Time-resolved fluorescence spectroscopy (TRFS) is a powerful analytical tool for quantifying the biochemical composition of organic and inorganic materials. The potentials of TRFS as nondestructive clinical tool for tissue diagnosis have been recently demonstrated. To facilitate the translation of TRFS technology to the clinical arena, algorithms for online TRFS data analysis are of great need. A fast model-free TRFS deconvolution algorithm based on the Laguerre expansion method has been previously introduced, demonstrating faster performance than standard multiexponential methods, and the ability to estimate complex fluorescence decay without any a-priori assumption of its functional form. One limitation of this method, however, was the need to select, a priori, the Laguerre parameter a and the expansion order, which are crucial for accurate estimation of the fluorescence decay. In this thesis, a new implementation of the Laguerre deconvolution method is introduced, in which a nonlinear least-square optimization of the Laguerre parameter is performed, and the selection of optimal expansion order is attained based on a Minimum Description Length (MDL) criterion. In addition, estimation of the zero-time delay between the recorded instrument response and fluorescence decay is also performed based on a normalized means square error criterion. The method was fully validated on fluorescence lifetime, endogenous tissue fluorophores, and human tissue. The automated Laguerre deconvolution method is expected to facilitate online applications of TRFS, such as clinical real-time tissue diagnosis.

Accurately assessing the extent of myocardial tissue injury induced by Myocardial infarction (MI) is critical to the planning and optimization of MI patient management. With this in mind, this study investigated the feasibility of using combined fluorescence and diffuse reflectance spectroscopy to characterize a myocardial infarct at the different stages of its development. An animal study was conducted using twenty male Sprague-Dawley rats with MI. In vivo fluorescence spectra at 337 nm excitation and diffuse reflectance between 400 nm and 900 nm were measured from the heart using a portable fiber-optic spectroscopic system. Spectral acquisition was performed on - (1) the normal heart region; (2) the region immediately surrounding the infarct; and (3) the infarcted region - one, two, three and four weeks into MI development. The spectral data were divided into six subgroups according to the histopathological features associated with various degrees / severities of myocardial tissue injury as well as various stages of myocardial tissue remodeling, post infarction. Various data processing and analysis techniques were employed to recognize the representative spectral features corresponding to various histopathological features associated with myocardial infarction. The identified spectral features were utilized in discriminant analysis to further evaluate their effectiveness in classifying tissue injuries induced by MI. In this study, it was observed that MI induced significant alterations (p < 0.05) in the diffuse reflectance spectra, especially between 450 nm and 600 nm, from myocardial tissue within the infarcted and surrounding regions. In addition, MI induced a significant elevation in fluorescence intensities at 400 and 460 nm from the myocardial tissue from the same regions. The extent of these spectral alterations was related to the duration of the infarction. Using the spectral features identified, an effective tissue injury classification algorithm was developed which produced a satisfactory overall classification result (87.8%). The findings of this research support the concept that optical spectroscopy represents a useful tool to non-invasively determine the in vivo pathophysiological features of a myocardial infarct and its surrounding tissue, thereby providing valuable real-time feedback to surgeons during various surgical interventions for MI.

