Дисертації з теми "Biomedical textile"

The objective of this research is to develop a series of material treatments and modifications, and, using a standardized set of tests, determine the extent of the ability of the modified material to enhance coagulation. This research focuses on materials commonly used in traditional textile based wound dressings; utilizing Streaming Potential studies, Scanning Electron Microscopy (SEM) and Thrombin Assays. The materials tested can be classified into 4 groups: control materials, modified PLA, SAMs treated glass, and TEOS treated materials. The control materials included: spun cotton and rayon yarn; continuous filament Nylon, Polypropylene (PP), and Polyethylene terephthalate (PET); heat cleaned glass (control glass); and PLA staple fibers. Contact angle measurements showed that both the control glass and the PET showed an increase in contact angle when treated with TEOS. This corresponds to a decrease and no improvement, respectively, in thrombogenicity for these materials in the thrombin assay. The remaining materials tested showed no change or a decrease in contact angle after TEOS treatment, and a corresponding increase in thrombogenicity. These results support previous studies that indicate an increase in wettability contributes to the enhancement of coagulation (16). While the streaming potential studies showed no correlation between thrombin formation or contact angle data, these tests provided an important launching platform for future studies utilizing the Streaming Potential Jar. Future work could benefit from the use of more physiologically relevant solutions, such as CaCl2, NaCl, or other blood substitutes (15). While no definitive correlations between test methods were elucidated, the results garnered from this research created a strong launching platform from which future materials research can continue.

Medical applications of technical textiles are an expanding field of research. One of the added values of these new materials would be that they were suitable to contain and release active compounds in a controlled and sustained manner. Drug incorporation and release from synthetic fibers is related to the interaction of the drug with the polymer and probably greatly depends on the surface chemistry of the fiber. Plasma technology is a tool that enables to modify physical and chemical properties of the first nanometers of the fibers without affecting the bulk of the material. Applied to the medical textile field, plasma treatment of polymer fibres can lead to the design of new textile-based drug delivery systems. The novelty of this PhD. Thesis rests upon the modification of the drug/fiber interactions by plasma treatment to allow the modulation of the loading and the release of active principles (pharmaceutics and cosmetics) from the textile-based drug delivery systems, without requiring the use of any further chemicals. This Thesis aims at the development of two families of textile-based drug delivery systems, based on a novel surface functionalization by plasma treatment, with suitable characteristics for topical use as medical devices, or for clinical application in soft tissue repair. It is therefore organized in two distinct parts. In both parts of this thesis a general scheme has been followed: we have investigated the surface modification of textile materials with different types of plasmas (atmospheric and low pressure plasma), characterizing the surface modifications achieved by different complementary techniques. The effects of the plasma treatment have been evaluated on the subsequent incorporation of active pharmaceuticals and cosmetics. In the last step, the drug release to a standard medium has been studied by "in-vitro" dissolution assays. The first part is focused on medical textiles for topical application. Therein, the surface modification of polyamide 66 elastic-compressive knitted fabrics has been studied by corona plasma and low pressure plasma. The work has studied in parallel laboratory prepared fabrics and industrially finished fabrics, with views on the potential implementation of the proposed process in the textile industrial chain. Plasma treatment improved the release kinetics of anti-inflammatory pharmaceutic (ketoprofen) and of lipolitic cosmetic (caffeine) active principles, loaded in the polyamide 66 fabrics. A fundamental study comparing three different molecules of the same chemical family (caffeine, theobromine, pentoxifylline) has been performed regarding loading and release of the drugs. The second part focuses on textiles used as implants for soft tissue repair (e.g. hernia). The fiber surface of a polypropylene mesh has been modified by corona plasma and low-pressure plasma. The treatments evaluated had a major effect on the loading of antibiotic (ampicillin) by increasing it three times. As in vitro release kinetics of the drug was very fast, coating of the ampillicin-loaded polypropylene meshes with a biocompatible polymer was investigated by plasma polymerization.
Las aplicaciones médicas de los textiles técnicos son un campo de investigación en expansión. Uno de los valores añadidos de estos nuevos materiales puede ser su capacidad para contener y liberar principios activos farmacéuticos y cosméticos de una forma controlada y sostenida. La incorporación de fármacos y su liberación a partir de fibras sintéticas está relacionada con la interacción del fármaco con el polímero y puede depender en gran medida de la química de superficie de la fibra. La tecnología de plasma es una herramienta que permite modificar las propiedades físicas y químicas de los primero nanómetros de la superficie de las fibras sin afectar el interior del material. Aplicado al campo de los textiles médicos, el tratamiento con plasma de fibras poliméricas podría conducir al diseño de nuevos sistemas de liberación de fármacos basados en soportes textiles. La novedad de esta Tesis Doctoral se basa en la modificación de las interacciones fármaco / fibra por tratamiento de plasma para permitir la modulación de la incorporación y la liberación de los principios activos (farmacéuticos y cosméticos) a partir de sistemas de administración de fármacos basados en material textil, sin requerir el uso de productos químicos adicionales. Esta Tesis tiene como objetivo el desarrollo de dos familias de sistemas de liberación de fármacos basados en soportes textiles, por funcionalización de la superficie mediante tratamiento de plasma, con características adecuadas bien para uso tópico como dispositivos médicos, bien para aplicación clínica en la reparación de tejidos blandos. Por tanto, esta Tesis se organiza en dos partes bien diferenciadas. En ambas partes de esta Tesis se ha seguido el siguiente esquema general: en primer lugar se ha investigado primero la modificación superficial de los materiales textiles con diferentes tipos de plasmas (plasma corona y plasma de presión atmosférica), caracterizando las modificaciones de la superficie obtenidas mediante diferentes técnicas instrumentales. Los efectos del tratamiento con plasma se han evaluado entonces sobre la incorporación de principios activos farmacéuticos o cosméticos. En el último paso, se ha estudiado la liberación del fármaco mediante ensayos de disolución "in vitro". La primera parte de la Tesis Doctoral se centra en los textiles médicos para aplicación tópica. Para ello, se ha estudiado la modificación de la superficie de tejidos de punto elástico-compresivos de poliamida 66 con plasma corona y plasma de baja presión. En este trabajo experimental se han estudiado en paralelo tejidos preparados en laboratorio y tejidos industrialmente acabados, con vistas a la posible implementación del proceso propuesto en la cadena de producción industrial textil. Se ha observado que el tratamiento con plasma mejora la cinética de liberación de un fármaco anti-inflamatorio (ketoprofeno) y de un principio activo cosmético lipolítico (cafeína), incorporados en los tejidos de poliamida 66 tratados con plasma. Se ha desarrollado un estudio fundamental comparando tres moléculas diferentes de la misma familia química (cafeína, teobromina y pentoxifilina) con respecto a la incorporación al material textil y a la liberación del principio activo. La segunda parte se centra en los textiles utilizados como implantes para la reparación de tejidos blandos (por ejemplo, hernias abdominales). La superficie de la fibra de una malla de polipropileno approvada para su uso clínico ha sido modificada por el plasma corona y plasma de baja presión. Los tratamientos estudiados tuvieron un efecto importante sobre la carga de un antibiótico (ampicilina) mostrando un importante incremento del porcentaje de impregnación. La cinética de liberación in vitro del antibiótico de la malla de polipropileno a un medio líquido isotonico fue rápida. También se investigó la posibilidad de realizar un recubrimiento de la malla de polipropileno cargada con ampicilina mediante polimerización por plasma.

