Orientador: Vital Paulino Costa / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Ciências Médicas / Made available in DSpace on 2018-08-23T16:05:44Z (GMT). No. of bitstreams: 1 Silva_FabricioReis_M.pdf: 2953091 bytes, checksum: 5a1660d55c27fcad4aeb1be8cceeb33f (MD5) Previous issue date: 2013 / Resumo: Objetivo: Avaliar a sensibilidade e especificidade dos classificadores de aprendizagem de máquina no diagnóstico de glaucoma usando dados do Spectral Domain OCT (SD-OCT) e perimetria automatizada acromática (PAA). Desenho do estudo: Estudo transversal observacional. Método: Sessenta e dois pacientes com glaucoma e 48 indivíduos normais foram incluídos. Todos os pacientes foram submetidos a exame oftalmológico completo, PAA (24-2 SITA; Humphrey Field Analyzer II, Carl Zeiss Meditec, Inc., Dublin, CA) e exame de imagem da camada de fibras nervosas utilizando SD-OCT (Cirrus HD-OCT; Carl Zeiss Meditec Inc., Dublin, California). Curvas ROC (Receiver Operating Characteristic) foram obtidas para todos os parâmetros do SD-OCT e índices globais da perimetria acromática (MD, PSD, GHT). Subsequentemente, os seguintes classificadores de aprendizagem de máquina (CAMs) foram testados usando parâmetros do OCT e PAA: Bagging (BAG), Naive-Bayes (NB), Multilayer Perceptron (MLP), Radial Basis Function (RBF), Random Forest (RAN), Ensemble Selection (ENS), Classification Tree (CTREE), Ada Boost M1(ADA), Support Vector Machine Linear (SVML) e Support Vector Machine Gaussian (SVMG). Áreas abaixo da curva ROC (aROC) obtidas com os parâmetros isolados do campo visual (CV) e OCT foram comparados com os CAMs usando dados combinados do OCT e CV. Resultados: A média de idade foi de 57,0 ± 9,2 anos para os indivíduos normais e 59,9 ± 9,0 anos para os pacientes glaucomatosos (p=0,103). As médias dos valores do Mean Deviation (MD) foram -4,1 ± 2,4 dB para glaucomatosos e -1,5 ± 1,6 dB para indivíduos normais (p<0,001). Os parâmetros do SD-OCT com maiores aROCs foram: quadrante inferior (aROC=0,813) e espessura média (aROC=0,807). O melhor parâmetro do CV foi PSD (aROC=0,915). Combinando os dados do OCT e do CV, aROCs dos CAMs variaram entre 0,777 (CTREE) e 0,946 (RAN). A maior aROC dos CAMs OCT+CV, obtida com RAN (0,946) foi significativamente maior que o melhor parâmetro do OCT (p=0,003), mas não houve diferença estatisticamente significante com o melhor parâmetro do CV (p=0,37). Conclusão: Os classificadores de aprendizagem de máquina treinados com dados do OCT e CV podem discriminar entre olhos normais e glaucomatosos com sucesso. A combinação das medidas do OCT e CV melhoraram a acurácia diagnóstica comparados aos parâmetros do OCT / Abstract: Purpose: To evaluate the sensitivity and specificity of machine learning classifiers (MLCs) for glaucoma diagnosis using Spectral Domain OCT (SD-OCT) and standard automated perimetry (SAP). Design: Observational cross-sectional study. Population: Sixty two glaucoma patients and 48 healthy individuals were included. Methods: All patients underwent a complete ophthalmologic examination, SAP and retinal nerve fiber layer (RNFL) imaging with SD-OCT (Cirrus HD-OCT; Carl Zeiss Meditec Inc., Dublin, California). Receiver operating characteristic (ROC) curves were obtained for all SD-OCT parameters and some SAP parameters (MD, PSD, GHT). Subsequently, the following MLCs were tested using parameters from SD-OCT and SAP: Bagging (BAG), Naive-Bayes (NB), Multilayer Perceptron (MLP), Radial Basis Function (RBF), Random Forest (RAN), Ensemble Selection (ENS), Classification Tree (CTREE), Ada Boost M1 (ADA), Support Vector Machine Linear (SVML) and Support Vector Machine Gaussian (SVMG). Areas under the ROC (aROC) obtained with SAP and OCT