OBJETIVO: Investigar a correlação entre as espessuras da mácula e da camada de fibras nervosas da retina (CFNR), medidas pelas tomografias de coerência óptica de domínio Fourier (FD - OCT) e de domínio do tempo (TD - OCT) e a perda de sensibilidade no campo visual (CV) em pacientes com atrofia em banda do nervo óptico. Comparar a habilidade diagnóstica dos dois instrumentos. MÉTODOS: 36 olhos de 36 pacientes com perda de CV permanente por compressão do quiasma óptico e 36 controles normais foram submetidos ao exame de CV pela perimetria automatizada padrão (Humphrey Field Analyzer TM; Carl Zeiss Meditec, Dublin, CA), ao FD - OCT (3 D OCT-1000TM Topcon Corp., Tokyo, Japan) e ao TD - OCT (StratusTM ;Carl Zeiss Meditec Inc, Dublin, California, USA). Foram comparados os protocolos análogos de ambos os equipamentos para avaliação da espessura macular.Foi realizado a divisão macular em quatro quadrantes e em metades, sendo calculado a média da espessura macular global, a média da espessura dos quadrantes e a média da espessura das metades maculares. A média global e setorial da medida da espessura da CFNR peripapilar também foi analisada. A perda de sensibilidade no CV foi inicialmente avaliada pelo defeito temporal médio. O CV foi ainda dividido em 6 setores de acordo com a distribuição da CFNR e em 16 pontos centrais para a realização da correlação estrutura-função entre os parâmetros medidos pelo FD - OCT e a perda de sensibilidade no CV em decibéis e 1/Lambert. Foi calculado o coeficiente de correlação de Sperman e a análise de regressão linear. As áreas sobre a curva ROC e valores fixos de sensibilidade e especificidade foram calculados para cada parâmetro estudado. RESULTADOS: As medidas da espessura macular e da CFNR pelo FD - OCT e TD - OCT foram capazes de discriminar olhos com atrofia em banda do nervo óptico dos controles normais. A espessura global e setorial dos parâmetros maculares e da CFNR mostraram diferenças significativas (p<0,001) entre os doentes e os controles e os dois aparelhos tiveram desempenho semelhante na discriminação entre pacientes e controles. Em ambos existiram correlações significativas entre a perda de sensibilidade do CV e as medidas da espessura macular e da CFNR. As medidas de espessura nos quadrantes e nas metades nasais da macula, avaliadas pelo FD - OCT tiveram os melhores desempenhos nas correlações com os defeitos de CV, sendo o parâmetro com a melhor correlação, a medida do quadrante infero-nasal da mácula e a perda de sensibilidade do CV central do quadrante temporal superior central (r = 0.78, R2 = 61%, p<0,001). CONCLUSÕES: A espessura macular e a espessura da CFNR medidas pelo FD - OCT e pelo TD - OCT se correlacionaram topograficamente com a perda de sensibilidade no CV de pacientes com hemianopsia temporal por compressäo quiasmática. A correlação entre os quadrantes maculares e a perda de sensibilidade no CV, foram melhores do que aquelas entre o CV e as medidas da CFNR, principalmente nas medidas realizadas pelo FD - OCT. Este estudo demonstrou a importância clínica das medidas maculares na correlação estrutura-função e na quantificaçäo do dano neural em pacientes com compressão quiasmática, podendo ser útil na monitorização destes pacientes / PURPOSE: To investigate the relationship between fourier-domain optical coherence tomography (FD - OCT) measured macular and retinal nerve fiber layer thickness (RNFL) and visual field sensitivity loss on standard automated perimetry in eyes with permanent temporal hemianopia from chiasmal compression and compare the ability of FD - OCT and time-domain optical coherence tomography (TD - OCT) to detect axonal loss in eyes with band atrophy of the optic nerve. METHODS: One eye of each of 36 patients with permanent temporal visual field defects and 36 age- and sex-matched healthy subjects. Subjects underwent standard automated perimetry and macular and RNFL thickness measurements with FD - OCT (3 D OCT-1000® Topcon Corp., Tokyo, Japan) and TD - OCT (Stratus; Carl Zeiss Meditec Inc, Dublin, California, USA). Macular thickness measurements as a global average, divided in four quadrants and in two halves as well as average and sectoral RNFL thickness around the optic disc were calculated. Visual field sensitivity loss was evaluated by the temporal mean defect; as deviations from normal in six sectors of the visual field and in 16 central visual field test points. Relationship between visual field sensitivity loss in decibel and 1/Lambert units and optical coherence tomography measurements were evaluated using Spearman correlation coefficients and by linear regression analysis. Receiver operating characteristic curves and sensitivities at fixed specificities were calculated for each parameter. RESULTS: Global and sectoral macular and RNFL thickness parameters showed a significant difference in eyes