Дисертації з теми "Microscopy images"

Modern research in microscopy image processing requires a deeper understanding of the influence of different factors on registration of this type of biomedical images. Analysis of this process requires smart software which should be able to obtain quantitative parameters of micro objectives with acceptable processing speed.

Image Super resolution is a widely-studied problem in computer vision, where the objective is to convert a lowresolution image to a high resolution image. Conventional methods for achieving super-resolution such as image priors, interpolation, sparse coding require a lot of pre/post processing and optimization. Recently, deep learning methods such as convolutional neural networks and generative adversarial networks are being used to perform super-resolution with results competitive to the state of the art but none of them have been used on microscopy images. In this thesis, a generative adversarial network, mSRGAN, is proposed for super resolution with a perceptual loss function consisting of a adversarial loss, mean squared error and content loss. The objective of our implementation is to learn an end to end mapping between the low / high resolution images and optimize the upscaled image for quantitative metrics as well as perceptual quality. We then compare our results with the current state of the art methods in super resolution, conduct a proof of concept segmentation study to show that super resolved images can be used as a effective pre processing step before segmentation and validate the findings statistically.
Image Super-resolution är ett allmänt studerad problem i datasyn, där målet är att konvertera en lågupplösningsbild till en högupplöst bild. Konventionella metoder för att uppnå superupplösning som image priors, interpolation, sparse coding behöver mycket föroch efterbehandling och optimering.Nyligen djupa inlärningsmetoder som convolutional neurala nätverk och generativa adversariella nätverk är användas för att utföra superupplösning med resultat som är konkurrenskraftiga mot toppmoderna teknik, men ingen av dem har använts på mikroskopibilder. I denna avhandling, ett generativ kontradiktorisktsnätverk, mSRGAN, är föreslås för superupplösning med en perceptuell förlustfunktion bestående av en motsatt förlust, medelkvadratfel och innehållförlust.Mål med vår implementering är att lära oss ett slut på att slut kartläggning mellan bilder med låg / hög upplösning och optimera den uppskalade bilden för kvantitativa metriks såväl som perceptuell kvalitet. Vi jämför sedan våra resultat med de nuvarande toppmoderna metoderna i superupplösning, och uppträdande ett bevis på konceptsegmenteringsstudie för att visa att superlösa bilder kan användas som ett effektivt förbehandling steg före segmentering och validera fynden statistiskt.

The purpose of this thesis is to develop a mathematical approach and associated software implementation for deconvolution of two-dimensional Transmission Electron Microscope (TEM) images. The focus is on TEM images of weakly scattering amorphous biological specimens that mainly produce phase contrast. The deconvolution is to remove the distortions introduced by the TEM detector that are modeled by the Modulation Transfer Function (MTF). The report tests deconvolution of the TEM detector MTF by Wiener _ltering and Tikhonov regularization on a range of simulated TEM images with varying degree of noise.The performance of the two deconvolution methods are quanti_ed by means of Figure of Merits (FOMs) and comparison in-between methods is based on statistical analysis of the FOMs.

We propose a test-bed application for synthesis and analysis of multi-layeredmicroscopy data with variation in depth of focus(DOF), where we considerthe problem of detecting object overlap.For the synthesis part, the objects are elliptical in appearance with the possibilityof setting dierent parameters like noise, resolution, illumination,circularity, area and orientation.For the analysis part, the approach allows the user to set several parameters,including sensitivity for error calculation and classier type for analysis.We provide a novel algorithm that exploits the multi-layered nature of theobject overlap problem in order to improve recognition. The variation of grayvalue for each pixel in dierent depth is used as feature source for classication.The classier divides the pixels in three dierent groups: backgroundpixels, pixels in single cells and pixels in overlapping parts.We provide experimental results on the synthesized data, where we add noiseof dierent density. In non-noisy environments the performance for accuracyof overlapping positions is 93% and the performance of the missed overlapsis around 99.98% for density of 150 cells.iv

In Contact Mode Atomic Force Microscopy (CM-AFM), a cantilever with a sharp tip on its end is employed to acquire topographic information. Such acquisition is normally made by monitoring the deflection of the cantilever when it is in contact with the surface being scanned and using deflection variations as a feedback signal to the control electronics in order to keep the deflection constant (also known as constant force imaging mode in the literature). However, there is a major problem with this approach since, in most cases, a constant force scanning is not possible: frictional forces, besides normal forces, may bend the cantilever. Such additional bending (deflection) needs to be considered in the formulation of the problem. The present dissertation investigates how these forces (frictional and normal) can give rise to a topographic artifact when scanning along the cantilever axis direction. Such artifact is even more dramatic when the friction coefficient of the sample changes from region to region. This effect is studied experimentally, with a sample composed of graphene monolayer atop silicon oxide. The observed artifact, caused by frictional forces, causes the graphene to appear either thicker or thinner than it really is depending on scan direction. A theoretical examination is also made both with analytical methods (Euler-Bernoulli beam theory) and a simulation on COMSOL Multiphysics package. The theory not only predicts the artifact, but also indicates how it can be completely avoided by changing the scanning angle to the perpendicular direction of the cantilever axis.
No Modo Contato da Microscopia de Força Atômica (CM-AFM), uma alavanca com uma ponta bastante afiada em sua extremidade é usada para adquirir informação topográfica. Tal aquisição normalmente é feita monitorando a deflexão da alavanca quando em contato com a superfície a ser varrida. Usa-se a variação da deflexão como um sinal de feedback que controla a eletrônica, mantendo a deflexão constante (conhecido como modo de força constante na literatura). Porém, existe um grande problema com essa abordagem, já que, na maioria dos casos, fazer uma varredura com força constante não é possível: forças de atrito, além da força normal, podem fletir a alavanca. Tal curvatura adicional (deflexão) deve ser considerada na formulação do problema. Essa dissertação investiga como essas forças (normal e de atrito) podem dar origem a um artefato de topografia quando é feito uma varredura ao longo do eixo da alavanca. Tal artefato é ainda mais dramático quando o coeficiente de atrito da amostra muda de região para região. Esse efeito é estudado experimentalmente, com uma amostra composta de uma monocamada de grafeno em cima de oxido de silício. O artefato observado, causado pelas forças de atrito, faz o grafeno aparecer mais espesso ou mais estreito do que realmente é, dependendo da direção de varredura. Uma verificação teórica também é feita usando métodos analíticos (teoria de vigas de Euler-Bernoulli) e simulações usando o pacote COMSOL Multiplysics. A teoria não apenas prediz o artefato, mas também indica como ele pode ser completamente evitado ao trocar o ângulo de varredura para perpendicular à direção do eixo da alavanca.

Thesis (M.S.)--University of Missouri--Rolla, 2007.
Vita. The entire thesis text is included in file. Title from title screen of thesis/dissertation PDF file (viewed December 7, 2007) Includes bibliographical references (p. 89-90).

Submitted by PIETRO CARELLI REIS DE OLIVEIRA CALTABIANO null (pietrocarelli@hotmail.com) on 2017-01-10T01:33:15Z No. of bitstreams: 1 Caltabiano_Pietro_Tese_2016.pdf: 23652706 bytes, checksum: 3e2ad76601dc03b6f3704911772a5005 (MD5)
Approved for entry into archive by LUIZA DE MENEZES ROMANETTO (luizamenezes@reitoria.unesp.br) on 2017-01-12T19:31:25Z (GMT) No. of bitstreams: 1 caltabiano_pcro_dr_guara.pdf: 23652706 bytes, checksum: 3e2ad76601dc03b6f3704911772a5005 (MD5)
Made available in DSpace on 2017-01-12T19:31:25Z (GMT). No. of bitstreams: 1 caltabiano_pcro_dr_guara.pdf: 23652706 bytes, checksum: 3e2ad76601dc03b6f3704911772a5005 (MD5) Previous issue date: 2016-11-25
Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)
Ferramentas de processamento e análise digital de imagens foram desenvolvidas com a finalidade de avaliar a evolução da textura morfológica e cristalográfica da microestrutura da liga de alumínio 7075 sob diferentes níveis de deformação plástica por compressão uniaxial. Amostras da liga de alumínio 7075-T6 passaram por um processo de recozimento pleno seguido de um estágio de compressão uniaxial, obtendo níveis de deformações entre 25 e 65%. As microestruturas das amostras foram avaliadas em função dos parâmetros morfológicos dos precipitados, da reorientação dos planos cristalográficos dos grãos e da orientação das subestruturas formadas durante o processo de deformação. Para a caracterização foram utilizadas técnicas de difração de raios-X, microscopia eletrônica, microscopia óptica utilizando técnicas de polarização linear e circular, microscopia correlativa e processamento digital de imagens. Os resultados de difração de raios-X indicaram uma reorientação do plano cristalográficos (200) para o (220) após a deformação, e as técnicas de microscopia eletrônica identificaram precipitados de Mg2Si, Al7Cu2Fe e Al6(FeCu) na liga. A análise morfológica dos precipitados indicou uma maior fragmentação dos precipitados devido à maior ativação do plano (331) a partir de 39% de deformação. Por meio do processamento de imagens foi encontrada uma tendência de correlação entre os planos cristalográficos e a fração de área das fases, enquanto que os parâmetros morfológicos das subestruturas formadas durante o processo de deformação permitiram avaliar apenas qualitativamente o nível de encruamento das amostras.
Digital image processing and analysis tools were developed to perform the AA 7075 crystallographic and morphologic texture evaluation under different level of plastic deformation by uniaxial compression. Samples of AA 7075-T6 were submitted to full annealing process followed by uniaxial compression, thus obtaining deformations between 25 and 65% of thickness. The samples microstructure evaluation was performed considering: precipitates morphological parameters, crystallographic lattices reorientation and deformation substructure orientation. The characterization technics were: X-ray diffraction, electron microscopy, optical microscopy with polarization light, correlative microscopy and digital images processing. Xray diffraction results showed that the crystallographic plane (200) was reoriented to (220) after compression. The EDS analysis identified precipitates of Mg2Si, Al7Cu2Fe e Al6(FeCu). The precipitates morphological analysis showed an increase in fragmentation due to plane (331) at 39% of deformation. The digital image process of the samples etched with Barker reagents indicated a correlation between area fraction and the diffraction peaks, and the deformation substructures analysis made viable a qualitative characterization of the hardening process.
FAPESP: 2011/00403-2

Thesis (M. Eng.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, June 2006.
Includes bibliographical references (leaf 41).
Cell morphology of tissue is naturally three-dimensional. Most current methods for tissue analysis use two dimensional histological images of the tissue samples, restricting the analysis to 2D. Existing approaches do not provide essential three-dimensional information such as cell volume, shape and structural orientation of cells within the tissue. This thesis investigates a method to extract three dimensional data using two-dimensional microscopy. We demonstrate that three dimensional cell structure can be acquired using two dimensional fluorescence microscopy and two-photon microscopy and explore the application of the analysis to studies of cardiac tissue.
by Francisco Cruz.
M.Eng.