parameters were compared with those obtained with MLCs using combined OCT+SAP data. Results: The mean age was 57.0 ± 9.2 years for healthy individuals and 59.9 ± 9.0 years for glaucoma patients (p=0.103). Mean Deviation (MD) values were -4.1 ± 2.4 dB for glaucoma patients and -1.5 ± 1.6 dB for healthy individuals (p<0.001). SD-OCT parameters with the greater aROCs were inferior quadrant (0.813) and average thickness (0.807). The best SAP parameter was PSD (aROC=0.915). Combining OCT and SAP data, MLCs'aROCs varied from 0.777(CTREE) to 0.946 (RAN). The best combined OCT+SAP aROC, obtained with RAN (0.946), was significantly larger than the best single OCT parameter (p=0.003), but was not significantly different from the aROC obtained with the best single SAP parameter (p=0.37). Conclusion: Machine learning classifiers trained with OCT and SAP data can successfully discriminate between healthy and glaucomatous eyes. The combination of OCT and SAP measurements improved the diagnostic accuracy compared with OCT data alone / Mestrado / Oftalmologia / Mestre em Ciências Médicas

Objetivo: A partir das medidas anatômicas isoladas (altura, diâmetro externo e interno) do BMI construir um novo índice prognóstico para a cirurgia de correção do Buraco Macular Idiopático (IPBM). Tipo de estudo: intervencional, série de casos. Pacientes e Métodos: Estudo realizado no Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo e no Instituto Suel Abujamra, São Paulo-SP, entre outubro de 2005 e outubro de 2007. Foram incluídos 36 olhos de 36 pacientes com BMI, que foram avaliados segundo as medidas apresentadas ao exame de TCO (Stratus - Zeiss, versão 4.01) antes da cirurgia do BMI. Utilizando o compasso do TCO, obteve-se a medida dos maiores diâmetros externo e interno e da altura dos BMI. Por meio dessas medidas, foi criado o IPBM. Após vitrectomia posterior com retirada de Membrana Limitante Interna (MLI), sem utilização de corantes, os pacientes foram acompanhados por seis meses. Após a cirurgia, os pacientes foram avaliados no primeiro e sétimo dias, duas semanas, um, três e seis meses. Ao final do seguimento, o IBPM e outras variáveis (sexo, idade, raça, estádio do BMI pela classificação biomicroscópica de Gass, tempo decorrido desde a piora da acuidade visual informada pelo paciente e a acuidade visual pré-operatória), foram correlacionadas com o resultado anatômico e a acuidade visual pós-operatória. Resultados: Vinte e nove (80,6%) dos 36 olhos com BMI obtiveram fechamento anatômico ao final de seis meses de acompanhamento (8,86 ± 4,23 meses). Dezenove (52,7%) dos BMI eram do estádio IV de Gass, com tempo de duração maior que um ano em 21 pacientes (58,3%). A AV LogMAR corrigida pré-operatória média foi de 1,10 (0,60 a 1,62) e a pós-operatória média foi de 0,69 (0,0 a 1,60). A média de melhora da AV foi de 3,94 linhas. Em relação ao fechamento anatômico, não houve significância em relação ao tempo de história da doença entre os grupos aberto (grupo 1) e fechado (grupo 2) (Teste t-Student, p=0,072). O diâmetro da base interna foi maior no grupo 1 em relação ao grupo 2 (Teste t-Student, p=0,007). Na análise do índice IPBM, houve diferença significativa entre o grupo 1 (média 0,49) e o grupo 2 (média 0,91). (Teste t-Student, p< 0,001) A análise de regressão logística apontou que BMIs com IPBM maior que 0,53 apresentam chance de fechamento anatômico 9,6 vezes maior (Odds Ratio= 9,6, p = 0,018). Pacientes com IPBM > 0,53 apresentaram AV pós-operatória ao final do sexto mês significativamente melhor do que pacientes com IPBM < 0,53 (Mann-Whitney, p=0,005). O ganho percentual da AV foi de 41,93% nos pacientes com IBPM>0,53, quando comparado com os 7,14% do