Des diagnostics optiques non-intrusifs sont développés afin d’améliorer la compréhension des phénomènes d’évaporation de gouttes de combustible multicomposant. La configuration simplifiée retenue est un jet linéaire de gouttes monodispersées. L’évaporation est réalisée soit à la température ambiante, soit à des températures plus élevées de l’ordre de 500°C. Le combustible utilisé est composé d’éthanol et d’acétone, dont les volatilités sont très différentes. Deux nouvelles techniques de mesures fondées sur la Fluorescence Induite par Laser (LIF) ont été développées au cours des ces travaux. La première, la LIF à trois couleurs permet de mesurer la température moyenne de gouttes binaires. La deuxième exploite la fluorescence émise par l’acétone excitée par un rayonnement laser à 266 nm afin de déterminer la composition instantanée du mélange. Les résultats expérimentaux ont été comparés à un modèle à composants discrets utilisant une approche 1D quasi-stationnaire / Non intrusive optical diagnostics are developed to investigate the mechanisms governing multicomponent droplets evaporation. The case of periodically and linearly arranged monodisperse droplets is considered in these experiments. Droplets are injected either at ambient temperature or into a hot air plume at about 500°C. The fuel mixture is made of ethanol and acetone, which exhibit very different volatilities. Two techniques based on Laser Induced Fluorescence (LIF) have been developed to investigate heat and mass transfers. The first one, the three colours LIF allows obtaining the average temperature of binary droplets. The second one used the fluorescence of acetone excited by an UV laser (266 nm) in order to quantify the acetone molar fraction. Finally, measurements are compared with a numerical model that is derived from a discreet component approach and the assumption of 1D quasi-steady evolution of the gas phase

Melanoma is a pigmented tumor that originates from the melanocytes; pigmented cells present throughout the body, including skin and iris. The cutaneous form is the most common type, and it represents about 5% of the skin tumors diagnosed in Brazil. Although it does not have a high incidence, it represents about 80% to 85% of all skin tumor deaths. The second most frequent type of melanoma is ocular. It represents 5% of all melanoma cases and is a potentially lethal disease, especially when it causes metastasis. The main therapeutic approach for melanomas, in general, is surgery, with resection of the cutaneous lesion or enucleation in the case of ocular melanoma. Other techniques such as adjuvant immunotherapy, palliative chemotherapy, and radiotherapy are also used. However, they have low efficacy and several side effects. Photodynamic therapy is a therapeutic modality based on the interaction of light at specific wavelength and photosensitizer, in the presence of molecular oxygen, leading the cell to death. As melanoma is a pigmented cancer, it usually does not respond well to photodynamic therapy due to the high absorption of light on the surface of the tumor, making volumetric eradication impossible. This project investigated optical strategies for the diagnosis and treatment of melanoma. For the diagnosis, it was evaluated the fluorescence lifetime technique to differentiate melanoma and normal skin. A sensitivity of 99.4%, specificity of 97.4% and accuracy of 98.4% were achieved using linear discrimination analysis. For the cutaneous melanoma treatment, PDT combined to optical clearing agents (OCAs) was investigated. Vascular and cell-target photosensitizers were evaluated combined or not to OCAs. OCA improved PDT response in all pigmented tumors treated, but the best results were achieved when a dual-photosensitizer treatment combined to OCA was performed. The treatment of conjunctival melanoma was conducted using 2-photon excitation photodynamic therapy. The advantage of this technique is the use of infrared light, in a wavelength that melanin has a low absorption, improving the light penetration into the tumor. The tumor histology shows that apoptosis was induced only at the treatment site, with no damage to the surrounding tissue. Additionally, a single TPE-PDT session could treat the entire tumor. / O melanoma é um tumor pigmentado que surge dos melanócitos, células pigmentadas presentes em todo o corpo, incluindo a pele e a íris. A forma cutânea é a mais comum e representa cerca de 5% dos tumores cutâneos diagnosticados no Brasil. Embora não tenha uma alta incidência, representa cerca de 80% a 85% de todas as mortes por tumor de pele. O segundo tipo de melanoma mais frequente