Electron crystallography is one of the methods for determining the structure of membrane proteins. However, the resolution that we get from electron crystallography of membrane proteins is limited by crystal disorder and inaccurate determination of contrast transfer function (CTF) parameters. To overcome these problems, we applied single particle refinement with local averaging for long range variation of the crystals and local region-based CTF correction for the tilted images. These two corrections are done on the Melibiose Permease (MelB) data sets and a resolution of 8.6Å is reported.
Elektronkristallografi är en av metoderna för att bestämma strukturen av membranproteiner.Den upplösning vi kan få från elektronkristallografi av membranproteiner begränsas av oordning i kristallen och felaktig bestämning av kontrastöverföringsfunktioner (CTF). För att lösa dessa problem tillämpade vi single particle refinement for lokala medelvärden av kristaller och lokala regionsbaserad CTF korrigering av tiltade dataset. Dessa två korrektioner görs på bilder av melibiospermeas (Melb) kristaller och en upplösning på 8.6Å erhölls.

It has been very good presentation with some comments on CTF measurment

Scanning probe microscopy allows the investigation and manipulation of matter at the atomic and molecular level, and is crucial in the development of new and novel techniques within nanoscience. However, to understand the information obtained from the various forms of scanning probe microscopy, a thorough theoretical understanding is necessary. Often this theoretical background is provided through density functional theory, which, while incredibly powerful, has limitations with regards to the size and complexity of the systems in which it can investigate. Thus, for more complicated systems, alternative techniques are desirable to be used both independently and alongside density functional theory. In this work, theoretical techniques are constructed that allow the information obtained from both scanning tunnelling microscopy and atomic force microscopy to be investigated for a variety of systems. These techniques are all based around Huckel molecular orbital theory or extended Huckel molecular orbital theory, and use a simple linear combination of atomic orbital basis, that allows rapid analysis of various systems. The main focus of the work is the scanning probe microscopy of the C60 fullerene molecule. Theoretical scanning tunnelling microscopy images are constructed for the cases where C60 is adsorbed on both the substrate and the scanning probe in the form of a functionalised tip, as well as when a tip-adsorbed molecule interacts with a sample-adsorbed molecule. The atomic force microscopy images of surface adsorbed C60 are considered, with the main focus centred on the repulsive interaction observed due to the Pauli exclusion principle. The structure of the scanning probe, and the effect this has on this imaging is examined, as well as considering the atomic force microscopy images obtained when two C60s interact. Molecules other than C60 are also considered, with the techniques developed used to interpret and understand the atomic force microscopy images obtained when a pentacene and a PTCDA molecule interact with a carbon monoxide functionalised tip. The theoretical work is accompanied throughout by a variety of experimental work, both from previously published work, and from unpublished work obtained by the University of Nottingham nanoscience group. Much focus is given to the interaction between C60 and the Si(111)-(7x7) reconstruction, both in the sense of a functionalised tip interacting with the surface, and with the interactions present where a C60 is adsorbed onto a surface. In doing so, previously postulated bonding sites for C60 on this surface have been verified.

Dissertação para obtenção do Grau de Mestre em Engenharia Biomédica
Most of the investigation in microbiology relies on microscope imaging and needs to be complemented with reliable methods of computer assisted image processing, in order to avoid manual analysis. In this work, a method to assist the study of the in vivo kinetics of protein expression from Escherichia coli cells was developed. Confocal fluorescence microscopy (CFM) and Differential Interference Contrast (DIC) microscopy images were acquired and processed using the developed method. This method comprises two steps: the first one is focused on the cells detection using DIC images. The latter aligns both DIC and CFM images and computes the fluorescence level emitted by each cell. For the first step, the Gradient Path Labelling (GPL) algorithm was used which produces a moderate over-segmented DIC image. The proposed algorithm, based on decision trees generated by the Classification and Regression Trees (CART) algorithm, discards the backgrounds regions and merges the regions belonging to the same cell. To align DIC/fluorescence images an exhaustive search of the relative position and scale parameters that maximizes the fluorescence inside the cells is made. After the cells have been located on the CFM images, the fluorescence emitted by each cell is evaluated. The discard classifier performed with an error rate of 1:81% 0:98% and the merge classifier with 3:25% 1:37%. The segmentation algorithm detected 93:71% 2:06% of the cells in the tested images. The tracking algorithm correctly followed 64:52% 16:02% of cells and the alignment method successfully aligned all the tested images.

Thin film arrays of molecules or supramolecules are active subjects of investigation because of their potential value in electronics, chemical sensing, catalysis, and other areas. Scanning probe microscopes (SPMs), including scanning tunneling microscopes (STMs) and atomic force microscopes (AFMs) are commonly used for the characterization and metrology of thin film arrays. As opposed to transmission electron microscopy (TEM), SPMs have the advantage that they can often make observations of thin films in air or liquid, while TEM requires highly specialized techniques if the sample is to be in anything but vacuum. SPM is a surface imaging technique, while TEM typically images a 2D projection of a thin 3D sample. Additionally, variants of SPM can make observations of more than just topography; for instance, magnetic force microscopy measures nanoscale magnetic properties. Thin film arrays are typically two-dimensionally periodic. A perfect, infinite two-dimensionally periodic array is mathematically constrained to belong to one of only 17 possible 2D plane symmetry groups. Any real image is both finite and imperfect. Crystallographic Image Processing (CIP) is an algorithm that Fourier transforms a real image into a 2D array of complex numbers, the Fourier coefficients of the image intensity, and then uses the relationship between those coefficients to first ascertain the 2D plane symmetry group that the imperfect, finite image is most likely to possess, and then adjust those coefficients that are symmetry-related so as to perfect the symmetry. A Fourier synthesis of the symmetrized coefficients leads to a perfectly symmetric image in direct space (when accumulated rounding and calculation errors are ignored). The technique is, thus, an averaging technique over the direct space experimental data that were selected from the thin film array. The image must have periodicity in two dimensions in order for this technique to be applicable. CIP has been developed over the past 40 years by the electron crystallography community, which works with 2D projections from 3D samples. Any periodic sample, whether it is 2D or 3D has an "ideal structure" which is the structure absent any crystal defects. The ideal structure can be considered one average unit cell, propagated by translation into the whole sample. The "real structure" is an actual sample containing vacancies, dislocations, and other defects. Typically the goal of electron and other types of microscopy is examination of the real structure, as the ideal structure of a crystal is already known from X-ray crystallography. High resolution transmission electron microscope image based electron crystallography, on the other hand, reveals the ideal crystal structure by crystallographic averaging. The ideal structure of a 2D thin film cannot be easily in a spatially selective fashion examined by grazing incidence X-ray or low energy electron diffraction based crystallography. SPMs straightforwardly observe thin films in direct space, but SPM accuracy is hampered by blunt or multiple tips and other unavoidable instrument errors. Especially since the film is often of a supramolecular system whose molecules are weakly bonded (via pi bonds, hydrogen bonds, etc.) both to the substrate and to each other, it is relatively easy for a molecule from the film to adhere to the scanning tip during the scan and become part of the tip during subsequent observation. If the thin film array has two-dimensional periodicity, CIP is a unique and effective tool both for image enhancement (determination of ideal structure) and for the quantification of overall instrument error. In addition, if a sample of known 2D periodicity is scanned, CIP can return information about the contribution of the instrument itself to the image. In this thesis we show how the technique is applied to images of two dimensionally periodic samples taken by SPMs. To the best of our knowledge, this has never been done before. Since 2D periodic thin film arrays have an ideal structure that is mathematically constrained to belong to one of the 17 plane symmetry groups, we can use CIP to determine that group and use it for a particularly effective averaging algorithm. We demonstrate that the use of this averaging algorithm removes noise and random error from images more effectively than translational averaging, also known as "lattice averaging" or "Fourier filtering". We also demonstrate the ability to correct systematic errors caused by hysteresis in the scanning process. These results have the effect of obtaining the ideal structure of the sample, averaging out the defects crystallographically, by providing an average unit cell which, when translated, represents the ideal structure. In addition, if one has recorded a scanning probe image of a 2D periodic sample of known symmetry, we demonstrate that it is possible to use the Fourier coefficients of the image transform to solve the inverse problem and calculate the point spread function (PSF) of the instrument. Any real scanning probe instrument departs from the ideal PSF of a Dirac delta function, and CIP allows us to quantify this departure as far as point symmetries are concerned. The result is a deconvolution of the "effective tip", which includes any blunt or multiple tip effects, as well as the effects caused by adhesion of a sample molecule to the scanning tip, or scanning irregularities unrelated to the physical tip. We also demonstrate that the PSF, once known, can be used on a second image taken by the same instrument under approximately the same experimental conditions to remove errors introduced during that second imaging process. The preponderance of two-dimensionally periodic samples as subjects of SPM observation makes the application of CIP to SPM images a valuable technique to extract a maximum amount of information from these images. The improved resolution of current SPMs creates images with more higher-order Fourier coefficients than earlier, "softer" images; these higher-order coefficients are especially amenable to CIP, which can then effectively magnify the resolution improvement created by better hardware. The improved resolution combined with the current interest in supramolecular structures (which although 3D usually start building on a 2D periodic surface) appears to provide an opportunity for CIP to significantly contribute to SPM image processing.