grupo com IPBM <0,53 (p=0,002). No que diz respeito à AV final LogMAR, todas as variáveis estudadas anteriormente foram submetidas ao teste de correlação de Pearson. Observou-se que o IPBM tem uma correlação negativa significante com a AV, e foi selecionado juntamente com a AV pré-operatória através de regressão linear como os melhores preditores de AV final (p<0,001 e p=0,005, respectivamente). O modelo aponta que 58,4% da AV pós-operatória está sendo explicada pelo IPBM e AV pré-operatória. Conclusões: Foi construído um novo índice Prognóstico do Buraco Macular Idiopático (IPBM) representado pela razão altura / diâmetro interno do BMI. Verificou-se que o IPBM pode ser utilizado como fator prognóstico de fechamento anatômico do BMI. O IPBM e a AV pré-operatória foram os fatores prognósticos com melhor relação para a AV no sexto mês após o tratamento cirúrgico do BMI. / Purpose: To create a new prognostic index for IMH surgery based on anatomical values of IMH height, external and internal diameters (MHPI). Type of Study: Prospective, interventional, case of series. Patients and Methods: 36 eyes with IMH of 36 patients followed at Hospital das Clinicas, University of São Paulo Medical School (HC-FMUSP) and Suel Abujamra Institute (ISA), São Paulo-SP, between October 2006 and October 2007, were included. IMH OCT measurements were obtained before surgery (Stratus - Zeiss version 4.01) Values of the larger external and internal diameters, and the IMH height were obtained using the OCT compass. The prognostic index of IMH (MHPI) was defined as the index height / internal base. MHPI was defined by using those OCT measurements. Patients underwent pars plana vitrectomy with ILM peeling without dye and were followed by 6 months. Patients were seen at days 1, 7, 14, and months 1, 2, 3, and 6 after surgery. At the end of the follow-up period, MHPI and, other variables (sex, age, ethnic group, stage of IMH following the biomicroscopic classification of Gass, the time of visual loss reported by the patient, and pre surgical visual acuity) were correlated with anatomical results and post-surgical visual acuity. Results: Twenty nine eyes (80.6%) of thirty six patients with IMH had anatomical closure at the end of the six-month follow-up (8.86 ± 4.23 months). Nineteen (52.7%) IMH were stage IV of Gass with more than one year duration in twenty one patients. Pre-surgical medium LogMAR VA was 1.10 (0.60 to 1.62) and post-surgical was 0.69 (0.0 to 1.60). Medium VA improvement was 3.94 lines. The internal base diameter (BINT) was larger in group 1 than in group 2 (t-Student Test, p=0.373). MHPI analisys showed significant difference between group 1 (average 0.49) and group 2 (average 0.91) (t-Student Test, p> 0.001). Logistical regression showed that IMH with MHPI higher than 0.53 present 9.6 times more risk of failure than those with MHPI lower than 0.53 (Mann-Whitney, p=0.005). The percentage gain of VA was 41.93% in patients with MHPI > 0.53, and 7.14% in patients with MHPI lower than 0.53 (p=0.002). In regards to the final LogMAR VA, all studied variables above submitted to Pearson correlation test. MHPI is inversely correlated with VA by linear regression with gradient procedure as best predictor of final VA (p< 0.001 and p= 0.005 respectively). The sample shows that 58.4% of post-surgery VA is being explained by the MHPI and pre-surgery VA. Conclusions: A new prognostic index for IMH surgery was defined as IMH height/internal diameter. We concluded that MHPI may be used as a prognostic factor for IMH anatomical closure after surgical treatment. MHPI and preoperatory VA were the best correlated prognostic factors for 6-month VA.

No description available.

