmetry and controlled for the overall amount of tongue–palate contact. The secondary objective was to use the proposed method to quantify asymmetry in the production of the linguopalatal consonants of British English, focusing on asymmetry observed in the region of the palate corresponding to the place of articulation. Regression analysis of 22,004 speech sounds, produced by four native speakers, indicated that the approximant /l/ is the most asymmetrical speech sound, followed by the central approximants /j, r/. Although fricatives had been hypothesized to be highly asymmetrical, they were not consistently more asymmetrical than plosives. In terms of the place of articulation of speech sounds, velar sounds were less asymmetrical than alveolars. It was possible to account for some of these findings by referring to the unilateral productions of approximants.

Published in Journal of the International Phonetic Association, 2025

Download here

Distinct Circle of Willis anatomical configurations in healthy preterm born adults: a 3D time-of-flight magnetic resonance angiography study

Abstract



Background
Preterm birth (< 37 weeks’ gestation) alters cerebrovascular development due to the premature transition from a foetal to postnatal circulatory system, with potential implications for future cerebrovascular health. This study aims to explore potential differences in the Circle of Willis (CoW), a key arterial ring that perfuses the brain, of healthy adults born preterm.
Methods
A total of 255 participants (108 preterm, 147 full-term) were included in the analysis. High-resolution three-dimensional Time-of-Flight Magnetic Resonance Angiography (3D TOF MRA) datasets were analysed, measuring vessel diameters and classifying segments into different groups of CoW anatomical variations. Statistical comparisons assessed the prevalence of each variant group between preterm and full-term populations, as well as the relationship between CoW variability, sex, and degree of prematurity.
Results
We identified 164 participants with variant CoW configurations. Unilateral segment hypoplasia (30%) and unilateral segment absence (29%) were the most common variations, with over 50% related to the posterior communicating artery (PComA). However, the incidence of absent segments was lower in preterm adults, who were more likely to exhibit variants associated with complete CoW configurations compared to full-term adults (p = 0.025). Preterm males had a higher probability of a group 1 variant (circles with one or more hypoplastic segments only) than the full-term group (p = 0.024). In contrast, preterm females showed higher odds of a group 4a variant (circles with one or more accessory segments, without any absent segments) in comparison to their full-term counterparts (p = 0.020).
Conclusions
Preterm birth is linked to a distinct vascular phenotype of CoW in adults born preterm, with a higher likelihood of a CoW configuration with hypoplastic segments but a lower likelihood of absent segments. Future work should focus on larger prospective studies and explore the implications of these findings for normal development and cerebrovascular disease. Furthermore, TOF MRA might be a useful adjunct in the neurovascular assessment of preterm-born individuals.

Published in BMC Medical Imaging, 2025

Download here

Matrix mechano-sensing at the invasive front induces a cytoskeletal and transcriptional memory supporting metastasis

Abstract



The extracellular matrix (ECM) controls tumour dissemination. We characterise ECM organization in human and mouse tumours, identifying three regions: tumour body, proximal invasive front and distal invasive front. Invasive areas show increased matrix density, fibre thickness, length, and alignment, with unique radial fibre orientation at the distal invasive front correlating with amoeboid invasive features. Using patient samples and murine models, we find that metastases recapitulate ECM features of the primary tumour. Ex vivo culture of murine cancer cells isolated from the different tumour regions reveals a spatial cytoskeletal and transcriptional memory. Several in vitro models recapitulate the in vivo ECM organisation showing that increased matrix induces 3D confinement supporting Rho-ROCK-Myosin II activity, while radial orientation enhances directional invasion. Spatial transcriptomics identifies a mechano-inflammatory program associated with worse prognosis across multiple tumour types. These findings provide mechanistic insights into how ECM organization shapes local invasion and distant metastasis.

Published in Nature Communications, 2025

Download here

Fibre tracing in biomedical images: An objective comparison between seven algorithms

Abstract



Obtaining the traces and the characteristics of elongated structures is an important task in computer vision pipelines. In biomedical applications, the analysis of traces of vasculature, nerves or fibres of the extracellular matrix can help characterise processes like angiogenesis or the effect of a certain treatment. This paper presents an objective comparison of six existing methodologies (Edge detection, CT Fire, Scale Space, Twombli, U-Net and Graph Based) and one novel approach called Trace Ridges to trace biomedical images with fibre-like structures. Trace Ridges is a fully automatic and fast algorithm that combines a series of image-processing algorithms including filtering, watershed transform and edge detection to obtain an accurate delineation of the fibre-like structures in a rapid time. To compare the algorithms, four biomedical data sets with vastly distinctive characteristics were selected. Ground truth was obtained by manual delineation of the fibre-like structures. Three pre-processing filtering options were used as a first step: no filtering, Gaussian low-pass and DnCnn, a deep-learning filtering. Three distance error metrics (total, average and maximum distance from the obtained traces to the ground truth) and processing time were calculated. It was observed that no single algorithm outperformed the others in all metrics. For the total distance error, which was considered the most significative, Trace Ridges ranked first, followed by Graph Based, U-Net, Twombli, Scale Space, CT Fire and Edge Detection. In terms of speed, Trace Ridges ranked second, only slightly slower than Edge Detection. Code is freely available at github.com/youssefarafat/Trace_Ridges.

Published in PLoS ONE, 2025

Download here

research

Cancer

Cancer

Published:

Circulation

Circulation and Microcirculation

Published:

Phonetics

Phonetics

Published:

Texture

Texture

Published:

resources

teaching

Teaching experience 1

This is a description of a teaching experience. You can use markdown like any other post.

Undergraduate course, University 1, Department, 2014

Teaching experience 2

This is a description of a teaching experience. You can use markdown like any other post.

Workshop, University 1, Department, 2015

team

techniques