Segmentação dos hipocampos em imagens de ressonância magnética usando modelos geométricos deformáveis e informações de textura

Abstract

Hippocampi are brain structures intimately related to cognition, hence they are of great clinical interest for the diagnosis and analysis of the progression of neurodegenerative diseases, such as Alzheimer’s disease (AD) and Parkinson’s disease. The hippocampal atrophy by age group is considered to be one of the main AD’s biomarkers. Since manual delineation (for further assessment) of hippocampi is a difficult task and prone to mistakes, due to the variability between distinct specialists delineations, a plethora of automated hippocampi segmentation methods using tridimensional (3D) Magnetic Resonance (MR) Images have been proposed in the literature, among which stand out those based on geometrically deformable models. Thus, the main objective of this work was the development of an automatic method to segmentate hippocampi in 3D MR images using geometrically deformable models and textural information from hippocampi regions. Such method allows aid in the diagnosis of neurodegenerative diseases. In addition, a web application was developed, aiming a more practical way of making this work available to specialists in the subject.