with band atrophy compared with controls. The strongest correlations were seen between visual field sensitivity loss and quadrantic or hemianopic nasal macular thickness measurements than with sectoral retinal nerve fiber layer thickness measurements. The highest correlation was observed between the inferonasal quadrant macular thickness and the visual field sensitivity loss in the superior temporal central visual field quadrant (r = 0.78, R2 = 61%, p <0.001). CONCLUSION: Both RNFL thickness and macular thickness FD - OCT and TD - OCT measurements were related topographically with visual field sensitivity loss in patients with temporal hemianopia from chiasmal compression and there is a stronger relationship in quadrantic macular compared to RNFL thickness measurements with FD - OCT. Macular thickness measurements could potentially be used to quantify neuronal loss in patients with chiasmal compression and could prove clinically useful for detection of damage and for monitoring these patients

Motivated by the classic works of Charles M. Stein, we focus on developing risk-estimation frameworks for denoising problems in both one-and two-dimensions. We assume a standard additive noise model, and formulate the denoising problem as one of estimating the underlying clean signal from noisy measurements by minimizing a risk corresponding to a chosen loss function. Our goal is to incorporate perceptually-motivated loss functions wherever applicable, as in the case of speech enhancement, with the squared error loss being considered for the other scenarios. Since the true risks are observed to depend on the unknown parameter of interest, we circumvent the roadblock by deriving finite-sample un-biased estimators of the corresponding risks based on Stein’s lemma. We establish the link with the multivariate parameter estimation problem addressed by Stein and our denoising problem, and derive estimators of the oracle risks. In all cases, optimum values of the parameters characterizing the denoising algorithm are determined by minimizing the Stein’s unbiased risk estimator (SURE). The key contribution of this thesis is the development of a risk-estimation approach for choosing the two critical parameters affecting the quality of nonparametric regression, namely, the order and bandwidth/smoothing parameters. This is a classic problem in statistics, and certain algorithms relying on derivation of suitable finite-sample risk estimators for minimization have been reported in the literature (note that all these works consider the mean squared error (MSE) objective). We show that a SURE-based formalism is well-suited to the regression parameter selection problem, and that the optimum solution guarantees near-minimum MSE (MMSE) performance. We develop algorithms for both glob-ally and locally choosing the two parameters, the latter referred to as spatially-adaptive regression. We observe that the parameters are so chosen as to tradeoff the squared bias and variance quantities that constitute the MSE. We also indicate the advantages accruing out of incorporating a regularization term in the cost function in addition to the data error term. In the more general case of kernel regression, which uses a weighted least-squares (LS) optimization, we consider the applications of image restoration from very few random measurements, in addition to denoising of uniformly sampled data. We show that local polynomial regression (LPR) becomes a special case of kernel regression, and extend our results for LPR on uniform data to non-uniformly sampled data also. The denoising algorithms are compared with other standard, performant methods available in the literature both in terms of estimation error and computational complexity. A major perspective provided in this thesis is that the problem of optimum parameter choice in nonparametric regression can be viewed as the selection of optimum parameters of a linear, shift-invariant filter. This interpretation is provided by deriving motivation out of the hallmark paper of Savitzky and Golay and Schafer’s recent article in IEEE Signal Processing Magazine. It is worth noting that Savitzky and Golay had shown in their original Analytical Chemistry journal article, that LS fitting of a fixed-order polynomial over a neighborhood of fixed size is equivalent to convolution with an impulse response that is fixed and can be pre-computed. They had provided tables of impulse response coefficients for computing the smoothed function and smoothed derivatives for different orders and neighborhood sizes, the resulting filters being referred to as Savitzky-Golay (S-G) filters. Thus, we provide the new perspective that the regression parameter choice is equivalent to optimizing for the filter impulse response length/3dB bandwidth, which are inversely related. We observe that the MMSE solution