é o ocular. Representa 5% de todos os casos de melanoma e é uma doença potencialmente letal, especialmente em casos de metástase. A principal abordagem terapêutica para melanomas, em geral, é a cirurgia, com ressecção da lesão cutânea ou enucleação no caso do melanoma ocular. Outras técnicas, como imunoterapia adjuvante, quimioterapia paliativa e radioterapia também são usadas, porém, apresentam baixa eficiência e muitos efeitos colaterais. A terapia fotodinâmica é uma modalidade terapêutica baseada na interação da luz em um comprimento de onda específico e um fotossensibilizador, na presença de oxigênio molecular, levando a célula à morte. Como o melanoma é um câncer pigmentado, geralmente não responde bem à terapia fotodinâmica devido à alta absorção de luz na superfície do tumor, impossibilitando a erradicação volumétrica. Este projeto investigou estratégias ópticas para o diagnóstico e tratamento do melanoma. Para o diagnóstico, foi avaliada a técnica de tempo de vida de fluorescência para distinguir melanoma de pele normal. Utilizando análise de discriminação linear, obteve-se uma sensibilidade de 99,4%, especificidade de 97,4% e precisão de 98,4%. Para o tratamento de melanoma cutâneo, a PDT combinada com clareadores ópticos (OCAs) foi investigada. Um fotossensibilizador que tem como alvo vaso sanguíneo e um fotossensibilizador de alvo celular foram avaliados combinados ou não com OCAs. OCAs são soluções hiperosmóticas que desidratam o tecido, diminuindo o espalhamento da luz e melhorando a penetração de luz em profundidade. OCA melhorou a resposta de PDT em todos os tumores melanóticos tratados, mas os melhores resultados foram obtidos quando a PDT foi realizada com a combinação dos fotossensibilizadores e clareador óptico em uma única sessão. O tratamento do melanoma conjuntival foi realizado utilizando a terapia fotodinâmica por excitação de 2 fótons (TPE-PDT). A vantagem desta técnica é o uso de luz na região do infravermelho, em um comprimento de onda que melanina tem baixa absorção, melhorando a penetração de luz no tumor. A histologia do tumor mostrou que a apoptose foi induzida apenas no local do tratamento, sem danos no tecido adjacente. Além disso, uma única sessão de TPE-PDT foi capaz de tratar todo o tumor.

Melanoma is a pigmented tumor that originates from the melanocytes; pigmented cells present throughout the body, including skin and iris. The cutaneous form is the most common type, and it represents about 5% of the skin tumors diagnosed in Brazil. Although it does not have a high incidence, it represents about 80% to 85% of all skin tumor deaths. The second most frequent type of melanoma is ocular. It represents 5% of all melanoma cases and is a potentially lethal disease, especially when it causes metastasis. The main therapeutic approach for melanomas, in general, is surgery, with resection of the cutaneous lesion or enucleation in the case of ocular melanoma. Other techniques such as adjuvant immunotherapy, palliative chemotherapy, and radiotherapy are also used. However, they have low efficacy and several side effects. Photodynamic therapy is a therapeutic modality based on the interaction of light at specific wavelength and photosensitizer, in the presence of molecular oxygen, leading the cell to death. As melanoma is a pigmented cancer, it usually does not respond well to photodynamic therapy due to the high absorption of light on the surface of the tumor, making volumetric eradication impossible. This project investigated optical strategies for the diagnosis and treatment of melanoma. For the diagnosis, it was evaluated the fluorescence lifetime technique to differentiate melanoma and normal skin. A sensitivity of 99.4%, specificity of 97.4% and accuracy of 98.4% were achieved using linear discrimination analysis. For the cutaneous melanoma treatment, PDT combined to optical clearing agents (OCAs) was investigated. Vascular and cell-target photosensitizers were evaluated combined or not to OCAs. OCA improved PDT response in all pigmented tumors treated, but the best results were achieved when a dual-photosensitizer treatment combined to OCA was performed. The treatment of conjunctival melanoma was conducted using 2-photon excitation photodynamic therapy. The advantage of this technique is the use of infrared light, in a wavelength that melanin has a low absorption, improving the light penetration into the tumor. The tumor histology shows that apoptosis was induced only at the treatment site, with no damage to the surrounding tissue. Additionally, a single TPE-PDT session could treat the entire tumor. / O melanoma é um tumor pigmentado