Motivation: Cartilage is a robust but flexible connective tissue found in most joints of the body. The collagen fibres present in the extracellular matrix of cartilage contribute to its tensile strength and stiffness. The purpose of this study is to develop and implement methods to determine the orientation and anisotropy of collagen fibres in 3-D images gen- erated with multi-photon microscopy. The motivation behind developing these techniques is to improve the foundation for further studies on understanding the characteristics of the cartilage matrix. This in turn would give a better foundation for developing artificial matrices and mechanical models, as well as improve diagnostics.Material and methods: The two methods developed in this study are based on analysing the frequency domain. One is an expansion of a previous developed method by Chaudhuri et al. [1]. This method is based on evaluating the average intensity at different directions in the frequency domain. The direction with the least average intensity is equivalent to the direction of the fibres. The other method is based on thresholding the frequency domain according to intensity followed by fitting an ellipsoid to the remaining data set. The direction of the collagen fibres is equivalent to the direction of the shortest axis of the ellipsoid. These methods are called the sector and ellipsoid method, respectively. To determine how robust these methods are a series of tests were developed. The focus of these tests was to determine if the methods are rotational invariant and if the results are influences by different preprocessing techniques. These preprocessing techniques are: median filtering, deconvolution and skeletonization of the original image containing the collagen fibres. It is also important to determine the sensitivity of the ellipsoid method according to the chosen threshold value. In addition data generated fibres and frequency domains were made to determine the accuracy of the methods.Results and conclusion: The sector method was not very robust. For most cases there is not one specific direction that has the least average intensity in the frequency domain. Instead there is a quite large minimum area. The ellipsoid method shows promising results. It managed to find the correct direction both for the data generated data sets, but also for the real images. It seems like no preprocessing nor frequency filtering, except for thresholding, is needed to still find the correct direction and its anisotropy. The only remark is that the automatically chosen threshold value was to low for one of the samples. This can probably be improved by making a slight change in the process for choosing a threshold value.

Analysing migrating cells in microscopy time-lapse images has already helped the understanding of many biological processes and may be of importance in the development of new medical treatments. Today’s biological experiments tend to produce a huge amount of dynamic image data and tracking the individual cells by hand has become a bottleneck for the further analysis work. A number of cell tracking methods have therefore been developed over the past decades, but still many of the techniques have a limited performance. The aim of this Master Project is to develop a particle filter algorithm that automatically detects and tracks a large number of individual cells in an image sequence. The solution is based on a Rao-Blackwellized particle filter for multiple object tracking. The report also covers a review of existing automatic cell tracking techniques, a review of well-known filter techniques for single target tracking and how these techniques have been developed to handle multiple target tracking. The designed algorithm has been tested on real microscopy image data of neutrophils with 400 to 500 cells in each frame. The designed algorithm works well in areas of the images where no cells touch and can in these situations also correct for some segmentation mistakes. In areas where cells touch, the algorithm works well if the segmentation is correct, but often makes mistakes when it is not. A target effectiveness of 77 percent and a track purity of 80 percent are then achieved.

Thesis (M. Eng.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 2006.
Includes bibliographical references.
In this project, we have proposed a decorrelator-based single antenna interference cancellation algorithm for the asynchronous GSM/GPRS network. The algorithm is tested according to the current SAIC/DARP performance requirement in the computer simulation, and is shown to give various gains in different test scenarios.
by Sheldon Y. Chan.
M.Eng.

The aim of blob detection is to find regions in a digital image that differ from their surroundings with respect to properties like intensity or shape. Bio-image analysis is a common application where blobs can denote regions of interest that have been stained with a fluorescent dye. In image-based in situ sequencing for ribonucleic acid (RNA) for example, the blobs are local intensity maxima (i.e. bright spots) corresponding to the locations of specific RNA nucleobases in cells. Traditional methods of blob detection rely on simple image processing steps that must be guided by the user. The problem is that the user must seek the optimal parameters for each step which are often specific to that image and cannot be generalised to other images. Moreover, some of the existing tools are not suitable for the scale of the microscopy images that are often in very high resolution and 3D. Machine learning (ML) is a collection of techniques that give computers the ability to ”learn” from data. To eliminate the dependence on user parameters, the idea is applying ML to learn the definition of a blob from labelled images. The research question is therefore how ML can be effectively used to perform the blob detection. A blob detector is proposed that first extracts a set of relevant and nonredundant image features, then classifies pixels as blobs and finally uses a clustering algorithm to split up connected blobs. The detector works out-of-core, meaning it can process images that do not fit in memory, by dividing the images into chunks. Results prove the feasibility of this blob detector and show that it can compete with other popular software for blob detection. But unlike other tools, the proposed blob detector does not require parameter tuning, making it easier to use and more reliable.
Syftet med blobdetektion är att hitta regioner i en digital bild som skiljer sig från omgivningen med avseende på egenskaper som intensitet eller form. Biologisk bildanalys är en vanlig tillämpning där blobbar kan beteckna intresseregioner som har färgats in med ett fluorescerande färgämne. Vid bildbaserad in situ-sekvensering för ribonukleinsyra (RNA) är blobbarna lokala intensitetsmaxima (dvs ljusa fläckar) motsvarande platserna för specifika RNA-nukleobaser i celler. Traditionella metoder för blob-detektering bygger på enkla bildbehandlingssteg som måste vägledas av användaren. Problemet är att användaren måste hitta optimala parametrar för varje steg som ofta är specifika för just den bilden och som inte kan generaliseras till andra bilder. Dessutom är några av de befintliga verktygen inte lämpliga för storleken på mikroskopibilderna som ofta är i mycket hög upplösning och 3D. Maskininlärning (ML) är en samling tekniker som ger datorer möjlighet att “lära sig” från data. För att eliminera beroendet av användarparametrar, är tanken att tillämpa ML för att lära sig definitionen av en blob från uppmärkta bilder. Forskningsfrågan är därför hur ML effektivt kan användas för att utföra blobdetektion. En blobdetekteringsalgoritm föreslås som först extraherar en uppsättning relevanta och icke-överflödiga bildegenskaper, klassificerar sedan pixlar som blobbar och använder slutligen en klustringsalgoritm för att dela upp sammansatta blobbar. Detekteringsalgoritmen fungerar utanför kärnan, vilket innebär att det kan bearbeta bilder som inte får plats i minnet genom att dela upp bilderna i mindre delar. Resultatet visar att detekteringsalgoritmen är genomförbar och visar att den kan konkurrera med andra populära programvaror för blobdetektion. Men i motsats till andra verktyg behöver den föreslagna detekteringsalgoritmen inte justering av sina parametrar, vilket gör den lättare att använda och mer tillförlitlig.

Chaine d'acquisition du signal image du mebis. Etude des sources de bruit associees aux detecteurs a semiconducteur et mise au point d'un processus de realisation de diodes de detection a barriere de surface. Conception d'une carte electronique compatible avec un microordinateur. Cette carte permet la numerisation, le stockage sur disquette et la visualisation des images fournies par le mebis

La détection et la caractérisation d’objets dans des images à faible rapport signal sur bruit est un problème courant dans de nombreux domaines tels que l’astronomie ou la microscopie. En astronomie, la détection des exoplanètes et leur caractérisation par imagerie directe depuis la Terre sont des sujets de recherche très actifs. Une étoile cible et son environnement proche (abritant potentiellement des exoplanètes) sont observés sur de courtes poses. En microscopie, l’holographie en ligne est une méthode de choix pour caractériser à faibles coûts les objets microscopiques. Basée sur l’enregistrement d’un hologramme, elle permet une mise au point numérique dans n’importe quel plan du volume 3-D imagé. Dans ces deux applications cibles, le problème est rendu difficile par le faible contraste entre les objets et le fond non stationnaire des images enregistrées.Dans cette thèse, nous proposons un algorithme non-supervisé dédié à la détection et à la caractérisation d’exoplanètes par une modélisation statistique des fluctuations du fond. Cette méthode est basée sur une modélisation de la distribution statistique des données à une échelle locale de patchs, capturant ainsi leur covariances spatiales. Testé sur plusieurs jeux de données de l’imageur haut-contraste SPHERE opérant au Très Grand Télescope Européen, cet algorithme atteint de meilleures performances que les méthodes de l’état de l’art. En particulier, les cartes de détection produites sont stationnaires et statistiquement fondées. La détection des exoplanètes peut ainsi être effectuée à probabilité de fausse alarme contrôlée. L’estimation de la distribution d’énergie spectrale des sources détectées est également non biaisée. L’utilisation d’un modèle statistique permet également de déduire des précisions photométriques et astrométriques fiables. Ce cadre méthodologique est ensuite adapté pour la détection de motifs spatialement étendus tels que les motifs de diffraction rencontrés en microscopie holographique qui sont également dominés par un fond non-stationnaire. Nous proposons aussi des approches robustes basées sur des stratégies de pondération afin de réduire l’influence des nombreuses valeurs aberrantes présentes sur les données réelles. Nous montrons sur des vidéos holographiques que les méthodes de pondération proposées permettent d’atteindre un compromis biais/variance. En astronomie, la robustesse améliore les performances de détection, en particulier à courtes séparations angulaires, où les fuites stellaires dominent. Les algorithmes développés sont également adaptés pour tirer parti de la diversité spectrale des données en plus de leur diversité temporelle, améliorant ainsi leurs performances de détection et de caractérisation. Tous les algorithmes développés sont totalement non-supervisés: les paramètres de pondération et/ou de régularisation sont estimés directement à partir des données. Au-delà des applications considérées en astronomie et en microscopie, les méthodes de traitement du signal introduites dans cette thèse sont générales et pourraient être appliquées à d’autres problèmes de détection et d’estimation
Detecting and characterizing objects in images in the low signal-to-noise ratio regime is a critical issue in many areas such as astronomy or microscopy. In astronomy, the detection of exoplanets and their characterization by direct imaging from the Earth is a hot topic. A target star and its close environment (hosting potential exoplanets) are observed on short exposures. In microscopy, in-line holography is a cost-effective method for characterizing microscopic objects. Based on the recording of a hologram, it allows a digital focusing in any plane of the imaged 3-D volume. In these two fields, the object detection problem is made difficult by the low contrast between the objects and the nonstationary background of the recorded images.In this thesis, we propose an unsupervised exoplanet detection and characterization algorithm based on the statistical modeling of background fluctuations. The method, based on a modeling of the statistical distribution of patches, captures their spatial covariances. It reaches a performance superior to state-of-the-art techniques on several datasets of the European high-contrast imager SPHERE operating at the Very Large Telescope. It produces statistically grounded and spatially-stationary detection maps in which detections can be performed at a constant probability of false alarm. It also produces photometrically unbiased spectral energy distributions of the detected sources. The use of a statistical model of the data leads to reliable photometric and astrometric accuracies. This methodological framework can be adapted to the detection of spatially-extended patterns in strong structured background, such as the diffraction patterns in holographic microscopy. We also propose robust approaches based on weighting strategies to reduce the influence of the numerous outliers present in real data. We show on holographic videos that the proposed weighting approach achieves a bias/variance tradeoff. In astronomy, the robustness improves the performance of our detection method in particular at close separations where the stellar residuals dominate. Our algorithms are adapted to benefit from the possible spectral diversity of the data, which improves the detection and characterization performance. All the algorithms developed are unsupervised: weighting and/or regularization parameters are estimated in a data-driven fashion. Beyond the applications in astronomy and microscopy, the signal processing methodologies introduced are general and could be applied to other detection and estimation problems