is such that the S-G filter chosen is of longer impulse response length (equivalently smaller cutoff frequency) at relatively flat portions of the noisy signal so as to smooth noise, and vice versa at locally fast-varying portions of the signal so as to capture the signal patterns. Also, we provide a generalized S-G filtering viewpoint in the case of kernel regression. Building on the S-G filtering perspective, we turn to the problem of dynamic feature computation in speech recognition. We observe that the methodology employed for computing dynamic features from the trajectories of static features is in fact derivative S-G filtering. With this perspective, we note that the filter coefficients can be pre-computed, and that the whole problem of delta feature computation becomes efficient. Indeed, we observe an advantage by a factor of 104 on making use of S-G filtering over actual LS polynomial fitting and evaluation. Thereafter, we study the properties of first-and second-order derivative S-G filters of certain orders and lengths experimentally. The derivative filters are bandpass due to the combined effects of LPR and derivative computation, which are lowpass and highpass operations, respectively. The first-and second-order S-G derivative filters are also observed to exhibit an approximately constant-Q property. We perform a TIMIT phoneme recognition experiment comparing the recognition accuracies obtained using S-G filters and the conventional approach followed in HTK, where Furui’s regression formula is made use of. The recognition accuracies for both cases are almost identical, with S-G filters of certain bandwidths and orders registering a marginal improvement. The accuracies are also observed to improve with longer filter lengths, for a particular order. In terms of computation latency, we note that S-G filtering achieves delta and delta-delta feature computation in parallel by linear filtering, whereas they need to be obtained sequentially in case of the standard regression formulas used in the literature. Finally, we turn to the problem of speech enhancement where we are interested in de-noising using perceptually-motivated loss functions such as Itakura-Saito (IS). We propose to perform enhancement in the discrete cosine transform domain using risk-minimization. The cost functions considered are non-quadratic, and derivation of the unbiased estimator of the risk corresponding to the IS distortion is achieved using an approximate Taylor-series analysis under high signal-to-noise ratio assumption. The exposition is general since we focus on an additive noise model with the noise density assumed to fall within the exponential class of density functions, which comprises most of the common densities. The denoising function is assumed to be pointwise linear (modified James-Stein (MJS) estimator), and parallels between Wiener filtering and the optimum MJS estimator are discussed.

Motivated by the classic works of Charles M. Stein, we focus on developing risk-estimation frameworks for denoising problems in both one-and two-dimensions. We assume a standard additive noise model, and formulate the denoising problem as one of estimating the underlying clean signal from noisy measurements by minimizing a risk corresponding to a chosen loss function. Our goal is to incorporate perceptually-motivated loss functions wherever applicable, as in the case of speech enhancement, with the squared error loss being considered for the other scenarios. Since the true risks are observed to depend on the unknown parameter of interest, we circumvent the roadblock by deriving finite-sample un-biased estimators of the corresponding risks based on Stein’s lemma. We establish the link with the multivariate parameter estimation problem addressed by Stein and our denoising problem, and derive estimators of the oracle risks. In all cases, optimum values of the parameters characterizing the denoising algorithm are determined by minimizing the Stein’s unbiased risk estimator (SURE). The key contribution of this thesis is the development of a risk-estimation approach for choosing the two critical parameters affecting the quality of nonparametric regression, namely, the order and bandwidth/smoothing parameters. This is a classic problem in statistics, and certain algorithms relying on derivation of suitable finite-sample risk estimators for minimization have been reported in the literature (note that all these works consider the mean squared error (MSE) objective). We show that a SURE-based formalism is well-suited to the regression parameter selection problem, and that the optimum solution guarantees near-minimum MSE (MMSE) performance. We develop algorithms for both glob-ally and locally choosing the two parameters, the latter referred to as spatially-adaptive regression. We observe that the parameters