Les implants cardio-vasculaires sont de plus en plus utilisés pour la réparation des pathologies vasculaires. Près de 300 000 remplacement de valve cardiaque sont réalisés par an à travers le monde. Le développement de ces implants est désormais primordial. L’objectif de ce travail de recherche est de développer des matériaux bio-textiles performants pouvant être utilisés comme implants médicaux par l’amélioration de leur bio intégration dans le milieu biologique. En effet, suite aux études in vivo menées au LPMT, des fibroblastes prolifèrent sur la surface des implants suite à une réaction inflammatoire. Ces dernières, lorsqu’elles se fixent en grandes quantités sont à l’origine du dysfonctionnement de la valve cardiaque en textile. L’état de l’art met en évidence la sensibilité de ces cellules à la topographie. De ce fait, le traitement consiste à modifier la topographie du tissu par la projection de micro particules en surface. Cette technique a été développée par le CRITT TJFU. Le travail de recherche porte sur : i) l’étude élémentaire de l’interaction du jet d’azote supercritique avec la surface d’un polymère, ii) l’étude de l’évolution des caractéristiques physiques : vitesse de particules, température du jet en fonction des conditions du tir et iii) l’étude de l’interaction du jet avec un textile. Ainsi, sous des conditions particulières de traitement, les textiles ont subi une modification de surface à l’échelle des fils. Cette modification est caractérisée par la présence des cratères d’impact et des effilochages. Cette topographie s’avère très intéressante pour limiter les fibroblastes dans le cas de tissu monofilament et pour limiter la réaction inflammatoire sur le tissu multifilament
Cardiovascular implants are increasingly used for the repair of vascular pathologies. Almost 300,000 heart valve replacements per year are performed around the world. Nowadays, the development of these implants become crucial. The objective of this research work is to develop high-performance bio-textile materials that can be used as medical implants by improving their bio-integration into the biological environment. In fact, following in vivo studies carried out at LPMT, fibroblasts proliferate on the surface of implants following an inflammatory reaction. When these cells proliferate in large quantities, they form a biological tissue that cause the dysfunction of the textile heart valve. Bibliographic studies demonstrate the sensitivity of these cells to topography. Therefore, the treatment consists in modifying the topography of the tissue by the projection of micro particles on the surface. This technique was developed by CRITT TJFU. This research work focuses on: i) the elementary study of the supercritical nitrogen jet interaction with the polymer surface, ii) the study of the physical characteristics evolution: particle speed, temperature of the jet as well as iii) the study of the jet interaction with the textile. Thus, under special processing conditions, the particles projected by the jet N2 SC generate craters on the surface of monofilament as well as multifilament fabric, allowing topographical modifications at the yarn scale. Our results showed a significant decrease in fibroblast proliferation with increasing textile roughness compared to untreated one. Moreover, the topography limits the inflammatory reaction on the multifilament fabrics

Soft-tissue sarcoma (STS) of the extremities forms a relatively uncommon yet aggressive group of neoplasms with high metastatic risk of the disease. The vast majority of STS metastases occur in the lungs. Due to the general poor prognosis of patients diagnosed with STS lung metastases, there is a clinical need to identify relevant prognostic factors as early as possible in the course of staging and treatment management. Recent evidence suggests that positron emission tomography (PET) using fluorodeoxyglucose (FDG) and magnetic resonance (MR) imaging texture features have the potential to predict the outcome of tumours through the assessment of their microenvironment heterogeneity characteristics. The goal of this work is therefore to investigate FDG-PET and MR texture features as potential early predictors of lung metastasis risk in STS cancer of the extremities.In this study, a dataset of 35 patients with histologically proven STS of the extremities was retrospectively analyzed. All patients received pre-treatment FDG-PET and MR scans. MR imaging data comprised of T1-weighted, T2 fat-saturation (T2FS) and short tau inversion recovery (STIR) sequences. The median follow-up period was 29 months (range: 4 to 85 months). Thirteen patients from the dataset developed lung metastases. Six texture features from the gray-level co-occurrence matrix (GLCM) were extracted from the FDG-PET, MR and fused FDG-PET/MR scans. In addition, the maximum standard uptake value (SUVmax) of the tumours was included in the feature set. The fusion of FDG-PET and MR scans was carried out using the discrete wavelet transform (DWT) and a band-pass frequencies enhancement technique. Statistical analysis was performed using Spearman's correlation (rho), and multivariable modeling using logistic regression. The prediction performance of the different multivariable models was assessed using bootstrap resampling by calculating the area under the receiver-operating characteristics curve (AUC) and Matthews' correlation coefficient (MCC). The highest univariate prediction of lung metastases was attributed to the SUVmax metric (rho=0.6382, p<0.0001). Most texture features extracted from fused scans had higher Spearman's correlation with lung metastases than those extracted from separate scans. On separate scans, FDG-PET texture features were generally dominant over MR texture features. The highest multivariable prediction of lung metastases was found using fused scans and the following 4-parameters model: 0.94*SUVmax − 0.401*PET-T2FS/STIR--Variance − 6.7*PET-T1--Contrast − 165*PET-T1--Homogeneity + 140. This model reached rho=0.8255, p<0.0001 on the entire dataset and AUC=0.956±0.002, MCC=0.829±0.002 in bootstrap testing sets. Overall, this work indicates the strong potential of FDG-PET and MR texture features for the prediction of lung metastases in STS cancer of the extremities. Substantial prediction improvements were found using texture features from fused scans and multivariable modeling strategies compared to texture features extracted from separate scans and univariate analysis. Potentially, this could improve patient outcomes by allowing better personalization of treatments and the application of pre-emptive strategies to mitigate disease spread.
Les sarcomes des tissus mous (STM) provenant des extrémités forment un groupe relativement rare de néoplasme avec un risque métastatique élevé. La grande majorité des métastases provenant des STM ont lieu dans les poumons, et le pronostique résultant est généralement faible. En ce sens, il est important d'identifier autant de facteurs pronostiques pertinents que possible au moment du diagnostique et de la gestion du traitement. Certains travaux récents ont permis de démontrer que les caractéristiques texturales d'images provenant de la tomographie par émission de positrons (TEP) utilisant le fluorodéoxyglucose (FDG) et l'imagerie par résonance magnétique (IRM) ont le potentiel de prédire l'évolution tumorale grâce à l'évaluation des propriétés d'hétérogénéité biologique des tumeurs. Donc, le but de ce travail est d'évaluer le potentiel des caractéristiques texturales d'images FDG-TEP et IRM en tant que prédicteur du risque de métastases aux poumons pour le cancer des STM provenant des extrémités. Dans cette étude, une cohorte de 35 patients diagnostiqués avec des STM aux extrémités a été rétrospectivement analysée. Tous les patients ont reçu un scan FDG-TEP et un scan IRM avant leur traitement. Les séquences IRM qui ont été utilisés dans l'analyse sont: T1, T2 par saturation des gras (T2FS) et STIR. Les patients ont été suivis sur une période médiane de 29 mois (intervalle: 4 à 85 mois). Treize patients de la cohorte ont développé des métastases aux poumons. Six caractéristiques texturales d'images provenant de la matrice de co-occurrence des niveaux de gris (GLCM) ont été extraites des scans FDG-PET, IRM et FDG-PET/IRM fusionnés. De plus, la valeur maximale de consommation standard des tumeurs (SUVmax) a été incluse dans l'analyse. La fusion des scans a été effectuée grâce à la transformée d'ondelettes discrètes et grâce à une technique de renforcement des fréquences passe-bandes. L'analyse statistique a été effectuée en utilisant la corrélation de Spearman (rho), et l'analyse multivariable en utilisant la régression logistique. Les performances de prédiction des différents modèles multivariables ont été évaluées en calculant 2 métriques à partir de la technique de ré-échantillonnage « bootstrap »: L'aire sous la courbe de fonctionnement (AUC) et le coefficient de corrélation de Matthews (MCC). La plus haute prédiction univariée est attribuée à SUVmax (rho=0.6382, p<0.0001). La plupart des caractéristiques texturales extraites des scans fusionnés possèdent des coefficients de corrélation Spearman plus haut que celles extraites des scans séparés. Dans le cas des scans séparés, les caractéristiques texturales provenant de FDG-TEP sont généralement dominantes par rapport à celles provenant des scans IRM. La plus haute prédiction multivariable est provenue des scans fusionnés avec le model suivant: 0.94*SUVmax − 0.401*PET-T2FS/STIR--Variance − 6.7*PET-T1--Contrast − 165*PET-T1--Homogeneity + 140. Ce model a atteint des résultats de rho=0.8255, p<0.0001 sur l'ensemble des patients et AUC=0.956±0.002, MCC=0.829±0.002 sur les ensembles de tests « bootstrap ». De façon générale, cette étude indique le fort potentiel des caractéristiques texturales provenant des images FDG-TEP et IRM pour prédire les métastases aux poumons dans le cas des patients atteints des STM aux extrémités. Une amélioration substantielle des prédictions a pu être obtenue en utilisant les caractéristiques texturales des scans fusionnés et des stratégies d'analyse multivariable comparativement aux caractéristiques texturales des scans séparés et à l'analyse univariée. Potentiellement, cela pourrait mener à l'application de stratégies préventives pour atténuer la propagation du cancer des STM et à l'application de traitements mieux adaptés aux besoins des patients.