Orientador: Carlos Lenz Cesar
Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Fisica Gleb Wataghin
Made available in DSpace on 2018-08-11T10:58:51Z (GMT). No. of bitstreams: 1 Thomaz_AndreAlexandrede_M.pdf: 10062018 bytes, checksum: 1e19c55cb5a4e709c2015e2d90f3ac13 (MD5) Previous issue date: 2007
Resumo: A pesquisa em fotônica biomedica está claramente tomando a direção do entendimento de processos biológicos a nível celular. A resolução necessária para atingir esse objetivo requer praticamente ferramentas fotônicas. Contudo, uma integração de diferentes ferramentes fotônicas e uma aproximação funcional serão necessárias para acessar os processos biomecânicos e bioquímicos celulares. Deste modo nós podemos observar eventos bioquímicos disparados mecanicamente ou eventos mecânicos disparados bioquimicamente, ou até mesmo observar simultâneamente eventos biomecânicos e bioquímicos disparados por outros meios, entre outros, eletricamente. Uma das grandes vantagens das ferramentas fotônicas é a sua facilidade de integração. Nós desenvolvemos uma ferramenta integrada incorporando pinça óptica simples com Microscopia Confocal "Single-photon" e Multifóton. O sistema consegue realizar microscopias de fluorescência excitada pela absorção de dois fótons e geração de segundo harmônico em conjunto com manipulações ópticas. Medidas de força, elasticidade e viscosidade de membranes esticadas podem ser monitoradas em tempo real pelas microscopias confocais, bem como protozoários capturados opticamente, como, por exemplo, Trypanosoma cruzi. Nós mostraremos vários exemplos do uso de tal ferramenta integrada e seu potencial para observar processos mecânicos e bioquímicos a nível celular
Abstract: The research in biomedical photonics is clearly evolving in the direction of the understanding of biological processes at the cell level. The spatial resolution to accomplish this task practically requires photonics tools. However, an integration of different photonic tools and a multimodal and functional approach will be necessary to access the mechanical and biochemical cell processes. This way we can observe mechanicaly triggered biochemical events or biochemicaly triggered mechanical events, or even observe simultaneously mechanical and biochemical events triggered by other means, e.g. electricaly. One great advantage of the photonic tools is its easiness for integration. Therefore, we developed such integrated tool by incorporating single Optical Tweezers with Confocal Single and Multiphoton Microscopies. This system can perform 2-photon excited fluorescence and Second Harmonic Generation microscopies together with optical manipulations. Force, elasticity and viscosity measurements of stretched membranes can be followed by real time confocal microscopies. Also opticaly trapped living protozoas, such as Trypanosoma cruzi. Integration with CARS microscopy is under way. We will show several examples of the use of such integrated instrument and its potential to observe mechanical and biochemical processes at cell level
Mestrado
Física
Mestre em Física

This thesis contributes to knowledge by developing algorithms that automatically detect and quantify structures of clinical interest in reflectance confocal microscopy (RCM) images, captured in-vivo and from excised skin tissue. The first part of the thesis presents an algorithm that detects the dermalepidermal junction (DEJ), characterised by papillae, in cubes of RCM images of in-vivo skin. A cube of images is a number of mosaic images captured at different depths parallel to the skin surface. A classifier, which makes use of texture and anatomical-based features was designed. The anatomical-based features are parameters that quantify the absence and presence of papillae across different images of the cube. The second part of the thesis analyses RCM images of excised tissue collected during Mohs surgery. These tissue samples include basal cell carcinoma (BCC) and non-diseased samples. An algorithm was developed to differentiate between (i) cancerous regions, (ii) regions of inflammation, and (iii) non-diseased regions. A classifier based on texture and nuclei-concentration features was designed. The nuclei concentration in cancerous sites is different from that in nondiseased sites and thus can be used to distinguish the two. The third part of the thesis analyses RCM video sequences of in-vivo skin imaged at the level of the DEJ. The boundaries of superficial skin capillaries can be delineated by visually observing the highly reflective red blood cells (RBCs) passing through the capillaries. An algorithm that automatically detects skin capillaries in RCM video sequences was developed. Additionally, an algorithm that quantifies the velocity of RBCs in cross-sectionally imaged capillaries is devised. The change in total capillary area (per unit frame area), individual capillary parameters and RBC velocity due to incremental ultra-violet radiation (UVR) doses are analysed in both fair and dark skinned volunteers. The work presented in this thesis has the potential to increase the acceptance of RCM in the dermatology clinic, both for diagnosis and for assessing treatment response of skin conditions located at (or above) the DEJ. Additionally, the thesis enhances the potential of using RCM images of excised samples instead of preparing the tissue for histological examinations during surgery.

Microscopy image processing is an emerging and quickly growing field in medical imaging research area. Recent advancements in technology including higher computation power, larger and cheaper storage modules, and more efficient and faster data acquisition devices such as whole-slide imaging scanners contributed to the recent microscopy image processing research advancement. Most of the methods in this research area either focus on automatically process images and make it easier for pathologists to direct their focus on the important regions in the image, or they aim to automate the whole job of experts including processing and classifying images or tissues that leads to disease diagnosis. This dissertation is consisted of four different frameworks to process microscopy images. All of them include methods for segmentation either as the whole suggested framework or the initial part of the framework for future feature extraction and classification. Specifically, the first proposed framework is a general segmentation method that works on histology images from different tissues and segments relatively solid nuclei in the image, and the next three frameworks work on cervical microscopy images, segmenting cervical nuclei/cells. Two of these frameworks focus on cervical tissue segmentation and classification using histology images and the last framework is a comprehensive segmentation framework that segments overlapping cervical cells in cervical cytology Pap smear images. One of the several commonalities among these frameworks is that they all work at the region level and use different region features to segment regions and later either expand, split or refine the segmented regions to produce the final segmentation output. Moreover, all proposed frameworks work relatively much faster than other methods on the same datasets. Finally, proving ground truth for datasets to be used in the training phase of microscopy image processing algorithms is relatively time-consuming, complicated and costly. Therefore, I designed the frameworks in such a way that they set most (if not all) of the parameters adaptively based on each image that is being processed at the time. All of the included frameworks either do not depend on training datasets at all (first three of the four discussed frameworks) or need very small training datasets to learn or set a few parameters.

Optical Coherence Tomography (OCT) is a noninvasive imaging modality that has significantly contributed to the quantitative assessment of ocular diseases. Another tool available to ophthalmic clinicians is in-vivo confocal microscopy, which allows anatomical structures to be observed live at the cellular level. Incorporating both of these modalities for imaging the cornea allows us to take structural measurements to characterize disease-related changes in corneal anatomy. Notable diseases that directly impact or correlate with corneal structures include glaucoma and diabetic neuropathy. Given glaucoma's impact as the second leading cause of blindness in the world, great efforts have been made in researching and understanding the disease. Correlations have been found between the central corneal thickness (CCT) and the risk of developing visual field loss in patients diagnosed with glaucoma. It should come as no surprise that measuring CCT among glaucoma suspects has also now become a clinical standard of practice. Diabetes is a group of metabolic diseases where the body experiences high blood sugar levels over prolonged periods of time. It is a prominent disease that affects millions of Americans each day. While not necessarily an ocular disease in its own right, it has been shown that diabetes can still affect the corneal structures. Diabetics have decreased corneal sensitivity and a significant link has been established between neuropathic severity in diabetic patients and corneal nerve fiber density. Given the availability of these imaging tools and the significant impact these prominent diseases have on society a growing focus has developed on relating corneal structure measurements and ophthalmic diseases. However, manually acquiring structural measures of the cornea can be a labor intensive and daunting task. Hence, experts have sought to develop automatic alternatives. The goals of our work includes the ability to automatically segment the corneal structures from anterior segment-optical coherence tomography (AS-OCT) and in-vivo confocal microscopy (IVCM) to provide useful structural information from the cornea. The major contributions of this work include 1) utilizing the information of AS-OCT imagery to segment the cornea layers simultaneously in 3D, 2) increasing the region-of-interest of IVCM imagery using a feature-based registration approach to develop a panorama from the images, 3) incorporating machine-learning techniques to segment the corneal nerves in the IVCM imagery, and 4) extracting structural measurements from the segmentation results to determine correlations between the structural measurements known to differ from the corneal structures in various subject groups.

Scanning tunnelling microscopy (STM) and atomic force microscopy (AFM) can produce images of molecules with extremely high resolution. However, Claims that dynamic force microscopy has the capability to resolve intermolecular bonds in real space continue to be vigorously debated. Several studies have now shown that tip flexibility, especially at very close tip-sample separations, is responsible for the striking intra- and intermolecular resolution observed with various scanning probe microscopy techniques. The apparent intermolecular features can be observed with dynamic force microscopy even when no bonding interaction is present, suggesting that such features are in fact an artefact and cannot be interpreted as a real-space image of an intermolecular bond. We have studied the interaction between fullerene (C60) molecules using a sum of pairwise Lennard-Jones (12-6) potentials, and investigated how flexibility in the tip can produce a bond like feature between the molecules in a C60 island where there is no chemical bond present except the weak van der Waals force. We also investigate how the potential between the molecules is dependent on their relative orientations. For a given configuration of the tip and the sample molecules, our results allow us to predict the form of the intermolecular potential that would be observed using non contact atomic force microscopy (NC-AFM). Our study on the Si(111)-(7x7) reconstructed surface using the same model provides a better understating on the origin of ‘sub-atomic’ contrast observed in experiment suggesting that the contrast can arise from a flexible tip exploring an asymmetric potential created due to the positioning of the surrounding surface atoms. We have also simulated NC-AFM images of 2D bi-isonicotinic acid lattice using the same model. The geometry of the lattice have been optimized using DFT before simulating AFM images. Simulation results are in a good agreement with the experiment. The theoretical work is accompanied by a variety of experimental results obtained by the group of Prof Philip Moriarty at the University of Nottingham.