are so chosen as to tradeoff the squared bias and variance quantities that constitute the MSE. We also indicate the advantages accruing out of incorporating a regularization term in the cost function in addition to the data error term. In the more general case of kernel regression, which uses a weighted least-squares (LS) optimization, we consider the applications of image restoration from very few random measurements, in addition to denoising of uniformly sampled data. We show that local polynomial regression (LPR) becomes a special case of kernel regression, and extend our results for LPR on uniform data to non-uniformly sampled data also. The denoising algorithms are compared with other standard, performant methods available in the literature both in terms of estimation error and computational complexity. A major perspective provided in this thesis is that the problem of optimum parameter choice in nonparametric regression can be viewed as the selection of optimum parameters of a linear, shift-invariant filter. This interpretation is provided by deriving motivation out of the hallmark paper of Savitzky and Golay and Schafer’s recent article in IEEE Signal Processing Magazine. It is worth noting that Savitzky and Golay had shown in their original Analytical Chemistry journal article, that LS fitting of a fixed-order polynomial over a neighborhood of fixed size is equivalent to convolution with an impulse response that is fixed and can be pre-computed. They had provided tables of impulse response coefficients for computing the smoothed function and smoothed derivatives for different orders and neighborhood sizes, the resulting filters being referred to as Savitzky-Golay (S-G) filters. Thus, we provide the new perspective that the regression parameter choice is equivalent to optimizing for the filter impulse response length/3dB bandwidth, which are inversely related. We observe that the MMSE solution is such that the S-G filter chosen is of longer impulse response length (equivalently smaller cutoff frequency) at relatively flat portions of the noisy signal so as to smooth noise, and vice versa at locally fast-varying portions of the signal so as to capture the signal patterns. Also, we provide a generalized S-G filtering viewpoint in the case of kernel regression. Building on the S-G filtering perspective, we turn to the problem of dynamic feature computation in speech recognition. We observe that the methodology employed for computing dynamic features from the trajectories of static features is in fact derivative S-G filtering. With this perspective, we note that the filter coefficients can be pre-computed, and that the whole problem of delta feature computation becomes efficient. Indeed, we observe an advantage by a factor of 104 on making use of S-G filtering over actual LS polynomial fitting and evaluation. Thereafter, we study the properties of first-and second-order derivative S-G filters of certain orders and lengths experimentally. The derivative filters are bandpass due to the combined effects of LPR and derivative computation, which are lowpass and highpass operations, respectively. The first-and second-order S-G derivative filters are also observed to exhibit an approximately constant-Q property. We perform a TIMIT phoneme recognition experiment comparing the recognition accuracies obtained using S-G filters and the conventional approach followed in HTK, where Furui’s regression formula is made use of. The recognition accuracies for both cases are almost identical, with S-G filters of certain bandwidths and orders registering a marginal improvement. The accuracies are also observed to improve with longer filter lengths, for a particular order. In terms of computation latency, we note that S-G filtering achieves delta and delta-delta feature computation in parallel by linear filtering, whereas they need to be obtained sequentially in case of the standard regression formulas used in the literature. Finally, we turn to the problem of speech enhancement where we are interested in de-noising using perceptually-motivated loss functions such as Itakura-Saito (IS). We propose to perform enhancement in the discrete cosine transform domain using risk-minimization. The cost functions considered are non-quadratic, and derivation of the unbiased estimator of the risk corresponding to the IS distortion is achieved using an approximate Taylor-series analysis under high signal-to-noise ratio assumption. The exposition is general since we focus on an additive noise model with the noise density assumed to fall within the exponential class of density functions, which comprises most of the common densities. The denoising function is assumed to be pointwise linear (modified James-Stein (MJS) estimator), and parallels between Wiener filtering and the optimum MJS estimator are discussed.