Glaucoma is a set of diseases that cause optic nerve damage and visual field loss. The most important risk factor for the development of glaucoma is elevated intraocular pressure. One approach used to alleviate the pressure increase is to surgically install glaucoma implants. Optical coherence tomography (OCT) is an imaging modality capable of acquiring cross-sectional images of tissue using back-reflected light. The images have a resolution of 10-15μm, and are thus best suited for visualizing tissue layers and structures. OCT images of some tissue types have few or no features in this size range but display a characteristic repetitive structure due to speckle. The purpose of this research was to show that OCT is capable of visualizing tissue changes, such as those associated with a healing response to glaucoma implants. A new OCT handheld probe was developed to facilitate in vivo imaging in rabbit eye studies. The OCT probe consisted of a mechanical scaffold designed to allow the imaging fiber to be held in a fixed position with respect to the rabbit eye, with minimal anesthesia. A piezo-electric lateral scanning device allowed the imaging fiber to be scanned across the tissue so that 2-D images may be acquired. Preliminary analysis of OCT images of two types of glaucoma implants indicates that OCT can visualize the development of fibrous encapsulation of the implant, tissue erosion and tube position in the anterior chamber. The application of statistical and spectral texture analysis techniques was investigated for differentiating tissue types based on the structural and speckle content in OCT images. Excellent correct classification rates were obtained for images of tissues and tissue phantoms that had slight visual differences and reasonable rates were obtained with nearly identical-appearing images of tissues and tissue phantoms. This study shows that OCT is capable of visualizing structural changes, associated with the healing response, on the order of tens to hundreds of microns. OCT also shows promise in being able to detect sub-resolution tissue healing response changes, by quantifying the changes in the speckle seen in OCT images, using texture analysis.

This dissertation presents work towards the development of a multiple finger, worn, dynamic display device, which utilizes a method of texture encoded information to haptically render graphical images for individuals who are blind or visually impaired. The device interacts directly with the computer screen, using the colors and patterns displayed by the image as a means to encode complex patterns of vibrotactile output, generating the texture feedback to render the image. In turn, the texture feedback was methodically designed to enable parallel processing of certain coarse information, speeding up the exploration of the diagram and improving user performance. The design choices were validated when individuals who are blind or visually impaired, using the multi-fingered display system, performed three-times better using textured image representations versus outline representations. Furthermore, in an open-ended object identification task, the display device saw on average two-times better performance accuracy than that previously observed for raised-line diagrams, the current standard for tactile diagrams.