Thesis: Ph. D., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2016.
This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections.
Cataloged from student-submitted PDF version of thesis.
Includes bibliographical references (pages 139-145).
I present a set of fast and scalable algorithms for segmenting very large 3D images of brain tissue. Currently, light and electron microscopy can now produce terascale 3D images within hours. Extracting the information about the shapes and connectivity of the neurons require fast and accurate image segmentation algorithms. Due to the sheer size of the problem, traditional approaches might be computationally infeasible. I focus on an segmentation pipeline that breaks up the segmentation problem into multiple stages, each of which can be improved independently. In the first step of the pipeline, convolutional neural networks are used to predict segment boundaries. Watershed transform is then used to obtain an over-segmentation, which is then reduced using agglomerative clustering algorithms. Finally, manual or computer-assisted proof reading is done by experts. In this thesis, I revisit the traditional approaches for training and applying convolutional neural networks, and propose: - A fast and scalable 3D convolutional network training algorithm suited for multi-core and many-core shared memory machines. The two main quantities of the algorithm are: (1) minimizing the required computation by using FFT-based convolution with memoization, and (2) parallelization approach that can utilize large number of CPUs while minimizing any required synchronization. - A high throughput inference algorithm that can utilize all available computational resources, CPUs and GPUs. I introduce a set of highly parallel algorithms for different layer types and architectures, and show how to combine them to achieve very high throughput. Additionally, I study the theoretical properties of the watershed transform of edge- weighed graphs and propose a liner-time algorithm. I propose a set of modification to the standard algorithm and a quasi-linear agglomerative clustering algorithm that can greatly reduce the over-segmentation produced by the standard watershed algorithm.
by Aleksandar Zlateski.
Ph. D.

Intravital microscopy (IVM) is a well-established imaging technique for real-time monitoring of microscale lung tissue dynamics. Although accepted as a gold standard in respiratory research, its characteristic image features are scarcely understood, especially when trying to determine the actual position of alveolar walls. To allow correct interpretation of these images with respect to the true geometry of the lung parenchyma, we analyzed IVM data of alveoli in a mouse model in comparison with simultaneously acquired optical coherence tomography images. Several IVM characteristics, such as double ring structures or disappearing alveoli in regions of liquid filling, could be identified and related to the position of alveoli relative to each other. Utilizing a ray tracing approach based on an idealized geometry of the mouse lung parenchyma, two major reflection processes could be attributed to the IVM image formation: partial reflection and total internal reflection between adjacent alveoli. Considering the origin of the reflexes, a model was developed to determine the true position of alveolar walls within IVM images. These results allow thorough understanding of IVM data and may serve as a basis for the correction of alveolar sizes for more accurate quantitative analysis within future studies of lung tissue dynamics.

La détection et la caractérisation automatisée des cellules constituent un enjeu important dans de nombreux domaines de recherche tels que la cicatrisation, le développement de l'embryon et des cellules souches, l’immunologie, l’oncologie, l'ingénierie tissulaire et la découverte de nouveaux médicaments. Étudier le comportement cellulaire in vitro par imagerie des cellules vivantes et par le criblage à haut débit implique des milliers d'images et de vastes quantités de données. Des outils d'analyse automatisés reposant sur la vision numérique et les méthodes non-intrusives telles que la microscopie à contraste de phase (PCM) sont nécessaires. Comme les images PCM sont difficiles à analyser en raison du halo lumineux entourant les cellules et de la difficulté à distinguer les cellules individuelles, le but de ce projet était de développer des algorithmes de traitement d'image PCM dans Matlab® afin d’en tirer de l’information reliée à la morphologie cellulaire de manière automatisée. Pour développer ces algorithmes, des séries d’images de myoblastes acquises en PCM ont été générées, en faisant croître les cellules dans un milieu avec sérum bovin (SSM) ou dans un milieu sans sérum (SFM) sur plusieurs passages. La surface recouverte par les cellules a été estimée en utilisant un filtre de plage de valeurs, un seuil et une taille minimale de coupe afin d'examiner la cinétique de croissance cellulaire. Les résultats ont montré que les cellules avaient des taux de croissance similaires pour les deux milieux de culture, mais que celui-ci diminue de façon linéaire avec le nombre de passages. La méthode de transformée par ondelette continue combinée à l’analyse d'image multivariée (UWT-MIA) a été élaborée afin d’estimer la distribution de caractéristiques morphologiques des cellules (axe majeur, axe mineur, orientation et rondeur). Une analyse multivariée réalisée sur l’ensemble de la base de données (environ 1 million d’images PCM) a montré d'une manière quantitative que les myoblastes cultivés dans le milieu SFM étaient plus allongés et plus petits que ceux cultivés dans le milieu SSM. Les algorithmes développés grâce à ce projet pourraient être utilisés sur d'autres phénotypes cellulaires pour des applications de criblage à haut débit et de contrôle de cultures cellulaires.
Automated cell detection and characterization is important in many research fields such as wound healing, embryo development, immune system studies, cancer research, parasite spreading, tissue engineering, stem cell research and drug research and testing. Studying in vitro cellular behavior via live-cell imaging and high-throughput screening involves thousands of images and vast amounts of data, and automated analysis tools relying on machine vision methods and non-intrusive methods such as phase contrast microscopy (PCM) are a necessity. However, there are still some challenges to overcome, since PCM images are difficult to analyze because of the bright halo surrounding the cells and blurry cell-cell boundaries when they are touching. The goal of this project was to develop image processing algorithms to analyze PCM images in an automated fashion, capable of processing large datasets of images to extract information related to cellular viability and morphology. To develop these algorithms, a large dataset of myoblasts images acquired in live-cell imaging (in PCM) was created, growing the cells in either a serum-supplemented (SSM) or a serum-free (SFM) medium over several passages. As a result, algorithms capable of computing the cell-covered surface and cellular morphological features were programmed in Matlab®. The cell-covered surface was estimated using a range filter, a threshold and a minimum cut size in order to look at the cellular growth kinetics. Results showed that the cells were growing at similar paces for both media, but their growth rate was decreasing linearly with passage number. The undecimated wavelet transform multivariate image analysis (UWT-MIA) method was developed, and was used to estimate cellular morphological features distributions (major axis, minor axis, orientation and roundness distributions) on a very large PCM image dataset using the Gabor continuous wavelet transform. Multivariate data analysis performed on the whole database (around 1 million PCM images) showed in a quantitative manner that myoblasts grown in SFM were more elongated and smaller than cells grown in SSM. The algorithms developed through this project could be used in the future on other cellular phenotypes for high-throughput screening and cell culture control applications.

Imagens multiespectrais são utilizadas em diferentes aplicações, que vão desde sensoriamento remoto a processos médicos. No caso de imagens multiespectrais oriundas de microscopia confocal de varredura à laser (Confocal Laser Scanning Microscopy-CLSM), a extração da informação se inicia pela conversão das assinaturas espectrais, em uma imagem RGB. Esta imagem é a referência para a seleção da região de interesse, da qual se obtém a assinatura espectral média, originada do arquivo multiespectral (LSM). Mesmo utilizando um padrão muito bem estabelecido de conversão, alguns pontos devem ser considerados: i) o processo de conversão reduz a informação, a uma ordem de 10-145%; ii) a cor é uma experiência sensorial, subjetiva e pessoal, interferindo na seleção da região de interesse e; iii) a assinatura é obtida pela média espectral, da região de interesse, selecionada manualmente.Assim, esta tese de doutorado propõem um método de mapeamento e visualização das informações de imagens multiespectrais, combinando um algoritmo de agrupamento não supervisionado(kmeans) e um algoritmo que define uma paleta de cores coerentes com a informação espectral das regiões mapeadas. Aplicou-se o método em três casos de estudos de tecidos vegetais: i) no pré-tratamento de paredes celulares da cana-de-açúcar; ii) na plasticidade foliar do Jacaranda caroba e; iii) no uso de assinaturas espectrais na classificação de plantas do Cerrado. Os resultados demonstraram que o método é bastante robusto, permitindo de forma inovadora a: visualização, análise e comparação de imagens multiespectrais qualitativa e quantitativamente, e que seu uso é viável em qualquer área de pesquisa que utilize imagens multiespectrais.
Multispectral images are used in different applications, ranging from remote sensing images to medical images. In the case of multispectral images derived from confocal laser scanning microscopy (CLSM), the extraction of information begins with the conversion of spectral signatures in an RGB image. This is the reference for selecting the region of interest, from which it gets the average spectral signature, originated from multispectral file (LSM). Even using a very well established pattern of conversion, some points should be considered: i) the conversion process reduces the information on the order of 10-145%; ii) the color is a sensory experience, subjective and personal, interfering in the selection of the interest region and; the signature is obtained by the spectral average, from interest region which is selected manually. Thus, this doctoral thesis proposes a method of mapping and visualization of multispectral imaging information, combining an unsupervised clustering algorithm (kmeans) and an algorithm that defines a consistent color palette with the spectral information of mapped regions. The proposed method was applied in three cases plant tissue studies: i) in the pre-treating the cell walls of sugarcane; ii) in the leaf plasticity of Jacaranda caroba; iii) in the use of spectral signatures in the Cerrado plant classification. The results showed that the proposed method is quite robust. It presents innovation to the visualization and analysis of multispectral images and makes possible a qualitative and quantitative comparison of a group of multispectral images. Besides that, its use is feasible in any area of research, which are using multispectral images.

This dissertation develops novel methods to reconstruct a three-dimensional surface together with a characterization of the surface composition given one or more images obtained from the same viewing direction. First, a vector space method to reconstruct a surface given a gradient field is developed using the linear relationship between a surface and its gradient field in discrete surface domains. The developed gradient field representation is generalized to alleviate the common assumption of uniform integrability in gradient fields to partial integrability, allowing adequate reconstruction of surfaces with non-integrable gradient fields. In addition, the developed technique is further explored for gradient field noise reduction, by embedding multiscale properties providing sparse gradient field representations. Next, the ambiguity in possible surface gradients obtained by a two-image photometric stereo analysis is resolved using a cyclic projections algorithm based on the set of possible gradient fields and the previously developed gradient field representation. An algorithm that provides accurate surface reconstructions and surface type characterizations given two images of an unknown composite surface is established. We then apply this algorithm to Scanning Electron Microscopy (SEM) images to extract specimen surface topography and material type information from a pair of Secondary Electron (SE) and Back-scattered Electron (BSE) images. We then use a similar cyclic projections algorithm to reconstruct a surface from a single image. The simulation results indicate that the developed algorithm solves this classical shape-from-shading problem in a robust and accurate manner for varying illumination conditions. Finally, we establish a unified surface reconstruction framework using previously developed techniques on a photometric stereo image triplet containing shadows. We categorize the surface pixels as those illuminated in all three images, only two images and only one image. We then establish through simulation results that the developed method uses the surface gradient information obtained from the brightness images efficiently and effectively, and provides an accurate surface reconstruction.