que surge dos melanócitos, células pigmentadas presentes em todo o corpo, incluindo a pele e a íris. A forma cutânea é a mais comum e representa cerca de 5% dos tumores cutâneos diagnosticados no Brasil. Embora não tenha uma alta incidência, representa cerca de 80% a 85% de todas as mortes por tumor de pele. O segundo tipo de melanoma mais frequente é o ocular. Representa 5% de todos os casos de melanoma e é uma doença potencialmente letal, especialmente em casos de metástase. A principal abordagem terapêutica para melanomas, em geral, é a cirurgia, com ressecção da lesão cutânea ou enucleação no caso do melanoma ocular. Outras técnicas, como imunoterapia adjuvante, quimioterapia paliativa e radioterapia também são usadas, porém, apresentam baixa eficiência e muitos efeitos colaterais. A terapia fotodinâmica é uma modalidade terapêutica baseada na interação da luz em um comprimento de onda específico e um fotossensibilizador, na presença de oxigênio molecular, levando a célula à morte. Como o melanoma é um câncer pigmentado, geralmente não responde bem à terapia fotodinâmica devido à alta absorção de luz na superfície do tumor, impossibilitando a erradicação volumétrica. Este projeto investigou estratégias ópticas para o diagnóstico e tratamento do melanoma. Para o diagnóstico, foi avaliada a técnica de tempo de vida de fluorescência para distinguir melanoma de pele normal. Utilizando análise de discriminação linear, obteve-se uma sensibilidade de 99,4%, especificidade de 97,4% e precisão de 98,4%. Para o tratamento de melanoma cutâneo, a PDT combinada com clareadores ópticos (OCAs) foi investigada. Um fotossensibilizador que tem como alvo vaso sanguíneo e um fotossensibilizador de alvo celular foram avaliados combinados ou não com OCAs. OCAs são soluções hiperosmóticas que desidratam o tecido, diminuindo o espalhamento da luz e melhorando a penetração de luz em profundidade. OCA melhorou a resposta de PDT em todos os tumores melanóticos tratados, mas os melhores resultados foram obtidos quando a PDT foi realizada com a combinação dos fotossensibilizadores e clareador óptico em uma única sessão. O tratamento do melanoma conjuntival foi realizado utilizando a terapia fotodinâmica por excitação de 2 fótons (TPE-PDT). A vantagem desta técnica é o uso de luz na região do infravermelho, em um comprimento de onda que melanina tem baixa absorção, melhorando a penetração de luz no tumor. A histologia do tumor mostrou que a apoptose foi induzida apenas no local do tratamento, sem danos no tecido adjacente. Além disso, uma única sessão de TPE-PDT foi capaz de tratar todo o tumor.

Introduction: Le cancer colorectal demeure un grave problème de santé publique au Canada. Les programmes de dépistage pourraient réduire l'incidence du cancer colorectal et la mortalité qui lui est associée. Une coloscopie de haute qualité est considérée comme un moyen rentable de prévenir le cancer en identifiant et en éliminant les lésions précurseurs du cancer. Bien que la coloscopie puisse servir de mesure préventive contre le cancer, la procédure peut imposer un fardeau supplémentaire à la santé publique par l'enlèvement et l'évaluation histologique de polypes colorectaux diminutifs et insignifiants, qui présentent un risque minime d'histologie avancée ou de cancer. La technologie de l'amélioration de l'image permettrait aux médecins de réséquer et de rejeter les polypes diminutifs ou de diagnostiquer et de laisser les polypes rectosigmoïdiens diminutifs sans examen histopathologique. Malgré la disponibilité de systèmes informatiques de caractérisation des polypes, la pratique du diagnostic optique reste limitée en raison de la crainte d'un mauvais diagnostic de cancer, d'une mauvaise surveillance des patients et des problèmes médico-légaux correspondants. Il est donc indispensable d'élaborer des stratégies alternatives de résection et d'élimination non optiques pour améliorer la précision et la sécurité du diagnostic optique et l'adapter à la pratique clinique. Ces stratégies doivent répondre à des critères cliniques simples et ne nécessitent pas de formation supplémentaire ni de dispositifs d'amélioration de l'image. De plus, la pratique sûre du diagnostic optique, la prise de décision appropriée concernant la technique de polypectomie ou l'intervalle de surveillance dépendent de l'estimation précise de la taille des polypes. La variabilité inter-endoscopistes dans la mesure de la taille des polypes exige le développement de méthodes fiables et validées pour augmenter la précision de la mesure de la taille. Une balance virtuelle intégrée à un endoscope haute définition est actuellement