Radiomics treats images as quantitative data and promises to improve cancer prediction in radiology and therapy response assessment in radiation oncology. However, there are a number of fundamental problems that need to be solved in order to potentially apply radiomic features in clinic. The first basic step in computed tomography (CT) radiomic analysis is the acquisition of images using selectable image acquisition and reconstruction parameters. Radiomic features have shown large variability due to variation of these parameters. Therefore, it is important to develop methods to address these variability issues in radiomic features due to each CT parameter. To this end, texture phantoms provide a stable geometry and Hounsfield Units (HU) to characterize the radiomic features with respect to image acquisition and reconstruction parameters. In this project, normalization methods were developed to address the variability issues in CT Radiomics using texture phantoms. In the first part of this project, variability in radiomic features due to voxel size variation was addressed. A voxel size resampling method is presented as a preprocessing step for imaging data acquired with variable voxel sizes. After resampling, variability due to variable voxel size in 42 radiomic features was reduced significantly. Voxel size normalization is presented to address the intrinsic dependence of some key radiomic features. After normalization, 10 features became robust as a function of voxel size. Some of these features were identified as predictive biomarkers in diagnostic imaging or useful in response assessment in radiation therapy. However, these key features were found to be intrinsically dependent on voxel size (which also implies dependence on lesion volume). The normalization factors are also developed to address the intrinsic dependence of texture features on the number of gray levels. After normalization, the variability due to gray levels in 17 texture features was reduced significantly. In the second part of the project, voxel size and gray level (GL) normalizations developed based on phantom studies, were tested on the actual lung cancer tumors. Eighteen patients with non-small cell lung cancer of varying tumor volumes were studied and compared with phantom scans acquired on 8 different CT scanners. Eight out of 10 features showed high (Rs > 0.9) and low (Rs < 0.5) Spearman rank correlations with voxel size before and after normalizations, respectively. Likewise, texture features were unstable (ICC < 0.6) and highly stable (ICC > 0.9) before and after gray level normalizations, respectively. This work showed that voxel size and GL normalizations derived from texture phantom also apply to lung cancer tumors. This work highlights the importance and utility of investigating the robustness of CT radiomic features using CT texture phantoms. Another contribution of this work is to develop correction factors to address the variability issues in radiomic features due to reconstruction kernels. Reconstruction kernels and tube current contribute to noise texture in CT. Most of texture features were sensitive to correlated noise texture due to reconstruction kernels. In this work, noise power spectra (NPS) was measured on 5 CT scanners using standard ACR phantom to quantify the correlated noise texture. The variability in texture features due to different kernels was reduced by applying the NPS peak frequency and the region of interest (ROI) maximum intensity as correction factors. Most texture features were radiation dose independent but were strongly kernel dependent, which is demonstrated by a significant shift in NPS peak frequency among kernels. Percent improvements in robustness of 19 features were in the range of 30% to 78% after corrections. In conclusion, most texture features are sensitive to imaging parameters such as reconstruction kernels, reconstruction Field of View (FOV), and slice thickness. All reconstruction parameters contribute to inherent noise in CT images. The problem can be partly solved by quantifying noise texture in CT radiomics using a texture phantom and an ACR phantom. Texture phantoms should be a pre-requisite to patient studies as they provide stable geometry and HU distribution to characterize the radiomic features and provide ground truths for multi-institutional validation studies.

The overall clinical motivation of this thesis is to differentiate between the different stages of liver disease stratifying into: no disease, mild disease, and severe fibrosis using Magnetic Resonance Imaging (MRI). As a related aim, we seek to differentiate as much as possible pericellular and nonpericellular fibrosis. This latter is clinically important, but currently no method exists that is able to perform this. Quickly, we realised that these aims push low level image analysis beyond their current bounds and so a great deal of the thesis is dedicated to extending such techniques before they can be applied. To work on the most fundamental low level image analysis concepts and algorithms we choose one of the most recent developments, namely continuous intrinsic dimensionality (ciD), which allows the continuous classification of homogeneous patches from 1D structures to intrinsically 2D structures. We show that the current formalism has several fundamental limitations and we propose a number of developments to improve on these. We re-evaluated feature energy statistics that were originally proposed in ciD, and additionally we examined the confidence one may have in stateof- the-art methods to estimate the orientation of features. We show that new statistical methods are required for feature energy, and that orientation predictability is more important than correctness of the estimation. This evaluation led us to the monogenic signal local orientation. Analysis of feature or texture energy is also a main contribution of this thesis. Within this framework we propose the Riesz-weighted phase congruency model. This is able to detect internal texture structures but it is not capable of delineating boundaries. Nevertheless, it proves an appropriate basis for texture quantification. Finally, we show that in contrast to using the standard established Kovesi approach, the developed texture measure leads to good results on the suboptimal T1w MRI liver image staging images. We show that we are able to differentiate automatically between the separate disease scores and between pericellular and non-pericellular fibrosis.

Positron Emission Tomography (PET) is an imaging modality that has become increasingly beneficial in Radiotherapy by improving treatment planning (1). PET reveals tumor volumes that are not well visualized on computed tomography CT or MRI, recognizes metastatic disease, and assesses radiotherapy treatment (1). It also reveals areas of the tumor that are more radiosensitive allowing for dose painting - a non-homogenous dose treatment across the tumor (1). However, PET is not without limitations. The quantitative unit of PET images, the Standardized Uptake Value (SUV), is affected by many factors such as reconstruction algorithm, patient weight, and tracer uptake time (2). In fact, PET is so sensitive that a patient imaged twice in a single day on the same machine and same protocol will produce different SUV values. The objective of this research was to increase the capabilities of PET by exploring other quantitative PET/CT measures for Radiotherapy treatment applications. The technique of quantitative image feature analysis, nowadays known as radiomics, was applied to PET and CT images. Image features were then extracted from PET/CT images and how the features differed between conventional and respiratory-gated PET/CT images in lung cancer was analyzed. The influence of noise on image features was analyzed by applying uncorrelated, Gaussian noise to PET/CT images and measuring how significantly noise affected features. Quantitative PET/CT measures outside of image feature analysis were also investigated. The correlation of esophageal metabolic tumor volumes (tumor volume demonstrating high metabolic uptake) and endoscopically implanted fiducial markers was studied. It was found that certain image features differed greatly between conventional and respiratory-gated PET/CT. The differences were mainly due to the effect of respiratory motion including affine motion, rotational motion and tumor deformation. Also, certain feature groups were more affected by noise than others. For instance, contour-dependent shape features exhibited the least change with noise. Comparatively, GLSZM features exhibited the greatest change with added noise. Discordance was discovered between the inferior and superior tumor fiducial markers and metabolic tumor volume (MTV). This demonstrated a need for both fiducial markers and MTV to provide a comprehensive view of a tumor. These studies called attention to the differences in features caused by factors such as motion, acquisition parameters, and noise, etc. Investigators should be aware of these effects. PET/CT radiomic features are indeed highly affected by noise and motion. For accurate clinical use, these effects must be account by investigators and future clinical users. Further investigation is warranted towards the standardization of PET/CT radiomic feature acquisition and clinical application.

Chronic Obstructive Pulmonary Disease (COPD), a growing health concern, is the fourth leading cause of death in the United States. While people habituated to smoking constitute the highest COPD susceptible population, people exposed to air pollution or other lung irritants also form a major group of potential COPD patients. COPD is a progressive disease that is characterized by the combination of chronic bronchitis, small airway obstruction, and emphysema that causes an overall decrease in the lung elasticity affecting the lung tissue. The current gold standard method to diagnose COPD is by pulmonary function tests (PFT) which measures the extent of COPD based on the lung volumes and is further classified into five severity stages. PFT measurements are insensitive to early stages of COPD and also its lack of reproducibility makes it hard to rely on, in assessing the disease progression. Alternatively, Pulmonary CT scans are considered as a major diagnostic tool in analyzing the COPD and CT measures are also closely related to the pathological extent of the disease. Quantification of COPD using features derived from CT images has been proven effective. The most common features are density based and texture based. We propose a new set of features called lung biomechanical features which capture the regional lung tissue deformation patterns during the respiratory cycle. We have tested these features on 75 COPD subjects and 15 normal subjects. We have done classification of COPD/Non COPD on the dataset using the three feature sets and also performed the classification all these subjects to their corresponding severity stage. It is shown that the lung biomechanical features were also able to classify COPD subjects with a good AUC. It is also shown that, by combining the best features from each feature set, there is an improvement in the classifier performance. Multiple regression analysis is performed to find the correlation between the CT derived features and PFT measurements.