Fundamentos: A caracterização precisa da interação da placa aterosclerótica no momento do implante do stent é crucial para o entendimento da complacência e da cicatrização vasculares. Objetivamos investigar se a composição da placa avaliada pela tomografia de coerência óptica (OCT), influencia as alterações agudas no procedimento índice do implante do stent e na cicatrização vascular no seguimento tardio. Métodos: Os pacientes tratados com um único tipo de stent eluidor de fármaco (cromo cobalto, eluidor de sirolimus e polímero bioabsorvível) foram incluídos prospectivamente, seguindo um protocolo com etapas de dilatações progressivas do vaso. As imagens de OCT sequenciais foram realizadas no procedimento índice (basal e a cada etapa do protocolo) e no seguimento tardio, co-registradas e analisadas a cada 0,6mm. A avaliação semiquantitativa da placa foi realizada dividindo-se secções transversas em 4 quadrantes, com cada quadrante rotulado de acordo com o seu componente mais prevalente (fibrótico, calcificado, lipídico, normal). A interação stent-vaso avaliada pela OCT foi utilizada como indicador substituto para lesão e cicatrização vasculares após o implante do stent. Resultados: Um total de 22 lesões (1stent/lesão) de 20 pacientes e 2298 seções transversas de OCT foram analisadas no procedimento índice. O reestudo com OCT foi realizado em 17 pacientes e 19 lesões (86%). O componente de placa predominante foi fibrótico (fibrótico = 46.84 ± 16%; lipídico = 17.63 ± 10.72%; calcificado = 4.63 ± 5.9%; normal = 29.16 ± 12.24; não analizável=1.74 ± 5.35%). Houve um aumento nas áreas da luz (10atm = 5.5 (4.5 - 7.4) mm2, 14-16atm = 6.0 (4.7 - 7.70) mm2, 20atm = 6.7 (5.5 - 8.2) mm2; P < 0.001) e do stent (10atm = 5.2 (4.3 - 7.0) mm2, 14-16atm = 5.7 (4.5 - 7.5) mm2, 20atm = 6.5 (5.3 - 7.9) mm2; P < 0.001), com um aumento na área do prolapso tecidual (10atm =0.09 (0.06 - 0.12) mm2, 14-16atm =0.10 (0.06 - 0.15) mm2, 20atm =0.15 (0.08 - 0.20) mm2; P < 0.01). Segmentos com muito tecido fibrocalcificado tiveram áreas luminais menores ao longo das etapas da intervenção. Por outro lado, placas com muito conteúdo lipídico ou vaso normal tiveram maiores ganhos nas medidas das áreas luminais mínimas ao longo das dilatações sequenciais. Além disso, placas com muito tecido fibrocalcificado no momento basal apresentaram menor crescimento neointimal no seguimento tardio, enquanto que o grau de conteúdo lipídico e de vaso normal não tiveram impacto sobre a formação do tecido neointimal. Os indicadores substitutos de lesão vascular após o implante do stent correlacionaram-se significativamente com o crescimento neointimal no seguimento tardio. Conclusões: A composição tecidual das placas subjacentes influencia significativamente o comportamento mecânico agudo e a longo prazo dos vasos coronarianos submetidos ao implante de stent. Além disso, a lesão vascular após o implante do stent está potencialmente ligada ao futuro crescimento neointimal no seguimento tardio / Background Accurate characterization of atherosclerotic plaque interaction during stent deployment is crucial to understand vascular compliance and healing. We sought to determine whether plaque composition assessed by optical coherence tomography (OCT), influences acute changes at index procedure and vascular healing at follow up. Methods Patients treated with a single drug-eluting stent type (cobalt chromium with bioabsorbable polymer eluting sirolimus stent) were prospectively included, following a pre-defined step-by-step progressive vessel dilatation. Sequential OCT imaging were performed at the index procedure (baseline and at each time point of the protocol) and at follow up, co-registered and analyzed every 0.6mm for quantitative measurements. Semi-quantitative plaque assessment was performed at baseline by dividing cross-sections into 4 quadrants, with each quadrant labeled according to its most prevalent component (fibrotic, calcific, lipid). OCT assessments of stent-vessel interactions were used as a surrogate for vessel injury and healing after stent implantation. Results A total of 22 lesions (1stent/lesion) of 20 patients and 2298 OCT crosssections were analyzed at the index procedure. For an average of 19.7 months (591.88 ± 60.52 days), 17 of the patients and 19 lesions (86%) underwent OCT imaging at follow up. The predominant percentage plaque component was fibrotic (fibrotic = 46.84 ± 16%; lipid = 17.63 ± 10.72%; calcific = 4.63 ± 5.9%; normal = 29.16 ± 12.24; non-analyzable = 1.74 ± 5.35%). There was an increase in lumen (10atm = 5.5 (4.5 - 7.4) mm2, 14-16atm = 6.0 (4.7 - 7.70) mm2, 20atm = 6.7 (5.5 - 8.2) mm2; P < 0.001) and stent (10atm = 5.2 (4.3 - 7.0) mm2, 14-16atm = 5.7 (4.5 - 7.5) mm2, 20atm = 6.5 (5.3 - 7.9) mm2; P < 0.001) areas, with an increase in tissue prolapse area (10atm =0.09 (0.06 - 0.12) mm2, 14-16atm =0.10 (0.06 - 0.15) mm2, 20atm =0.15 (0.08 - 0.20) mm2; P < 0.01). Segments with high fibrocalcific content tended to have decreased minimal luminal areas along the intervention time-points. Conversely, plaques with high lipid content had increased minimal luminal areas during sequential dilatations. Moreover, plaques with high fibrocalcific tissue at baseline had significantly smaller neointimal growth at follow-up, whereas the degree of lipid content or normal tri-layered vessel had no impact on neointimal formation. OCT surrogates of vessel injury after coronary stenting significantly correlated with neointimal growth at follow-up. Conclusions: Tissue composition of underlying plaques significantly influences the acute mechanical and the long-term behavior of coronary vessels undergoing stent implantation. In addition, vessel injury after coronary stenting is potentially linked to future neointimal growth at follow-up