Abstract The development of a non-invasive glucose monitoring technique is very important because it would tremendously diminish the need to puncture the skin when taking blood samples and help diabetic patients in controlling their blood glucose levels and in treating Diabetes Mellitus. The focus of this thesis is on measuring the effect of glucose on the light scattering properties of a tissue-simulating phantom and biological tissues in vitro. Optical coherence tomography (OCT), the pulsed photoacoustic (PA) technique, and the time-of-flight (TOF) technique are used in the measurements and their capabilities for detecting changes in the scattering properties are evaluated and compared with each other. The theoretical background of the techniques, light propagation and PA wave generation are briefly explained. The glucose-induced changes in light scattering are also reviewed. The measurement results with the OCT and the PA technique from Intralipid, pig whole blood, and mouse skin tissue samples show that the glucose-induced changes are larger in the biological tissues than in the Intralipid phantom. The PA measurements show that although the PA signals are stronger at a wavelength of 532 nm than at 1064 nm, the glucose-induced change in the peak-to-peak value of the PA signal measured from pig whole blood is larger at a wavelength of 1064 nm than at 532 nm. The TOF measurements with a streak camera show that the scattering-related changes in the registered pulse shapes occur mainly in the rising part of the pulses. The utilization of fiber-optic measurement heads enabled the detection of back-scattered photons at different distances from the emitting fiber. Although all the techniques are able to detect changes induced by large glucose concentrations (0–5000 mg/dl) in Intralipid, the effect of glucose on the scattering properties of Intralipid is so weak that the techniques failed to detect changes with lower (50–500 mg/dl) concentrations. The measurements of biological samples with the PA technique and with the OCT also demonstrate capabilities to measure glucose concentrations in the physiologically relevant range (18–450 mg/dl) as well. The results compare well with earlier literature and also confirm some earlier findings.

Abstract Optical coherence tomography (OCT) is a modern rapidly developing technique for non-invasive imaging of the internal structure of optically non-uniform objects based on the principles of low-coherent interferometry. However, multiple scattering of light in the objects under study brings distortions to the images obtained by OCT. The analysis of formation of the OCT signals is required for understanding the role of multiple scattering in this formation and providing recommendations for optimal configuration of a measuring setup. In the present thesis formation of the OCT signals and images is analyzed implementing Monte Carlo simulations of light propagation in scattering media. Blood, intralipid solution, human skin and paper samples are chosen as the objects under study due to the interest in the diagnostics of these objects in biomedicine and paper industry. Multilayer models of skin phantoms, skin and paper were developed in the frames of the present study for simulation of OCT signals and two-dimensional OCT images of these objects. The contribution of different scattering orders as well as different fractions of photons (least and multiply scattered, diffusive and non-diffusive) to these images was found allowing to evaluate the maximal depth of non-distorted imaging in each particular case. The simulated OCT images were compared to the experimental ones demonstrating qualitative similarity. This fact allowed the author to analyze qualitatively the influence of parameters of the OCT setup on the images which have also been acquired in this work. The formation of the OCT images of paper samples with various refractive index matching liquids was also studied.

The utilisation of near-infraredR aman spectroscopyfo r the discrimination of cancersa nd pre-cancers from normal tissue in the acro-digestive tract has been evaluated. A commercially available Raman microspectrometehr as been modified to provide optimum throughput, sensitivity and fluorescence suppression for epithelial tissue measurements. Laser excitation at 830nmw as demonstratedto be optimum. High quality (SN ratio 15-20) NIR-Raman spectrah ave been acquired from oesophageaal nd laryngeal tissues in time scales under 30 seconds. Pathological groupings covering the full range of normal and neoplastic tissues in the organs of interest have been studied. Both fresh (snap frozen) and formalin fixed tissue samples were investigated,f irstly to indicate whether tissue-typesc an be distinguishedi n vivo and secondlyt o demonstrateth e use of Raman spectroscopya s a tool for classificationi n the pathology lab. Results using multivariate statistical techniques to distinguish between spectra from specimens exhibiting different tissue pathologies have been extremely promising. Crossvalidation of the spectral predictive models has shown that three groups of larynx tissue can be separated with sensitivities and specificities of between 86 and 90% and 87 and 95% respectively. Oesophageal prediction models have demonstrated sensitivities and specificities of 84 to 97% and 93 to 98% respectively for a three-group consensus model and 73 to 100% and 92 to 100% for an eight-groupc onsensusm odel. Epithelial tissues including stomach, tonsil, endometrium, bladder and prostate have been studiedt o identify further tissuesw hereR amans pectroscopym ay be employedf or detection of disease.S pectraw ere similar to those obtainedf rom oesophagusa nd larynx, although sufficiently different for distinct discriminant models to be required. This work has demonstratedth e genericn atureo f Ramans pectroscopyfo r the detectiona nd classification of cancersa nd pre-cancerousle sionsi n many tissues.T he evidencep rovided by this study indicatest hat utilisation of Ramans pectroscopyfo r non-invasived etectiona nd classification of diseaseis a distinct possibility. Potentiald ifficulties in the transferabilityf rom in vitro to in vivo have been evaluated and no significant barriers have been observed. However, further in vivo probe development and optimisation will be required before 'optical biopsy' with Ramans pectroscopyc anb ecomea reality.