I metodi di texture analysis al giorno d'oggi si basano prevalentemente su ROIs (regions of interest), il che può essere limitativo nel caso di malattie neurodegenerative come la malattia di Alzheimer. A tale proposito, in questa tesi si descrive un approccio voxelwise recentemente proposto in letteratura basato sulla tecnica chiamata VGLCM-TOP-3D, la quale mostra una migliore sensibilità ai cambiamenti tenui a livello di tessitura. Tramite uno studio multicenter è stata attuata una comparazione di affidabilità, confrontando tecnica ROI based con quella voxelwise. Infine, si è sfruttata la tecnica sulla malattia di Alzheimer, osservando soprattutto i cambiamenti di tessitura nel passaggio da mild cognitive impairmente a malattia di Alzheimer.

Questa tesi si propone di innovare lo stato dell’arte dei metodi di analisi dell’eterogeneità in lesioni polmonari attualmente utilizzati, affiancando l’analisi funzionale (emodinamica) a quella morfologica, grazie allo sviluppo di nuove feature specifiche. Grazie alla collaborazione tra il Computer Vision Group (CVG) dell’Università di Bologna e l’Unità Operativa di Radiologia dell’IRCCS-IRST di Meldola (Istituto di Ricovero e Cura a Carattere Scientifico – Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori), è stato possibile analizzare un adeguato numero di casi reali di pazienti affetti da lesioni polmonari primitive, effettuando un’analisi dell’eterogeneità sia su sequenze di immagini TC baseline sia contrast-enhanced, consentendo quindi un confronto tra eterogeneità morfologica e funzionale. I risultati ottenuti sono infine discussi sulla base del confronto con le considerazioni di natura clinica effettuate in cieco da due esperti radiologi dell’IRCCS-IRST.

Alzheimer’s disease (AD) is a chronic and progressive, irreversible syndrome that deteriorates the cognitive functions. Official death certificates of 2013 reported 84,767 deaths from Alzheimer’s disease, making it the 6th leading cause of death in the United States. The rate of AD is estimated to double by 2050. The neurodegeneration of AD occurs decades before symptoms of dementia are evident. Therefore, having an efficient methodology for the early and proper diagnosis can lead to more effective treatments. Neuroimaging techniques such as magnetic resonance imaging (MRI) can detect changes in the brain of living subjects. Moreover, medical imaging techniques are the best diagnostic tools to determine brain atrophies; however, a significant limitation is the level of training, methodology, and experience of the diagnostician. Thus, Computer aided diagnosis (CAD) systems are part of a promising tool to help improve the diagnostic outcomes. No publications addressing the use of Feedforward Artificial Neural Networks (ANN), and MRI image attributes for the classification of AD were found. Consequently, the focus of this study is to investigate if the use of MRI images, specifically texture and frequency attributes along with a feedforward ANN model, can lead to the classification of individuals with AD. Moreover, this study compared the use of a single view versus a multi-view of MRI images and their performance. The frequency, texture, and MRI views in combination with the feedforward artificial neural network were tested to determine if they were comparable to the clinician’s performance. The clinician’s performances used were 78 percent accuracy, 87 percent sensitivity, 71 percent specificity, and 78 percent precision from a study with 1,073 individuals. The study found that the use of the Discrete Wavelet Transform (DWT) and Fourier Transform (FT) low frequency give comparable results to the clinicians; however, the FT outperformed the clinicians with an accuracy of 85 percent, precision of 87 percent, sensitivity of 90 percent and specificity of 75 percent. In the case of texture, a single texture feature, and the combination of two or more features gave results comparable to the clinicians. However, the Gray level co-occurrence matrix (GLCOM), which is the combination of texture features, was the highest performing texture method with 82 percent accuracy, 86 percent sensitivity, 76 percent specificity, and 86 percent precision. Combination CII (energy and entropy) outperformed all other combinations with 78 percent accuracy, 88 percent sensitivity, 72 percent specificity, and 78 percent precision. Additionally, a combination of views can increase performance for certain texture attributes; however, the axial view outperformed the sagittal and coronal views in the case of frequency attributes. In conclusion, this study found that both texture and frequency characteristics in combinations with a feedforward backpropagation neural network can perform at the level of the clinician and even higher depending on the attribute and the view or combination of views used.

Le immagini termiche all’ infrarosso sono utilizzate da molte decadi per monitorare la distribuzione di temperatura della pelle umana. Anormalità come infiammazioni ed infezioni causano un aumento locale della temperatura, visibile sulle immagini sotto forma di hot spot. In tal senso la termografia ad infrarossi può essere utilizzata nel campo ortopedico per rilevare le zone sovra-caricate dalla protesi. Per effettuare una valutazione precisa dell’interfaccia moncone-invasatura può essere utile combinare i dati termografici e i dati antropometrici (superficie tridimensionale del moncone), relativi ai singoli pazienti. Di ciò si occupa tale studio, che dopo aver fornito una panoramica sulla termografia e sulla reverse engineering, sperimenta delle tecniche semplici e low-cost per combinare i dati termici e i dati antropometrici. Buoni risultati si riescono ad ottenere utilizzando un Kinect come scanner e un software open-source per il texture mapping. I termogrammi 3D ricreati costituiscono un ottimo strumento di valutazione, per medici e tecnici ortopedici, circa il design dell’invasatura.