disponible pour le calcul automatique de la taille des polypes, mais sa faisabilité clinique n'a pas encore été établie. En dehors des points susmentionnés, une coloscopie de haute qualité nécessite l'examen complet de la muqueuse colique, ainsi que la visualisation de la valve iléocæcale et de l'orifice appendiculaire. À ce jour, aucune solution informatique n'a été capable d'assister les endoscopistes pendant les coloscopies en temps réel en détectant et en différenciant les points de repère cæcaux de façon automatique. Objectifs: Les objectifs de cette thèse sont : 1) d'étudier l'effet de la limitation du diagnostic optique aux polypes de 1 à 3 mm sur la sécurité du diagnostic optique pour le traitement des polypes diminutifs et l'acceptation par les endoscopistes de son utilisation dans les pratiques en temps réel tout en préservant ses potentiels de temps et de rentabilité ; 2) élaborer et examiner des stratégies non optiques de résection et d'élimination qui peuvent remplacer le diagnostic optique tout en offrant les mêmes possibilités d'économie de temps et d'argent ; 3) examiner la précision relative d'un endoscope à échelle virtuelle pour mesurer la taille des polypes ; 4) former, valider et tester un modèle d'intelligence artificielle qui peut prédire la complétude d'une procédure de coloscopie en identifiant les points de repère anatomiques du cæcum (c'est-à-dire la valve iléo-cæcale et l'orifice appendiculaire) et en les différenciant les uns des autres, des polypes et de la muqueuse normale. Méthodes: Pour atteindre le premier objectif de cette thèse, une analyse post-hoc de trois études prospectives a été réalisée pour évaluer la proportion de patients chez lesquels des adénomes avancés ont été découverts et le diagnostic optique a entraîné une surveillance retardée dans trois groupes de taille de polypes : 1–3, 1–5, et 1–10 mm. Pour atteindre le second objectif de cette thèse, deux stratégies non optiques ont été développées et testées dans deux études prospectives: une stratégie de résection et d'élimination basée sur la localisation qui utilise la localisation anatomique des polypes pour classer les polypes du côlon en non-néoplasiques ou néoplasiques à faible risque et une stratégie de résection et d'élimination basée sur les polypes qui attribue des intervalles de surveillance en fonction du nombre et de la taille des polypes. Dans les trois études, la concordance de l'attribution d'intervalles de surveillance basée sur un diagnostic optique à haute confiance ou sur des stratégies non optiques avec les recommandations basées sur la pathologie, ainsi que la proportion d'examens pathologiques évités et la proportion de communications immédiates d'intervalles de surveillance, ont été évaluées. Le troisième objectif de cette thèse a été abordé par le biais d'une étude de faisabilité pilote prospective qui a utilisé la mesure de spécimens de polypes immédiatement après leur prélèvement, suite à une polypectomie par un pied à coulisse Vernier comme référence pour comparer la précision relative des mesures de la taille des polypes entre les endoscopistes et un endoscope à échelle virtuelle. Enfin, le quatrième objectif de cette thèse a été évalué par l'enregistrement et l'annotation prospective de vidéos de coloscopie. Des images non modifiées de polype, de valve iléo-caecale, d'orifice appendiculaire et de muqueuse normale ont été extraites et utilisées pour développer et tester un modèle de réseau neuronal convolutionnel profond pour classer les images pour les points de repère qu'elles contiennent. Résultats: La réduction du seuil du diagnostic optique favoriserait la sécurité du diagnostic optique en diminuant de manière significative le risque d'écarter un polype avec une histologie avancée ou la mauvaise surveillance d'un patient avec de tels polypes. En outre, les stratégies non optiques de résection et d'élimination pourraient dépasser le critère de référence d'au moins 90% de concordance dans l'attribution des intervalles de surveillance post-polypectomie par rapport aux décisions basées sur l'évaluation pathologique. De plus, il a été démontré que l'endoscope à échelle virtuelle est plus précis que l'estimation visuelle de la taille des polypes en temps réel. Enfin, un modèle d'apprentissage profond s'est révélé très efficace pour détecter les repères cæcaux, les polypes et la muqueuse normale, à la fois individuellement et en combinaison. Discussion: La prédiction histologique optique des polypes de 1 à 3 mm est une approche efficace pour améliorer la sécurité et la faisabilité de la stratégie de résection et d'écartement dans la pratique. Les