Optical Coherence Tomography Velocimetry (OCTV) is a technique based on principals developed for medical imaging and is used to measure the velocity as a function of sample depth for a range of complex fluid systems, revealing insight into rheological properties that are otherwise ignored by bulk rheometry methods. The technique fulfils the requirements needed to study a wide range of complex systems, being capable of measuring absolute velocity and velocity fluctuations with high spatial and temporal resolution, as well as in opaque materials. Improvements to the instrumentation behind OCTV are described in detail, with improvements to signal to noise being achieved through better detection and the use of modulation techniques. The technique is also demonstrated in a range of complex fluids including dense suspensions of hard spheres, DNA solutions, polyacrylamide solutions, bacteria and shampoo amongst others. Non-linear flow behaviour is revealed that includes shear banding, wall slip and elastic turbulence. X-ray scattering techniques are also discussed, with results from recent experiments at the Diamond synchrotron being presented. The commercial potential of OCTV is discussed in depth, with analysis of the market of scientific instruments for measuring flow, competitor analysis and a review of relevant intellectual property. Methods to commercialise OCTV are discussed, with licensing to an existing competitor being identified as the means with the least risk and greatest chance of success.

Dental erosion has been defined as a chemical process that involves the dissolution of enamel and dentine by acid(s) not derived from bacteria when the surrounding aqueous phase is under-saturated with tooth mineral. Clinically however, dental erosion hardly ever occurs exclusively from the other modes of tooth wear such as abrasion and attrition. Hence the term erosive wear has been proposed and used to describe erosion-facilitated wear.With the prevalence of erosive wear being reported to be on the rise among children and adolescents in many countries, a plethora of oral health products such as dentifrices and mouth rinses had been put forward with claims of increasing the resistance of the enamel and / or dentine from being demineralised. However, the efficacy of these oral health products is still widely questionable as the studies carried out were mainly in vitro and / or in situ studies and the contradictory results were reported by different groups of researchers.The above mentioned discrepancies are often due to non-standardised in vitro and in situ study designs of erosion studies. Parameters such as type of acid used, concentration of acid used, duration of acid exposure, inclusion or exclusion of abrasion in the study model, ex vivo or in vivo acid challenge, ex vivo or in vivo treatment with oral health product and the nature of pellicle on the specimen are often varied.The availability of an detection tool which could be used to measure demineralisation on natural tooth surface in vivo would exclude many of the inevitable variability in in situ and in vitro study design, such as the simulation of the intra oral conditions in which the erosion challenge and intervention take place and the nature of the pellicle formed on the specimens. Hence with an in vivo detection tool, it is hopeful that the efficacy of any interventions would be evaluated more accurately and its results expounded to a wider context. It would be useful if the tool was sensitive to the very early stages of the erosion process as this would entail shorter and more economical study designs. Arising from the problems discussed above, potential non-invasive assessment methods that could be used clinically to measure demineralisation were explored and it was found that two optical methods, Optical Coherence Tomography (OCT) and Quantitative Light-induced Fluorescence (QLF) were potential methods for the tasks.The studies described in this thesis were divided into three main big sections; the first being the in vitro validation work of the two optical methods (Chapter 5 and 6), the second, validation of these methods in situ (Chapter 7) and finally the assessment of the efficacy of a high fluoride dentifrice on early enamel erosion using these two methods (Chapter 8).In the in vitro validation (Chapter 5), it was demonstrated that both QLF and OCT were able to detect erosion-interval related changes in natural surface samples eroded with orange juice for 60 minutes. However, results of Multiple Linear Regression and Paired t test suggest that QLF was more sensitive than OCT in the detection of demineralisation changes in this particular experimental setting. QLF demonstrated a R2 value of 0.590 while the best of the OCT outcome measure demonstrated a R2 value of 0.319.Further in vitro study (Chapter 6) was performed to explore the use of a moistened-exposed surface as a reference method for both OCT and QLF as it was