La microscopie acoustique quantitative (MAQ) est une modalité d'imagerie bien établie qui donne accès à des cartes paramétriques 2D représentatives des propriétés mécaniques des tissus à une échelle microscopique. Dans la plupart des études sur MAQ, l'échantillons est scanné ligne par ligne (avec un pas de 2µm) à l'aide d'un transducteur à 250 MHz. Ce type d'acquisition permet d'obtenir un cube de données RF 3D, avec deux dimensions spatiales et une dimension temporelle. Chaque signal RF correspondant à une position spatiale dans l'échantillon permet d'estimer des paramètres acoustiques comme par exemple la vitesse du son ou l'impédance. Le temps d'acquisition en MAQ est directement proportionnel à la taille de l'échantillon et peut aller de quelques minutes à quelques dizaines de minutes. Afin d'assurer des conditions d'acquisition stables et étant donnée la sensibilité des échantillons à ces conditions, diminuer le temps d'acquisition est un des grand défis en MAQ. Afin de relever ce défi, ce travail de thèse propose plusieurs solutions basées sur l'échantillonnage compressé (EC) et la théories des signaux ayant un faible nombre de degré de liberté (finite rate of innovation - FRI, en anglais). Le principe de l'EC repose sur la parcimonie des données, sur l'échantillonnage incohérent de celles-ci et sur les algorithmes d'optimisation numérique. Dans cette thèse, les phénomènes physiques derrière la MAQ sont exploités afin de créer des modèles adaptés aux contraintes de l'EC et de la FRI. Plus particulièrement, ce travail propose plusieurs pistes d'application de l'EC en MAQ : un schéma d'acquisition spatiale innovant, un algorithme de reconstruction d'images exploitant les statistiques des coefficients en ondelettes des images paramétriques, un modèle FRI adapté aux signaux RF et un schéma d'acquisition compressée dans le domaine temporel
Quantitative acoustic microscopy (QAM) is a well-accepted modality for forming 2D parameter maps making use of mechanical properties of soft tissues at microscopic scales. In leading edge QAM studies, the sample is raster-scanned (spatial step size of 2µm) using a 250 MHz transducer resulting in a 3D RF data cube, and each RF signal for each spatial location is processed to obtain acoustic parameters, e.g., speed of sound or acoustic impedance. The scanning time directly depends on the sample size and can range from few minutes to tens of minutes. In order to maintain constant experimental conditions for the sensitive thin sectioned samples, the scanning time is an important practical issue. To deal with the current challenge, we propose the novel approach inspired by compressed sensing (CS) and finite rate of innovation (FRI). The success of CS relies on the sparsity of data under consideration, incoherent measurement and optimization technique. On the other hand, the idea behind FRI is supported by a signal model fully characterized as a limited number of parameters. From this perspective, taking into account the physics leading to data acquisition of QAM system, the QAM data can be regarded as an adequate application amenable to the state of the art technologies aforementioned. However, when it comes to the mechanical structure of QAM system which does not support canonical CS measurement manners on the one hand, and the compositions of the RF signal model unsuitable to existing FRI schemes on the other hand, the advanced frameworks are still not perfect methods to resolve the problems that we are facing. In this thesis, to overcome the limitations, a novel sensing framework for CS is presented in spatial domain: a recently proposed approximate message passing (AMP) algorithm is adapted to account for the underlying data statistics of samples sparsely collected by proposed scanning patterns. In time domain, as an approach for achieving an accurate recovery from a small set of samples of QAM RF signals, we employ sum of sincs (SoS) sampling kernel and autoregressive (AR) model estimator. The spiral scanning manner, introduced as an applicable sensing technique to QAM system, contributed to the significant reduction of the number of spatial samples when reconstructing speed of sound images of a human lymph node.[...]

Modern fluorescence microscopy modalities, such as light-sheet microscopy, are capable of acquiring large three-dimensional images at high data rate. This creates a bottleneck in computational processing and analysis of the acquired images, as the rate of acquisition outpaces the speed of processing. Moreover, images can be so large that they do not fit the main memory of a single computer. Another issue is the information loss during image acquisition due to limitations of the optical imaging systems. Analysis of the acquired images may, therefore, find multiple solutions (or no solution) due to imaging noise, blurring, and other uncertainties introduced during image acquisition. In this thesis, we address the computational processing time and memory issues by developing a distributed parallel algorithm for segmentation of large fluorescence-microscopy images. The method is based on the versatile Discrete Region Competition (Cardinale et al., 2012) algorithm, which has previously proven useful in microscopy image segmentation. The present distributed implementation decomposes the input image into smaller sub-images that are distributed across multiple computers. Using network communication, the computers orchestrate the collective solving of the global segmentation problem. This not only enables segmentation of large images (we test images of up to 10^10 pixels) but also accelerates segmentation to match the time scale of image acquisition. Such acquisition-rate image segmentation is a prerequisite for the smart microscopes of the future and enables online data inspection and interactive experiments. Second, we estimate the segmentation uncertainty on large images that do not fit the main memory of a single computer. We there- fore develop a distributed parallel algorithm for efficient Markov- chain Monte Carlo Discrete Region Sampling (Cardinale, 2013). The parallel algorithm provides a measure of segmentation uncertainty in a statistically unbiased way. It approximates the posterior probability densities over the high-dimensional space of segmentations around the previously found segmentation
Moderne Fluoreszenzmikroskopie, wie zum Beispiel Lichtblattmikroskopie, erlauben die Aufnahme hochaufgelöster, 3-dimensionaler Bilder. Dies führt zu einen Engpass bei der Bearbeitung und Analyse der aufgenommenen Bilder, da die Aufnahmerate die Datenverarbeitungsrate übersteigt. Zusätzlich können diese Bilder so groß sein, dass sie die Speicherkapazität eines einzelnen Computers überschreiten. Hinzu kommt der aus Limitierungen des optischen Abbildungssystems resultierende Informationsverlust während der Bildaufnahme. Bildrauschen, Unschärfe und andere Messunsicherheiten können dazu führen, dass Analysealgorithmen möglicherweise mehrere oder keine Lösung für Bildverarbeitungsaufgaben finden. Im Rahmen der vorliegenden Arbeit entwickeln wir einen verteilten, parallelen Algorithmus für die Segmentierung von speicherintensiven Fluoreszenzmikroskopie-Bildern. Diese Methode basiert auf dem vielseitigen "Discrete Region Competition" Algorithmus (Cardinale et al., 2012), der sich bereits in anderen Anwendungen als nützlich für die Segmentierung von Mikroskopie-Bildern erwiesen hat. Das hier präsentierte Verfahren unterteilt das Eingangsbild in kleinere Unterbilder, welche auf die Speicher mehrerer Computer verteilt werden. Die Koordinierung des globalen Segmentierungsproblems wird durch die Benutzung von Netzwerkkommunikation erreicht. Dies erlaubt die Segmentierung von sehr großen Bildern, wobei wir die Anwendung des Algorithmus auf Bildern mit bis zu 10^10 Pixeln demonstrieren. Zusätzlich wird die Segmentierungsgeschwindigkeit erhöht und damit vergleichbar mit der Aufnahmerate des Mikroskops. Dies ist eine Grundvoraussetzung für die intelligenten Mikroskope der Zukunft, und es erlaubt die Online-Betrachtung der aufgenommenen Daten, sowie interaktive Experimente. Wir bestimmen die Unsicherheit des Segmentierungsalgorithmus bei der Anwendung auf Bilder, deren Größe den Speicher eines einzelnen Computers übersteigen. Dazu entwickeln wir einen verteilten, parallelen Algorithmus für effizientes Markov-chain Monte Carlo "Discrete Region Sampling" (Cardinale, 2013). Dieser Algorithmus quantifiziert die Segmentierungsunsicherheit statistisch erwartungstreu. Dazu wird die A-posteriori-Wahrscheinlichkeitsdichte über den hochdimensionalen Raum der Segmentierungen in der Umgebung der zuvor gefundenen Segmentierung approximiert

With the emergence of new imaging techniques, researchers are always eager to push the boundaries by examining objects either smaller or further away than what was previously possible. The development of image analysis techniques has greatly helped to introduce objectivity and coherence in measurements and decision making. It has become an essential tool for facilitating both large-scale quantitative studies and qualitative research. In this Thesis, methods were developed for analysis of low-resolution (in respect to the size of the imaged objects) three-dimensional (3D) images with low signal-to-noise ratios (SNR) applied to images from cryo-electron tomography (cryo-ET) and fluorescence microscopy (FM). The main focus is on methods of low complexity, that take into account both grey-level and shape information, to facilitate large-scale studies. Methods were developed to localise and represent complex macromolecules in images from cryo-ET. The methods were applied to Immunoglobulin G (IgG) antibodies and MET proteins. The low resolution and low SNR required that grey-level information was utilised to create fuzzy representations of the macromolecules. To extract structural properties, a method was developed to use grey-level-based distance measures to facilitate decomposition of the fuzzy representations into sub-domains. The structural properties of the MET protein were analysed by developing a analytical curve representation of its stalk. To facilitate large-scale analysis of structural properties of nerve cells, a method for tracing neurites in FM images using local path-finding was developed. Both theoretical and implementational details of computationally heavy approaches were examined to keep the time complexity low in the developed methods. Grey-weighted distance definitions and various aspects of their implementations were examined in detail to form guidelines on which definition to use in which setting and which implementation is the fastest. Heuristics were developed to speed up computations when calculating grey-weighted distances between two points. The methods were evaluated on both real and synthetic data and the results show that the methods provide a step towards facilitating large-scale studies of images from both cryo-ET and FM.