approches non optiques de résection et d'élimination offrent également des alternatives viables au diagnostic optique lorsque les endoscopistes ne sont pas en mesure de répondre aux conditions de mise en œuvre systématique du diagnostic optique, ou lorsque la technologie d'amélioration de l'image n'est pas accessible. Les stratégies de résection et de rejet, qu'elles soient optiques ou non, pourraient réduire les coûts supplémentaires liés aux examens histopathologiques et faciliter la communication du prochain intervalle de surveillance le même jour que la coloscopie de référence. Un endoscope virtuel à échelle réduite faciliterait l'utilisation du diagnostic optique pour la détection des polypes diminutifs et permet une prise de décision appropriée pendant et après la coloscopie. Enfin, le modèle d'apprentissage profond peut être utile pour promouvoir et contrôler la qualité des coloscopies par la prédiction d'une coloscopie complète. Cette technologie peut être intégrée dans le cadre d'une plateforme de vérification et de génération de rapports qui élimine le besoin d'intervention humaine. Conclusion: Les résultats présentés dans cette thèse contribueront à l'état actuel des connaissances dans la pratique de la coloscopie concernant les stratégies pour améliorer l'efficacité de la coloscopie dans la prévention du cancer colorectal. Cette étude fournira des indications précieuses pour les futurs chercheurs intéressés par le développement de méthodes efficaces de traitement des polypes colorectaux diminutifs. Le diagnostic optique nécessite une formation complémentaire et une mise en œuvre à l'aide de modules de caractérisation informatisés. En outre, malgré la lenteur de l'adoption des solutions informatiques dans la pratique clinique, la coloscopie assistée par l'IA ouvrira la voie à la détection automatique, à la caractérisation et à la rédaction semi-automatique des rapports de procédure. / Introduction: Colorectal cancer remains a critical public health concern in Canada. Screening programs could reduce the incidence of colorectal cancer and its associated mortality. A high-quality colonoscopy is appraised to be a cost-effective means of cancer prevention through identifying and removing cancer precursor lesions. Although colonoscopy can serve as a preventative measure against cancer, the procedure can impose an additional burden on the public health by removing and histologically evaluating insignificant diminutive colorectal polyps, which pose a minimal risk of advanced histology or cancer. The image-enhance technology would enable physicians to resect and discard diminutive polyps or diagnose and leave diminutive rectosigmoid polyps without histopathology examination. Despite the availability of computer-based polyp characterization systems, the practice of optical diagnosis remains limited due to the fear of cancer misdiagnosis, patient mismanagement, and the related medicolegal issues. Thus, alternative non-optical resection and discard strategies are imperative for improving the accuracy and safety of optical diagnosis for adaptation to clinical practice. These strategies should follow simple clinical criteria and do not require additional education or image enhanced devices. Furthermore, the safe practice of optical diagnosis, adequate decision-making regarding polypectomy technique, or surveillance interval depends on accurate polyp size estimation. The inter-endoscopist variability in polyp sizing necessitates the development of reliable and validated methods to enhance the accuracy of size measurement. A virtual scale integrated into a high-definition endoscope is currently available for automated polyp sizing, but its clinical feasibility has not yet been demonstrated. In addition to the points mentioned above, a high-quality colonoscopy requires the complete examination of the entire colonic mucosa, as well as the visualization of the ileocecal valve and appendiceal orifice. To date, no computer-based solution has been able to support endoscopists during live colonoscopies by automatically detecting and differentiating cecal landmarks. Aims: The aims of this thesis are: 1) to investigate the effect of limiting optical diagnosis to polyps 1–3mm on the safety of optical diagnosis for the management of diminutive polyps and the acceptance of endoscopists for its use in real-time practices while preserving its time- and cost-effectiveness potentials; 2) to develop and examine non-optical resect and discard strategies that can replace optical diagnosis while offering the same time- and cost-saving potentials; 3) to examine the relative accuracy of a virtual