felt that it would be beneficial if a coated reference area which posts as an additional step for error could be done away with. Results of this study showed that the moistened-exposed surface could be used a reference method for QLF only but not for OCT. Furnished with the findings of the above mentioned in vitro validation studies, an in situ validation of both the optical methods was performed (Chapter 7). It was found that OCT and QLF were able to longitudinally measure in situ demineralisation on polished and natural surfaced enamel which were subjected to 150 minutes of in vivo exposure to orange juice. Similar to the results of the in vitro study, QLF was found to be more sensitive than OCT in the detection of demineralisation changes on natural surfaced enamel. The last in situ study (Chapter 8) was to evaluate whether there was a protective effect of treating human enamel with a high-concentrated fluoride dentifrice during an active erosion phase and whether OCT and QLF were able to detect the protective effect. It was found that treating natural surface enamel with a 5000ppm NaF dentifrice increased its resistance against concurrent in vivo erosive challenge with an orange juice challenge. QLF was able to detect the protective effect of the 5000ppm NaF dentifrice on natural surface enamel against early in vivo erosion with an orange juice challenge regime while OCT did not.

Optical coherence tomography (OCT) and Photoacoustic microscopy (PAM) are two noninvasive, high-resolution, three-dimensional, biomedical imaging modalities based on different contrast mechanisms. OCT detects the light backscattered from a biological sample either in the time or spectral domain using an interferometer to form an image. PAM is sensitive to optical absorption by detecting the light-induced acoustic waves to form an image. Due to their complementary contrast mechanisms, OCT and PAM are suitable for being combined to achieve multimodal imaging. In this dissertation, an optical coherence photoacoustic microscopy (OC-PAM) system was developed for in vivo multimodal retinal imaging with a pulsed broadband NIR light source. To test the capabilities of the system on multimodal ophthalmic imaging, the retina of pigmented rats was imaged. The OCT images showed the retinal structures with quality similar to conventional OCT, while the PAM images revealed the distribution of melanin in the retina since the NIR PAM signals are generated mainly from melanin in the posterior segment of the eye. By using the pulsed broadband light source, the OCT image quality highly depends on the pulse-to-pulse stability of the light source without averaging. In addition, laser safety is always a concern for in vivo applications, especially for eye imaging with a pulsed light source. Therefore, a continuous wave (CW) light source is desired for OC-PAM applications. An OC-PAM system using an intensity-modulated CW superluminescent diode was then developed. The system was tested for multimodal imaging the vasculature of a mouse ear in vivo by using Gold Nanorods (GNRs) as contrast agent for PAM, as well as excised porcine eyes ex vivo. Since the quantitative information of the optical properties extracted from the proposed NIR OC-PAM system is potentially able to provide a unique technique to evaluate the existence of melanin and lipofuscin specifically, a phantom study has been conducted and the relationship between image intensity of OCT and PAM was interpreted to represent the relationship between the optical scattering property and optical absorption property. It will be strong evidence for practical application of the proposed NIR OC-PAM system.

A multilayered system is a good model for many optical systems. An optical coherence tomography (OCT) system can provide fundamental information about the refractive index distribution of the sample and enables images to be corrected for geometric accuracy, but this requires the separation of refractive index and physical thickness for each layer from the measured optical paths. In this thesis a novel approach for simultaneous extraction of index and thickness of multi layer systems and the last medium index of refraction in a single experiment using only the object’s spectral response available by any Fourier Domain OCT system without using any additional outside measurements is introduced. The method is based on a novel matrix equation that uses the reflected spectrum from the object and the measured optical thickness. In the presence of slight error of measuring optical thickness, the parameters extraction has wavenumber dependency. A novel method is used to select the suitable set of spectral components that reduces the extracted parameters error. The parameters extraction method is followed by a fitting process for optimized results. The method works the best for low contrast index distribution even in the presence of relatively large optical thickness measurement error.