Cette thèse a pour objectif l’étude d’un lien effectif potentiel entre la motilité cellulaire, la mécanique cellulaire, et l’activité biochimique de ces mêmes cellules. Ce couplage a été étudié dans divers systèmes biologiques, et aussi bien dans des cultures de cellules qu’à l’intérieur de tissus plus complexes. Notamment, nous avons particulièrement cherché à détecter un couplage électromécanique dans des neurones qui pourrait être impliqué dans la propagation du message nerveux.Pour ce faire, nous avons dû développer deux microscopes optiques à la sensibilité extrême. Ces microscopes se composent de deux parties principales. La première sert à détecter des mouvements axiaux plus petits que la longueur d’onde optique, soit en dessous de 100 nanomètres. La deuxième partie permet la détection d’un signal de fluorescence, offrant la possibilité de suivre l’évolution biochimique de la cellule. Avec ces deux microscopes multimodaux, il est donc possible de suivre de manière simultanée un contraste de motilité, un contraste mécanique, un contraste structurel et un contraste biochimique. Si l’un de ces systèmes est basé sur la tomographie de cohérence optique plein champ et permet de faire de telles mesures en 3-D et en profondeur dans les tissus biologiques, le second ne permet que des mesures dans des cultures de cellules, mais est bien plus robuste au bruit mécanique. Dans ce manuscrit, nous allons essentiellement décrire le développement de ces deux appareils, et préciser les contrastes auxquels ils sont sensibles spécifiquement.Nous développerons également deux des applications principales de ces microscopes que nous avons étudié dans le détail au cours de cette thèse. La première application développe l’intérêt d’un de nos microscopes pour la détection sans marquage des principaux composants cellulaires et structuraux de la cornée et de la rétine. La seconde application tend à détecter et à suivre des ondes électromécaniques dans des neurones de mammifères / This PhD project aims to explore the relationship that might exist between the dynamic motility and mechanical behavior of different biological systems and their biochemical activity. In particular,we were interested in detecting the electromechanical coupling that may happen in active neurons, and may assist in the propagation of the action potential. With this goal in mind, we have developed two highly sensitive optical microscopes that combine one modality that detects sub-wavelength axial displacements using optical phase imaging and another modality that uses a fluorescence path. Therefore, these multimodal microscopes can combine a motility, a mechanical,a structural and a biochemical contrast at the same time. One of this system is based ona multimodal combination of full-field optical coherence tomography (FF-OCT) and allows the observation of such contrast inside thick and scattering biological tissues. The other setup provides a higher displacement sensitivity, but is limited to measurements in cell cultures. In this manuscript, we mainly discuss the development of both systems and describe the various contrastst hey can reveal. Finally, we have largely used our systems to investigate diverse functions of the eye and to look for electromechanical waves in cell cultures. The thorough description of both biological applications is also provided in the manuscript

Purpose: The aim of this study was to investigate differences in the nerve fiber layer and glaucoma-induced structural optic nerve head (ONH) damage in patients with normal- (NTG) and high-tension (HTG) glaucoma. Methods: In this retrospective pair-matched comparative study, 22 NTG and 22 HTG eyes were matched according to the same glaucomatous damage based on rim volume, rim area and disk size, as measured by Heidelberg retinal tomography (HRT III). Visual fields (VF) were assessed by Humphrey perimetry, and nerve fiber layer thickness was determined both by scanning laser polarimetry (GDxVCC) and spectral-domain optical coherence tomography (SD-OCT). Comparisons of all measured parameters were made between NTG and HTG groups. Results: Based on HRT results, both NTG and HTG eyes displayed comparable structural damage to the ONH (NTG/HTG, mean: disk area, 2.30/2.31 mm 2 , p = 0.942; rim area, 1.02/0.86 mm 2 , p = 0.082; rim volume, 0.19/0.17 mm 3 , p = 0.398). NTG eyes had significantly less VF damage than HTG eyes (NTG/HTG, mean deviation: –4.23/–12.12 dB, p = 0.002; pattern standard deviation: 5.39/8.23 dB, p = 0.022). The inferior nerve fiber layer of NTG patients was significantly thicker than that of HTG patients (NTG/HTG, mean: GDx inferior: 53.5/46.3 μm, p = 0.046). SD-OCT revealed a significantly thicker nerve fiber in NTG compared with HTG patients in all quadrants (NTG/HTG, total mean: 72.72/58.45 μm, p = 0.002). Conclusion: At comparable glaucomatous stages, nerve fiber loss was more advanced in HTG patients compared with NTG patients.