scale endoscope for measuring polyp size; 4) to train, validate, and test an artificial intelligence-empower model that can predict the completeness of a colonoscopy procedure by identifying cecal anatomical landmarks (i.e., ileocecal valve and appendiceal orifice) and differentiating them from one another, polyps, and normal mucosa. Methods: To achieve the first aim of this thesis, a post-hoc analysis of three prospective studies was performed to evaluate the proportion of patients in which advanced adenomas were found and optical diagnosis resulted in delayed surveillance in three polyp size groups: 1‒3, 1‒5, and 1‒10 mm. To achieve the second aim of this thesis, two non-optical strategies were developed and tested in two prospective studies: a location-based resect and discard strategy that uses anatomical polyp location to classify colon polyps into non-neoplastic or low-risk neoplastic and a polyp-based resect and discard strategy that assigns surveillance intervals based on polyp number and size. In all three studies, the agreement of assigning surveillance intervals based on high-confidence optical diagnosis or non-optical strategies with pathology-based recommendations, as well as the proportion of avoided pathology examinations and the proportion of immediate surveillance interval communications, was evaluated. The third aim of this thesis was addressed through a prospective pilot feasibility study that used the measurement of polyp specimens immediately after retrieving, following a polypectomy by a Vernier caliper as a reference to compare the relative accuracy of polyp size measurements between endoscopists and a virtual scale endoscope. Finally, the fourth aim of this thesis was assessed through prospective recording and annotation of colonoscopy videos. Unaltered images of polyp, ileocecal valve, appendiceal orifice and normal mucosa were extracted and used to develop and test a deep convolutional neural network model for classifying images for the containing landmarks. Results: Reducing the threshold of optical diagnosis would promote the safety of optical diagnosis by significantly decreasing the risk of discarding a polyp with advanced histology or the mismanagement of a patient with such polyps. Additionally, the non-optical resect and discard strategies could surpass the benchmark of at least 90% agreement in the assignment of post-polypectomy surveillance intervals compared with decisions based on pathologic assessment. Moreover, the virtual scale endoscope was demonstrated to be more accurate than visual estimation of polyp size in real-time. Finally, a deep learning model proved to be highly effective in detecting cecal landmarks, polyps, and normal mucosa, both individually and in combination. Discussion: Optical histology prediction of polyps 1‒3 mm in size is an effective approach to enhance the safety and feasibility of resect and discard strategy in practice. Non-optical resect and discard approaches also offer feasible alternatives to optical diagnosis when endoscopists are unable to meet the conditions for routine implementation of optical diagnosis, or when image-enhanced technology is not accessible. Both optical and non-optical resect and discard strategies could reduce additional costs related to histopathology examinations and facilitate the communication of the next surveillance interval in the same day as the index colonoscopy. A virtual scale endoscope would facilitate the use of optical diagnosis for the detection of diminutive polyps and allows for appropriate decision-making during and after colonoscopy. Additionally, the deep learning model may be useful in promoting and monitoring the quality of colonoscopies through the prediction of a complete colonoscopy. This technology may be incorporated as part of a platform for auditing and report generation that eliminates the need for human intervention. Conclusion: The results presented in this thesis will contribute to the current state of knowledge in colonoscopy practice regarding strategies for improving the efficacy of colonoscopy in the prevention of colorectal cancer. This study will provide valuable insights for future researchers interested in developing effective methods for treating diminutive colorectal polyps. Optical diagnosis requires further training and implementation using computer-based characterization modules. Furthermore, despite the slow adoption of computer-based solutions in clinical practice, AI-empowered colonoscopy will eventually pave the way for automatic detection, characterization, and semi-automated completion of procedure reports in the future.