Questo studio delinea la fattibilità di una nuova tecnica chiamata Istologia Virtuale dove lo stato del tessuto viene valutato direttamente estraendo ed analizzando valori Hounsfield da CT cardiache. Il tessuto viene prelevato in diverse posizioni del setto intraventricolare e della parete esterna del ventricolo sinistro per valutarne lo stato. Viene proposto un flusso di lavoro riproducibile e ripetibile per ottenere campioni con un volume di 1 cc partendo da immagini 2D. I campioni estratti vengono analizzati per creare un modello rappresentativo dei soggetti sani e viene poi confrontato con quello di soggetti affetti da rottura del setto intraventricolare post-infarto e cardiomiopatia ipetrofica, per diagnosticare la patologia direttamente dallo stato dei tessuti e non dalla sintomatologia. Vengono presentati e discussi i migliori parametri per caratterizzare il profilo densitometrico specifico di una porzione 3D di tessuto. Anche se i risultati per media, deviazione standard ed entropia sono incoraggianti, non sono sufficienti per distinguere accuratamente le patologie, quindi grazie ad un'ulteriore estrazione di features ci si è spostati ad un problema di Machine Learning. I nuovi parametri sono legati all'intensità del singolo pixel, proprietà dei profili densitometrici, e alla texture. Le misure di intensità sono caratteristiche 3D mentre per la texture, i risultati sono una media di slices 2D. Lo sbilanciamento delle classi ostacola la corretta classificazione di alcuni dati appartenenti ai soggetti affetti da patologie, ma le tecniche di Data Augmentation permettono di fornire una prova di ciò che sarà possibile in futuro con una corte bilanciata di soggetti. Uno studio di Feature Importance viene eseguito per capire quali caratteristiche sono più rappresentative nel dividere il dataset e il risultato è che la moda del profilo è una delle migliori per classificare i campioni seguita correlazione fra due pixels e dal contrasto.

INTRODUZIONE. In campo oncologico la Risonanza Magnetica multiparametrica (RMmp) sta diventando sempre più importante in quanto sequenze multiple, quali T2 e DWI, permettono di rilevare la presenza di tumore, ovvero di zone ipo-iper-intense rispetto al tessuto circostante. L’analisi delle immagini mediante tecniche di texture analysis può essere utilizzata per cercare dei potenziali biomarker che aiutino il radiologo nella diagnosi svolgendo un ruolo di second opinion per detection e diagnosis di una lesione ancor prima di aver eseguito la biopsia. MATERIALI E METODI. In questa Tesi sono state analizzate le sequenze T2 e DWI di RMmp-3T per identificare le neoplasie prostatiche mediante tecniche di texture analysis con estrazione di feature su patch locali centrate sui pixel d’interesse, la cui distribuzione spaziale è stata visualizzata ed analizzata mediante colormap su cui è stato applicato un algoritmo di segmentazione automatica. Sono state calcolate sette feature e per ciascuna nove descrittori statistici. RISULTATI. Dalla feature più significativa (entropia) nel caratterizzare l’eterogeneità del tessuto, sono state rilevate automaticamente le lesioni e confrontate con quelle segmentate manualmente dal medico, valutando mutualmente le performance ottenute (lesioni rilevate nel 96.55% dei casi con una media del 72.42% di lesioni con il radiologo come ground-truth e con una media del 53.33% di lesioni viceversa). Le feature locali risultano inoltre ben correlate con i biomarker clinici per la stratificazione delle lesioni (ρ=0.695 tra Skewness della Media e PSALT e ρ=0.790 tra la Kurtosi della Mediana e PSALT). CONCLUSIONI. I risultati ottenuti incoraggiano futuri approfondimenti nello studio della texture analysis che deve comprendere un’analisi multiparametrica di tutte le feature ed essere estesa a tutte le sequenze disponibili dalla RMmp. La metodologia sviluppata ha le caratteristiche per embrionali di un sistema di Computer Aided Detection/Diagnosis (CAD).

Made available in DSpace on 2016-08-17T14:53:26Z (GMT). No. of bitstreams: 1 Leonardo Martins.pdf: 1400853 bytes, checksum: 3b6aa06e1c4b580a53150460124fdeaa (MD5) Previous issue date: 2007-12-07
Coordenação de Aperfeiçoamento de Pessoal de Nível Superior
Breast cancer is a serious public health problem in several countries of the world. Computer-Aided Detection/Diagnosis systems (CAD/CADx) have been used with relative success in aid to health care professionals. The goal of such systems is not to replace the professional, but join forces in order to early detect the different types of cancer. The main contribution of this work is to present a methodology for detecting masses in digitized mammograms using the algorithm Growing Neural Gas for the segmentation of the image and Ripley’s K function to describe the texture of segmented objects. The classification of these objects is accomplished through a Support Vector Machine (SVM), which separates them into two groups: masses and non-masses. The methodology obtained 89,30% of accuracy and a rate of 0,93 false-positive per image.
O câncer de mama apresenta-se como um grave problema de saúde pública em vários países do mundo. Sistemas de Detecção e Diagnóstico baseados em computador (CAD/CADx) vêm sendo usados com relativo sucesso no auxílio aos profissionais de saúde. O objetivo de tais sistemas não é substituir o profissional, mas unir forças com o objetivo de detectar precocemente os diferentes tipos de câncer. A principal contribuição deste trabalho é apresentar uma metodologia para detecção de massas em imagens mamográficas digitais, utilizando para tanto o algoritmo Growing Neural Gas para a segmentação da imagem e a função K de Ripley para descrever a textura dos objetos segmentados. A classificação desses objetos é feita através de uma Máquina de Vetor de Suporte (Support Vector Machine - SVM), a qual separa os mesmos em dois grupos: massa e não-massa. A metodologia obteve 89,30% de acerto e uma taxa de 0,93 falso-positivos por imagem.

This biomedical engineering thesis explores the opportunities and challenges of 2D+t contrast echocardiography for left ventricle functional analysis, both clinically and within a computer vision atlas-based deformable template model framework. A database was created for the experiments in this thesis, with 21 studies of contrast Dobutamine Stress Echo, in all 4 principal planes. The database includes clinical variables, human expert hand-traced myocardial contours and visual scoring. First the problem is studied from a clinical perspective. Quantification of endocardial global and local function using standard measures shows expected values and agreement with human expert visual scoring, but the results are less reliable for myocardial thickening. Next, the problem of segmenting the endocardium with a computer is posed in a standard landmark and atlas-based deformable template model framework. The underlying assumption is that these models can emulate human experts in terms of integrating previous knowledge about the anatomy and physiology with three sources of information from the image: texture, geometry and kinetics. Probabilistic atlases of contrast echocardiography are computed, while noting from histograms at selected anatomical locations that modelling texture with just mean intensity values may be too naive. Intensity analysis together with the clinical results above suggest that lack of external boundary definition may preclude this imaging technique for appropriate measuring of myocardial thickening, while endocardial boundary definition is appropriate for evaluation of wall motion. Geometry is presented in a Principal Component Analysis (PCA) context, highlighting issues about Gaussianity, the correlation and covariance matrices with respect to physiology, and analysing different measures of dimensionality. A popular extension of deformable models ---Active Appearance Models (AAMs)--- is then studied in depth. Contrary to common wisdom, it is contended that using a PCA texture space instead of a fixed atlas is detrimental to segmentation, and that PCA models are not convenient for texture modelling. To integrate kinetics, a novel spatio-temporal model of cardiac contours is proposed. The new explicit model does not require frame interpolation, and it is compared to previous implicit models in terms of approximation error when the shape vector changes from frame to frame or remains constant throughout the cardiac cycle. Finally, the 2D+t atlas-based deformable model segmentation problem is formulated and solved with a gradient descent approach. Experiments using the similarity transformation suggest that segmentation of the whole cardiac volume outperforms segmentation of individual frames. A relatively new approach ---the inverse compositional algorithm--- is shown to decrease running times of the classic Lucas-Kanade algorithm by a factor of 20 to 25, to values that are within real-time processing reach.