As dental caries is one of the most common diseases, the early and noninvasive detection of carious lesions plays an important role in public health care. Optical coherence tomography (OCT) with its ability of depth-resolved, high-resolution, noninvasive, fast imaging has been previously recognized as a promising tool in dentistry. Additionally, polarization sensitive imaging provides quantitative measures on the birefringent tissue properties and can be utilized for imaging dental tissue, especially enamel and dentin. By imaging three exemplary tooth samples ex vivo with proximal white spot, brown spot, and cavity, we show that the combination of polarization sensitive OCT and the degree of polarization uniformity (DOPU) algorithm is a promising approach for the detection of proximal carious lesions due to the depolarization contrast of demineralized tissue. Furthermore, we investigate different sizes of the DOPU evaluation kernel on the resulting contrast and conclude a suitable value for this application. We propose that DOPU provides an easy to interpret image representation and appropriate contrast for possible future screening applications in early caries diagnostics.

The common way to diagnose hard and soft tissue irregularities in the oral cavity is initially the visual inspection by an experienced dentist followed by further medical examinations, such as radiological imaging and/or histopathological investigation. For the diagnosis of oral hard and soft tissues, the detection of early transformations is mostly hampered by poor visual access, low specificity of the diagnosis techniques, and/or limited feasibility of frequent screenings. Therefore, optical noninvasive diagnosis of oral tissue is promising to improve the accuracy of oral screening. Considering this demand, a rigid handheld endoscopic scanner was developed for optical coherence tomography (OCT). The novelty is the usage of a commercially near-infrared endoscope with fitting optics in combination with an established spectral-domain OCT system of our workgroup. By reaching a high spatial resolution, in vivo images of anterior and especially posterior dental and mucosal tissues were obtained from the oral cavity of two volunteers. The convincing image quality of the endoscopic OCT device is particularly obvious for the imaging of different regions of the human soft palate with highly scattering fibrous layer and capillary network within the lamina propria.

This work deals with possibilities of registration of retinal images from different mo-dalities, concretely optical coherence tomography (OCT), scanning laser ophthalmoscopy (SLO) and fundus camera. In first stage is the interest focused on registration of SLO and fundus images, which will serve to determine area of interest for consecutive registration of OCT data. The final stage is finding correct location of OCT B-scans in fundus image. On the basis of the studied methods of registration was chosen method making use of computation of correlation coefficient for both cases. For finding optimal parameters of registration is used searching through whole space of parameters. In partial stages of the work was created algorithm for alignment of B-scans followed by detection of blood vessels and also simple algorithm for detection of blood vessels from fundus image. For more transparent registration the graphical user interface was created, which allows loading input images and displaying the result in several possible forms.

The master thesis is focused on creating a methodology for quantification of the nerve fiber layer on photographs of the retina. The introductory part of the text presents a medical motivation of the thesis and mentions several studies dealing with this issue. Furthermore, the work describes available textural features and compares their ability to quantify the thickness of the nerve fiber layer. Based on the described knowledge, the methodology to make different regression models enabling prediction of the retinal nerve fiber layer thickness was developed. Then, the methodology was tested on the available image dataset. The results showed, that the outputs of regression models achieve a high correlation between the predicted output and the retinal nerve fiber layer thickness measured by optical coherence tomography. The conclusion discusses an usability of the applied solution.