Includes abstract.
Includes bibliographical references (leaves 140-144).
Automated microscopy for the detection of tuberculosis (TB) in sputum smears seeks to address the strain on technicians in busy TB laboratories and to achieve faster diagnosis in countries with a heavy TB burden. As a step in the development of an automated microscope, the project described here was concerned with microscope auto-positioning; this primarily involves generating a point of reference on a slide, which can be used to automatically bring desired fields on the slide to the field-of-view of the microscope for re-examination. The study was carried out using a conventional microscope and Ziehl- Neelsen (ZN) stained sputum smear slides. All images were captured at 40x magnification. A digital replication, the virtual slide map, of an actual slide was constructed by combining the manually acquired images of the different fields of the slide. The geometric hashing scheme was found to be suitable for auto-stitching a large number of images (over 300 images) to form a virtual slide map. An object recognition algorithm, which was also based on the geometric hashing technique, was used to localise a query image (the current field-of-view) on the virtual slide map. This localised field-of-view then served as the point of reference. The true positive (correct localisation of a query image on the virtual slide map) rate achieved by the algorithm was above 88% even for noisy query images captured at slide orientations up to 26°. The image registration error, computed as the average mean square error, was less than 14 pixel2 (corresponding to 1.02 μm2 and 0.001% error in an image measuring 1030 x 1300 pixels) corresponding to a root mean square registration error of 3.7 pixels. Superior image registration accuracy was obtained at the expense of time using the scale invariant feature transform (SIFT), with a image registration error of 1 pixel2 (0.07 μm2). The object recognition algorithm is inherently robust to changes in slide orientation and placement, which are likely to occur in practice as it is impossible to place the slide in exactly the same position on the microscope at different times. Moreover, the algorithm showed high tolerance to illumination changes and robustness to noise.

El diagnóstico final de la mayoría de tipos de cáncer lo realiza un médico experto en anatomía patológica que examina muestras tisulares o celulares sospechosas extraídas del paciente. Actualmente, esta evaluación depende en gran medida de la experiencia del médico y se lleva a cabo de forma cualitativa mediante técnicas de imagen tradicionales como la microscopía óptica. Esta tarea tediosa está sujeta a altos grados de subjetividad y da lugar a niveles de discordancia inadecuados entre diferentes patólogos, especialmente en las primeras etapas de desarrollo del cáncer. La espectroscopía infrarroja por Transformada de Fourier (siglas FTIR en inglés) es una tecnología ampliamente utilizada en la industria que recientemente ha demostrado una capacidad creciente para mejorar el diagnóstico de diferentes tipos de cáncer. Esta técnica aprovecha las propiedades del infrarrojo medio para excitar los modos vibratorios de los enlaces químicos que forman las muestras biológicas. La principal señal generada consiste en un espectro de absorción que informa sobre la composición química de la muestra iluminada. Los microespectrómetros FTIR modernos, compuestos por complejos componentes ópticos y detectores matriciales de alta sensibilidad, permiten capturar en un laboratorio de investigación común imágenes hiperespectrales de alta calidad que aúnan información química y espacial. Las imágenes FTIR son estructuras de datos ricas en información que se pueden analizar individualmente o junto con otras modalidades de imagen para realizar diagnósticos patológicos objetivos. Por lo tanto, esta técnica de imagen emergente alberga un alto potencial para mejorar la detección y la graduación del riesgo del paciente en el cribado y vigilancia de cáncer. Esta tesis estudia e implementa diferentes metodologías y algoritmos de los campos interrelacionados de procesamiento de imagen, visión por ordenador, aprendizaje automático, reconocimiento de patrones, análisis multivariante y quimiometría para el procesamiento y análisis de imágenes hiperespectrales FTIR. Estas imágenes se capturaron con un moderno microscopio FTIR de laboratorio a partir de muestras de tejidos y células afectadas por cáncer colorrectal y de piel, las cuales se prepararon siguiendo protocolos alineados con la práctica clínica actual. Los conceptos más relevantes de la espectroscopía FTIR se investigan profundamente, ya que deben ser comprendidos y tenidos en cuenta para llevar a cabo una correcta interpretación y tratamiento de sus señales especiales. En particular, se revisan y analizan diferentes factores fisicoquímicos que influyen en las mediciones espectroscópicas en el caso particular de muestras biológicas y pueden afectar críticamente su análisis posterior. Todos estos conceptos y estudios preliminares entran en juego en dos aplicaciones principales. La primera aplicación aborda el problema del registro o alineación de imágenes hiperespectrales FTIR con imágenes en color adquiridas con microscopios tradicionales. El objetivo es fusionar la información espacial de distintas muestras de tejido medidas con esas dos modalidades de imagen y centrar la discriminación en las regiones seleccionadas por los patólogos, las cuales se consideran más relevantes para el diagnóstico de cáncer colorrectal. En la segunda aplicación, la espectroscopía FTIR se lleva a sus límites de detección para el estudio de las entidades biomédicas más pequeñas. El objetivo es evaluar las capacidades de las señales FTIR para discriminar de manera fiable diferentes tipos de células de piel que contienen fenotipos malignos. Los estudios desarrollados contribuyen a la mejora de métodos de decisión objetivos que ayuden al patólogo en el diagnóstico final del cáncer. Además, revelan las limitaciones de los protocolos actuales y los problemas intrínsecos de la tecnología FTIR moderna, que deberían abordarse para permit
The final diagnosis of most types of cancers is performed by an expert clinician in anatomical pathology who examines suspicious tissue or cell samples extracted from the patient. Currently, this assessment largely relies on the experience of the clinician and is accomplished in a qualitative manner by means of traditional imaging techniques, such as optical microscopy. This tedious task is subject to high degrees of subjectivity and gives rise to suboptimal levels of discordance between different pathologists, especially in early stages of cancer development. Fourier Transform infrared (FTIR) spectroscopy is a technology widely used in industry that has recently shown an increasing capability to improve the diagnosis of different types of cancer. This technique takes advantage of the ability of mid-infrared light to excite the vibrational modes of the chemical bonds that form the biological samples. The main generated signal consists of an absorption spectrum that informs of the chemical composition of the illuminated specimen. Modern FTIR microspectrometers, composed of complex optical components and high-sensitive array detectors, allow the acquisition of high-quality hyperspectral images with spatially-resolved chemical information in a common research laboratory. FTIR images are information-rich data structures that can be analysed alone or together with other imaging modalities to provide objective pathological diagnoses. Hence, this emerging imaging technique presents a high potential to improve the detection and risk stratification in cancer screening and surveillance. This thesis studies and implements different methodologies and algorithms from the related fields of image processing, computer vision, machine learning, pattern recognition, multivariate analysis and chemometrics for the processing and analysis of FTIR hyperspectral images. Those images were acquired with a modern benchtop FTIR microspectrometer from tissue and cell samples affected by colorectal and skin cancer, which were prepared by following protocols close to the current clinical practise. The most relevant concepts of FTIR spectroscopy are thoroughly investigated, which ought to be understood and considered to perform a correct interpretation and treatment of its special signals. In particular, different physicochemical factors are reviewed and analysed, which influence the spectroscopic measurements for the particular case of biological samples and can critically affect their later analysis. All these knowledge and preliminary studies come into play in two main applications. The first application tackles the problem of registration or alignment of FTIR hyperspectral images with colour images acquired with traditional microscopes. The aim is to fuse the spatial information of distinct tissue samples measured by those two imaging modalities and focus the discrimination on regions selected by the pathologists, which are meant to be the most relevant areas for the diagnosis of colorectal cancer. In the second application, FTIR spectroscopy is pushed to their limits of detection for the study of the smallest biomedical entities. The aim is to assess the capabilities of FTIR signals to reliably discriminate different types of skin cells containing malignant phenotypes. The developed studies contribute to the improvement of objective decision methods to support the pathologist in the final diagnosis of cancer. In addition, they reveal the limitations of current protocols and intrinsic problems of modern FTIR technology, which should be tackled in order to enable its transference to anatomical pathology laboratories in the future.
El diagnòstic final de la majoria de tipus de càncer ho realitza un metge expert en anatomia patològica que examina mostres tissulars o cel¿lulars sospitoses extretes del pacient. Actualment, aquesta avaluació depèn en gran part de l'experiència del metge i es porta a terme de forma qualitativa mitjançant tècniques d'imatge tradicionals com la microscòpia òptica. Aquesta tasca tediosa està subjecta a alts graus de subjectivitat i dóna lloc a nivells de discordança inadequats entre diferents patòlegs, especialment en les primeres etapes de desenvolupament del càncer. L'espectroscòpia infraroja per Transformada de Fourier (sigles FTIR en anglès) és una tecnologia àmpliament utilitzada en la indústria que recentment ha demostrat una capacitat creixent per millorar el diagnòstic de diferents tipus de càncer. Aquesta tècnica aprofita les propietats de l'infraroig mitjà per excitar els modes vibratoris dels enllaços químics que formen les mostres biològiques. El principal senyal generat consisteix en un espectre d'absorció que informa sobre la composició química de la mostra il¿luminada. Els microespectrómetres FTIR moderns, compostos per complexos components òptics i detectors matricials d'alta sensibilitat, permeten capturar en un laboratori d'investigació comú imatges hiperespectrals d'alta qualitat que uneixen informació química i espacial. Les imatges FTIR són estructures de dades riques en informació que es poden analitzar individualment o juntament amb altres modalitats d'imatge per a realitzar diagnòstics patològics objectius. Per tant, aquesta tècnica d'imatge emergent té un alt potencial per a millorar la detecció i la graduació del risc del pacient en el cribratge i vigilància de càncer. Aquesta tesi estudia i implementa diferents metodologies i algoritmes dels camps interrelacionats de processament d'imatge, visió per ordinador, aprenentatge automàtic, reconeixement de patrons, anàlisi multivariant i quimiometria per al processament i anàlisi d'imatges hiperespectrals FTIR. Aquestes imatges es van capturar amb un modern microscopi FTIR de laboratori a partir de mostres de teixits i cèl¿lules afectades per càncer colorectal i de pell, les quals es van preparar seguint protocols alineats amb la pràctica clínica actual. Els conceptes més rellevants de l'espectroscòpia FTIR s'investiguen profundament, ja que han de ser compresos i tinguts en compte per dur a terme una correcta interpretació i tractament dels seus senyals especials. En particular, es revisen i analitzen diferents factors fisicoquímics que influeixen en els mesuraments espectroscòpiques en el cas particular de mostres biològiques i poden afectar críticament la seua anàlisi posterior. Tots aquests conceptes i estudis preliminars entren en joc en dues aplicacions principals. La primera aplicació aborda el problema del registre o alineació d'imatges hiperespectrals FTIR amb imatges en color adquirides amb microscopis tradicionals. L'objectiu és fusionar la informació espacial de diferents mostres de teixit mesurades amb aquestes dues modalitats d'imatge i centrar la discriminació en les regions seleccionades pels patòlegs, les quals es consideren més rellevants per al diagnòstic de càncer colorectal. En la segona aplicació, l'espectroscòpia FTIR es porta als seus límits de detecció per a l'estudi de les entitats biomèdiques més xicotetes. L'objectiu és avaluar les capacitats dels senyals FTIR per discriminar de manera fiable diferents tipus de cèl¿lules de pell que contenen fenotips malignes. Els estudis desenvolupats contribueixen a la millora de mètodes de decisió objectius que ajuden el patòleg en el diagnòstic final del càncer. A més, revelen les limitacions dels protocols actuals i els problemes intrínsecs de la tecnologia FTIR moderna, que haurien d'abordar per permetre la seva transferència als laboratoris d'anatomia patològica en el futur.
Peñaranda Gómez, FJ. (2018). Application of artificial vision algorithms to images of microscopy and spectroscopy for the improvement of cancer diagnosis [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/99748
TESIS