This work explores the blending of e-textile technology with the porous electrode of polymer electrolyte membrane fuel cells (PEMFCs) and with smart wound patches to allow monitoring and in-situ diagnostics. This work includes contributions to understanding water transport and conductivity in the carbon cloth gas diffusion layer (GDL), and further developing thread-based relative humidity (RH) and temperature sensors, which can be sewn on a cloth GDL in PEMFCs. We also explore the application of the developed RH and temperature sensors in wearable biomonitoring. First, an experimental prototype is developed for evaluating water transport, thermal conductivity and electrical conductivity of carbon cloth GDLs under different hydrophobic coatings and compressions. Second, we demonstrate the addition of external threads to the carbon cloth GDL to (1) facilitate water transport and (2) measure local RH and temperature with a minimal impact on the physical, microstructural and transport properties of the GDL. We illustrate the roll-to-roll process for fabricating RH and temperature sensors by dip-coating commodity threads into a carbon nanotubes (CNTs) suspension. The thread-based sensors response to RH and temperature in the working environment of PEMFCs is investigated. As a proof-of-concept, the local temperature of carbon cloth GDL is monitored in an ex-situ experiment. Finally, we optimized the coating parameters (e.g. CNTs concentration, surfactant concentration and a number of dipping) for the thread-based sensors. The response of the thread-based sensors in room conditions is evaluated and shows a linear resistance decrease to temperature and a quadratic resistance increase to RH. We also evaluated the biocompatibility of the sensors by performing cell cytotoxicity and studying wound healing in an animal model. The novel thread-based sensors are not only applicable for textile electrochemical devices but also, show a promising future in wearable biomonitoring applications.
Graduate

Articular cartilage is the connective tissue that lines the ends of long bones in diarthrodial joints, providing a low-friction load-bearing surface that can withstand a lifetime of loading cycles under normal conditions. Despite these unique and advantageous properties, the tissue possesses a limited capacity for self-repair due to its lack of vasculature and innervation. Total joint replacement is a well-established treatment for degenerative joint disease; however, the materials used in these procedures have a limited lifespan in vivo and will likely fail over time, requiring additional - and increasingly complicated - revision surgeries. For younger or more active patients, this risk is unacceptable. Unfortunately, alternative surgical options are not currently available, leaving pain management as the only viable treatment. In seeking to discover a new therapeutic strategy, the goal of this dissertation was to develop a functional tissue-engineered cartilage construct that may be used to resurface an entire diseased or damaged joint.

A three-dimensional (3-D) woven textile structure, produced on a custom-built miniature weaving loom, was utilized as the basis for producing novel composite scaffolds and cartilage tissue constructs that exhibited initial properties similar to those of native articular cartilage. Using polyglycolic acid (PGA) fibers combined with chondrocyte-loaded agarose or fibrin hydrogels, scaffolds were engineered with anisotropic, inhomogeneous, viscoelastic, and nonlinear characteristics prior to cultivation. However, PGA-based constructs showed a rapid loss of mechanical functionality over a 28 day culture period suggesting that the inclusion of other, less degradable, biomaterial fibers could provide more stable properties.

Retaining the original 3-D architecture and fiber/hydrogel composite construction, poly (epsilon-caprolactone) (PCL)-based scaffolds demonstrated initial biomechanical properties similar to those of PGA-based scaffolds. Long-term culture of 3-D PCL/fibrin scaffolds seeded with human adipose-derived stem cells (ASCs) showed that scaffolds maintained their baseline properties as new, collagen-rich tissue accumulated within the constructs.

In an attempt to improve the bioactivity of the PCL scaffold and further induce chondrogenic differentiation of seeded ASCs, we produced a hybrid scaffold system by embedding the 3-D woven structure within a porous matrix derived from native cartilage. We then demonstrated how this multifunctional scaffold could be molded, seeded, and cultured in order to produce an anatomically accurate tissue construct with potential for resurfacing the femoral head of a hip.

In summary, these findings provide valuable insight into a new approach for the functional tissue engineering of articular cartilage. The results of this work will hopefully lead to the discovery of new strategies for the long-term treatment of cartilage pathology.

Dissertation

Dissertação de mestrado em Química Têxtil
Os avanços na tecnologia dos biosensores tem vindo a revolucionar os cuidados de saúde, nomeadamente no diagnóstico de condições patológicas de forma rápida e eficaz, tornando o uso destes dispositivos uma realidade. O objetivo deste trabalho foi desenvolver tecidos condutores, sintéticos e naturais, compostos por poliamida (PA) ou algodão (CO), respetivamente, utilizando o polipirrole (PPy) e a polianilina (PANi) como polímeros condutores. Numa fase inicial, foi realizada uma polimerização in situ utilizando os respetivos monómeros (pirrole, Py e anilina, ANi) e um agente oxidante (persulfato de amónio, APS). A otimização dos vários parâmetros, como a concentração dos monómeros (Py e ANi), agente oxidante (APS), tempo de tingimento/polimerização e homogeneidade do tingimento foram analisados de acordo com um plano experimental criado, forma a obter tecidos têxteis com elevada condutividade. Aplicou-se um produto de acabamento, composto por poliuretano, de forma a conservar as principais características dos tecidos. Os tecidos foram caracterizados em termos de microestrutura, hidrofobicidade, composição química, solidez da cor à lavagem doméstica e industrial, solidez da cor à fricção e citotoxicidade. Adicionalmente, os tecidos de poliamida foram sujeitos a um pré-tratamento com plasma com o objetivo de melhorar a sua hidrofilicidade. As análises de SEM permitiram verificar a integridade estrutural das fibras/tecidos após o processo de polimerização. A medição dos ângulos de contato demonstrou que o processo de polimerização promoveu um aumento da hidrofobicidade em algumas condições. Na análise elementar por XPS foi possível identificar os picos característicos dos tecidos de poliamida ou algodão e dos polímeros condutores (PPy ou PANi). Relativamente aos ensaios de solidez da cor à lavagem doméstica e industrial e à fricção, verificou-se que os tecidos apresentaram valores baixos, porém, na generalidade, isso não afetou nem a condutividade superficial nem volumétrica. Excecionalmente, os tecidos de algodão tingidos com PANi alteraram as suas propriedades semicondutoras para isoladoras após o ensaio de solidez da cor à lavagem doméstica e industrial. Todos os tecidos, com exceção dos tecidos naturais tingidos com PANi, independentemente da aplicação do produto de acabamento, não apresentaram nenhum efeito citotóxico, demonstrando a sua biocompatibilidade para serem usados em aplicações biomédicas. Em conclusão, este estudo demonstra o desenvolvimento e caracterização de tecidos têxteis, sintéticos e naturais, condutores com um grande potencial para serem utilizados em futuras aplicações biomédicas.
The latest advances in biosensor technology has been revolutionize healthcare and diagnosis of pathological condition and its fast and specific treatment, making these devices a reality. The purpose of this work was to developed synthetic and natural conductive fabrics composed of polyamide (PA) and cotton (CO), respectively, for future biomedical applications, using PPy and PANi as conductive polymers. Firstly, the polymerization of the conductive polymers was made in situ using the correspondent monomers (pyrrole, Py and aniline, ANi) and an oxidizing agent (ammonium persulfate, APS). The experimental parameters like molar concentrations of monomers (Py and ANi) and APS, dyeing/polymerization time and dyeing homogeneity were adjusted according the results obtained from an experimental design created to achieve high values of conductivity. A finishing product composed of polyurethane was also applied to preserve fabrics main properties. Fabrics were characterized in terms of microstructure, hydrophobicity, chemical composition, color fastness of domestic and industrial washing, color fastness to rubbing and cytotoxicity. A pre-treatment by plasma was performed at PA to improve hydrophilicity. SEM analysis allowed to confirm the morphological integrity of the fibers/fabrics upon the polymerization process. Contact angle measurements have shown that, under certain conditions, the polymerization process affects the hydrophobicity of the final product. The elements analysis conducted by XPS allowed to identify the characteristic peaks of PA/CO fabrics and of the conductive polymers (PPy or PANi). Related to the color fastness to domestic and industrial washing and to rubbing, both types of fabrics have shown low values in both tests, but surface and volumetric conductivities were not affected after testing. Exceptionally, CO fabrics containing PANi changed their semiconductive properties to insulator after color fastness to domestic and industrial testing. All fabrics, with the exception of natural fabrics containing PANi, independently of the finishing product application, did not show any cytotoxic effect evidencing their biocompatibility to be used in biomedical applications. In conclusion, this study demonstrates the development and characterization of synthetic and natural conductive fabrics with great potential to be used in future biomedical applications.

