Metodologia Computacional para Detecção e Diagnóstico Automáticos de Estrabismo em Vídeos Digitais utilizando o Cover Test

• Ano de defesa: 2017
• Autor(a) principal: VALENTE, Thales Levi Azevedo
• Orientador(a): SILVA, Aristófanes Corrêa
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Maranhão
• Programa de Pós-Graduação: PROGRAMA DE PÓS-GRADUAÇÃO EM ENGENHARIA DE ELETRICIDADE/CCET
• Departamento: DEPARTAMENTO DE ENGENHARIA DA ELETRICIDADE/CCET
• País: Brasil
• Palavras-chave em Português: Diagnóstico de estrabismo;Cover test;Processamento de imagens; Vídeos digitais
• Palavras-chave em Inglês: Diagnosis of strabismus; Cover test; Image processing; Digital videos
• Área do conhecimento CNPq: Engenharia Médica
• Link de acesso: http://tedebc.ufma.br:8080/jspui/handle/tede/1292

Medical image processing can contribute to the detection and diagnosis of human body anomalies, as an important toll to minimizes the degree of uncertainty in diagnosis, and provides specialists with an additional source of information. Strabismus is one of these anomalies, which affects approximately 4% of the population. Strabismus modify the mechanism of vision, resulting in aesthetical and sensory problems. The first case is reversible at any age. Regarding sensory problems, late treatment may lead to an irreversible condition. The cover test is one of the types of exams used to diagnose this pathology. The use of low cost computational resources to aid in the diagnosis and treatment within the strabismus subspecialty is not yet a reality. Therefore, this work presents a computational methodology to automatically pre-diagnosis strabismus through digital videos featuring cover test. To do this, the methodology was divided in 8 stages: Acquisition (1), Detection of eyes region (2), Location of pupil (3), Location of limbus (4) , Tracking eyes (5), Detection of occluder (6) , Detection (7) and Diagnosis of strabismus (8). The effectiveness of the method in the indication of the diagnosis was evaluated by comparisons made with the diagnosis provided by the expert. For detection of strabismus, the method proposed achieved 100% specificity, sensitivity of 80% and 93.33% accuracy and was demonstrated to be 87% accurate in diagnosing strabismus, while acknowledging measures lower than 1 prismatic diopter, and an average error of 2.57 prismatic diopters in deviation measure. So, was demonstrated the feasibility of using computational resources based on images processing techniques to achieve diagnosing strabismus through cover test.