La cryo-tomographie électronique (cryo-ET) est une technique d'imagerie capable de produire des vues 3D de spécimens biologiques. Cette technologie permet d’imager de larges portions de cellules vitrifiées à une résolution nanométrique. Elle permet de combiner plusieurs échelles de compréhension de la machinerie cellulaire, allant des interactions entre les groupes de protéines à leur structure atomique. La cryo-ET a donc le potentiel d'agir comme un lien entre l'imagerie cellulaire in vivo et les techniques atteignant la résolution atomique. Cependant, ces images sont corrompues par un niveau de bruit élevé et d'artefacts d'imagerie. Leur interprétabilité dépend fortement des méthodes de traitement d'image. Les méthodes computationelles existantes permettent actuellement d'identifier de larges macromolécules telles que les ribosomes, mais il est avéré que ces détections sont incomplètes. De plus, ces méthodes sont limitées lorsque les objets recherchés sont de très petite taille ou présentent une plus grande variabilité structurelle. L'objectif de cette thèse est de proposer de nouvelles méthodes d'analyse d'images, afin de permettre une identification plus robuste des macromolécules d'intérêt. Nous proposons deux méthodes computationelles pour atteindre cet objectif. La première vise à réduire le bruit et les artefacts d'imagerie, et fonctionne en ajoutant et en supprimant de façon itérative un bruit artificiel à l'image. Nous fournissons des preuves mathématiques et expérimentales de ce concept qui permet d'améliorer le signal dans les images de cryo-ET. La deuxième méthode s'appuie sur les progrès récents de l'apprentissage automatique et les méthodes convolutionelles pour améliorer la localisation des macromolécules. La méthode est basée sur un réseau neuronal convolutif et nous montrons comment l'adapter pour obtenir des taux de détection supérieur à l'état de l'art
Cryo electron tomography (cryo-ET) is an imaging technique capable of producing 3D views of biological specimens. This technology enables to capture large field of views of vitrified cells at nanometer resolution. These features allow to combine several scales of understanding of the cellular machinery, from the interactions between groups of proteins to their atomic structure. Cryo-ET therefore has the potential to act as a link between in vivo cell imaging and atomic resolution techniques. However, cryo-ET images suffer from a high amount of noise and imaging artifacts, and the interpretability of these images heavily depends on computational image analysis methods. Existing methods allow to identify large macromolecules such as ribosomes, but there is evidence that the detections are incomplete. In addition, these methods are limited when searched objects are smaller and have more structural variability. The purpose of this thesis is to propose new image analysis methods, in order to enable a more robust identification of macromolecules of interest. We propose two computational methods to achieve this goal. The first aims at reducing the noise and imaging artifacts, and operates by iteratively adding and removing artificial noise to the image. We provide both mathematical and experimental evidence that this concept allows to enhance signal in cryo-ET images. The second method builds on recent advances in machine learning to improve macromolecule localization. The method is based on a convolutional neural network, and we show how it can be adapted to achieve better detection rates than the current state-of- the-art

Corneal diseases are one of the major causes of visual impairment and blindness worldwide. Used for diagnoses, a laser confocal microscope provides a sequence of images, at incremental depths, of the various corneal layers and structures. From these, ophthalmologists can extract clinical information on the state of health of a patient’s cornea. However, many factors impede ophthalmologists in forming diagnoses starting with the large number and variable quality of the individual images (blurring, non-uniform illumination within images, variable illumination between images and noise), and there are also difficulties posed for automatic processing caused by eye movements in both lateral and axial directions during the scanning process. Aiding ophthalmologists working with long sequences of corneal image requires the development of new algorithms which enhance, correctly order and register the corneal images within a sequence. The novel algorithms devised for this purpose and presented in this thesis are divided into four main categories. The first is enhancement to reduce the problems within individual images. The second is automatic image classification to identify which part of the cornea each image belongs to, when they may not be in the correct sequence. The third is automatic reordering of the images to place the images in the right sequence. The fourth is automatic registration of the images with each other. A flexible application called CORNEASYS has been developed and implemented using MATLAB and the C language to provide and run all the algorithms and methods presented in this thesis. CORNEASYS offers users a collection of all the proposed approaches and algorithms in this thesis in one platform package. CORNEASYS also provides a facility to help the research team and Ophthalmologists, who are in discussions to determine future system requirements which meet clinicians’ needs.
The data and image files accompanying this thesis are not available online.

Corneal diseases are one of the major causes of visual impairment and blindness worldwide. Used for diagnoses, a laser confocal microscope provides a sequence of images, at incremental depths, of the various corneal layers and structures. From these, ophthalmologists can extract clinical information on the state of health of a patient’s cornea. However, many factors impede ophthalmologists in forming diagnoses starting with the large number and variable quality of the individual images (blurring, non-uniform illumination within images, variable illumination between images and noise), and there are also difficulties posed for automatic processing caused by eye movements in both lateral and axial directions during the scanning process. Aiding ophthalmologists working with long sequences of corneal image requires the development of new algorithms which enhance, correctly order and register the corneal images within a sequence. The novel algorithms devised for this purpose and presented in this thesis are divided into four main categories. The first is enhancement to reduce the problems within individual images. The second is automatic image classification to identify which part of the cornea each image belongs to, when they may not be in the correct sequence. The third is automatic reordering of the images to place the images in the right sequence. The fourth is automatic registration of the images with each other. A flexible application called CORNEASYS has been developed and implemented using MATLAB and the C language to provide and run all the algorithms and methods presented in this thesis. CORNEASYS offers users a collection of all the proposed approaches and algorithms in this thesis in one platform package. CORNEASYS also provides a facility to help the research team and Ophthalmologists, who are in discussions to determine future system requirements which meet clinicians’ needs.

Confocal microscopy demonstrated its value in the diagnosis of Acanthamoeba and fungal keratitis which considered sight-threatening corneal diseases. However, it can be difficult to find and train confocal microscopy graders to accurately detect Acanthamoeba cysts and fungal filaments in the images. Use of an automated system could overcome this problem and help to start the correct treatment more quickly. Also, response to treatment can be difficult to assess in infectious keratitis using clinical examination alone, but there is evidence that the morphology of filaments and cysts may change over time with the use of correct treatment. An automated system to analyse confocal microscopy images for such changes would also assist clinicians in determining whether the ulcer is improving, or whether a change of treatment is needed. This research proposes a fully automated novel system with GUI to detect cysts and hyphae (filaments) and measure useful quantitative parameters for them through many stages; Image enhancement, image segmentation, quantitative analysis for detected cysts and hyphae, and registration and tracking of ordered sequence of images. The performance of the proposed segmentation procedure is evaluated by comparing between the manual and the automated traced images of the dataset that was provided by the Manchester Royal Eye Hospital. The positive predictive values rate of cysts for Acanthamoeba images was 76%. For detected hyphae in Fusarium images, many standard measurements were computed. The accuracy of their values was quantified by calculating the percent error rate for each measurement and which ranged from 23% to 49%.

Ovarian cancer is a deadly disease owing to the non-specific symptoms and suspected rapid progression, leading to frequent late stage detection and poor prognosis. Medical imaging methods such as CT, MRI and ultrasound as well as serum testing for cancer markers have had extremely poor performance for early disease detection. Due to the poor performance of available screening methods, and the impracticality and ineffectiveness of taking tissue biopsies from the ovary, women at high risk for developing ovarian cancer are often advised to undergo prophylactic salpingo-oophorectomy. This surgery results in many side effects and is most often unnecessary since only a fraction of high risk women go on to develop ovarian cancer. Better understanding of the early development of ovarian cancer and characterization of morphological changes associated with early disease could lead to the development of an effective screening test for women at high risk. Optical imaging methods including optical coherence tomography (OCT) and multiphoton microscopy (MPM) are excellent tools for studying disease progression owing to the high resolution and depth sectioning capabilities. Further, these techniques are excellent for optical biopsy because they can image in situ non-destructively. In the studies described in this dissertation OCT and MPM are used to identify cellular and tissue morphological changes associated with early tumor development in a mouse model of ovarian cancer. This work is organized into three specific aims. The first aim is to use the images from the MPM phenomenon of second harmonic generation to quantitatively examine the morphological differences in collagen structure in normal mouse ovarian tissue and mouse ovarian tumors. The second aim is to examine the differences in endogenous two-photon excited fluorescence in normal mouse ovarian tissue and mouse ovarian tumors. The third and final aim is to identify changes in ovarian microstructure resulting from early disease development by imaging animals in vivo at three time points during a long-term survival study.

In this thesis we present the development of machine learning algorithms for single cell analysis in an in-vitro diagnostic platform for Cellply, a startup that operates in precision medicine. We researched the state of the art of deep learning for biomedical image analysis, and we analyzed the impact that convolutional neural networks have had in object detection tasks. Then we compared neural networks that are currently used for cell detection, and we chose the one (i.e. Stardist) that is able to perform a more efficient detection also in a crowded cells context. We could train models using Stardist algorithm in the open-source platform ZeroCostDL4Mic, using code and GPU in Colab environment. We trained different models, intended for distinct applications, and we evaluated them using metrics such as precision and recall. These are our results: • a model for single channel brightfield images taken from samples of Covid patients, that guarantees a precision of about 0.98 and a recall of about 0.96 • a model for multi-channel images (i.e. a stack of multiple images, each one highlighting different contents) taken from experiments about natural killer cells, with precision and recall of about 0.81 • a model for multi-channel images taken from samples of AML (Acute Myeloid Leukemia) patients, with precision and recall of about 0.73 • a simpler model, trained to detect the main area (named "well") on which cells can be found, in order to discard what is out of this area. This model has a precision of about 1 and a recall of about 0.98. Finally, we wrote Python code in order to read a text input file that contains the necessary information to run a specified trained model for cell detection, with certain parameters, on a given set of images of a certain experiment. The output of the code is a .csv file where measurements related to every detected “object of interest” (i.e. cells or other particles) are stored. We also talk about future developments in this field.