碩士
國立清華大學
電機工程學系
91
For some biomedical purpose, a specific region needs to be extracted from a target 3-D image stack which usually consists of many 2-D slices. In previous work, a template-driven segmentation scheme by using snake is presented to find the approximate outline and position of the specific region on each slice of the target image stack based on the template (reference) 3-D image stack. In this thesis, a modified auto-segmentation process will be proposed. A ROI-finding scheme and a texture-based segmentation algorithm are used to avoid the original drawbacks. There are mainly four stages presented in this thesis to achieve the refinement: (i) ROI (region of interest)─finding process is used to filter out unnecessary area and locate the specific region. (ii) Texture-based multi-resolution segmentation scheme is used to segment target images by using local spectral histogram. (iii) Volume-smoothing method is used to smooth segmentation mask of each slice along depth-axis. (iv) Morphology is used to make the suggested contour much smoother. This process is more efficient than extracting the specific region by hand. Moreover, the problem of the blurred edges in the specific region is also addressed.

The ability to detect atherosclerotic plaque from optical coherence tomography (OCT) images by visual inspection is usually limited. We developed a texture based segmentation method using supervised and unsupervised classification to detect atherosclerotic plaque from OCT images without any reliance on visual inspection. Our Supervised method involves extraction of statistical textural features using the Spatial Gray Level Dependence Matrix (SGLDM) method, feature extraction and feature selection method, and application on supervised algorithm (K-nn). Our second method is based on unsupervised classification involves extraction of statistical textural features using the SGLDM method, application of an unsupervised clustering algorithm (K-means) on these features, and mapping of the segmented regions of features back to the actual image. We verified our results by visually comparing them to photographs of the vascular tissue with atherosclerotic plaque that we used to generate our OCT images. Our method could be potentially used in clinical cardiovascular OCT imaging.

Computed tomography (CT) imaging enables in vivo assessment of lung parenchyma and several lung diseases. CT scans are key in particular for the diagnosis of 1) chronic obstructive pulmonary disease (COPD), which is the fourth leading cause of death worldwide, and largely overlaps with pulmonary emphysema; and 2) lung cancer, which is the first leading cause of cancer-related death, and manifests in its early stage with the presence of lung nodules. Most lung CT image analysis methods to-date have relied on supervised learning requiring manually annotated local regions of interest (ROIs), which are slow and labor-intensive to obtain. Machine learning models requiring less or no manual annotations are important for a sustainable development of computer-aided diagnosis (CAD) systems. This thesis focused on exploiting CT scans for lung disease characterization via two learning strategies: 1) fully unsupervised learning on a very large amount of unannotated image patches to discover novel lung texture patterns for pulmonary emphysema; and 2) weakly-supervised learning to generate voxel-level localization of lung nodules from CT whole-slice labels. In the first part of this thesis, we proposed an original unsupervised approach to learn emphysema-specific radiological texture patterns. We have designed dedicated spatial and texture features and a two-stage learning strategy incorporating clustering and graph partitioning. Learning was performed on a cohort of 2,922 high-resolution full-lung CT scans, which included a high prevalence of smokers and COPD subjects. Experiments lead to discovering 10 highly-reproducible spatially-informed lung texture patterns and 6 quantitative emphysema subtypes (QES). Our discovered QES were associated independently with distinct risk of symptoms, physiological changes, exacerbations and mortality. Genome-wide association studies identified loci associated with four subtypes. Then we designed a deep-learning approach, using unsupervised domain adaptation with adversarial training, to label the QES on cardiac CT scans, which included approximately 70% of the lung. Our proposed method accounted for the differences in CT image qualities, and enabled us to study the progression of QES on a cohort of 17,039 longitudinal cardiac and full-lung CT scans. Overall, the discovered QES provide novel emphysema sub-phenotyping that may facilitate future study of emphysema development, understanding the stages of COPD and the design of personalized therapies. In the second part of the thesis, we have designed a deep-learning method for lung nodule detection with weak labels, using classification convolutional neural networks (CNNs) with skip-connections to generate high-quality discriminative class activation maps, and a novel candidate screening framework to reduce the number of false positives. Given that the vast majority of annotated nodules are benign, we further exploited a data augmentation framework with a generative adversarial network (GAN) to address the issue of data imbalance for lung cancer prediction. Our weakly-supervised lung nodule detection on 1,000s CT scans achieved competitive performance compared to a fully-supervised method, while requiring 100 times less annotations. Our data augmentation framework enabled synthesizing nodules with high fidelity in specified categories, and is beneficial for predicting nodule malignancy scores and hence improving the accuracy / reliability of lung cancer screening.