Advanced Analysis of Bipolar Disorder Through Computer Vision Technology

This research paper focuses on the analysis of Bipolar Disorder (BD) through the application of structural magnetic resonance imaging (MRI) and a supervised learning framework. The primary goal is to identify specific abnormalities in brain cells that serve as indicators of BD. BD is linked to disruptions in cellular metabolism within specific brain systems, particularly the anterior limbic brain network. These disruptions manifest as anomalies in brain activation patterns and specific neurochemical measures. To achieve this objective, a 3D texture analysis approach was employed. Texture features were extracted from the MRI data, and a supervised learning algorithm, Support Vector Machine (SVM), was used to classify samples as either healthy or abnormal. Initially, 117 texture features were extracted, and a Genetic Algorithm (GA), inspired by natural selection, was applied to optimize the feature selection process. The GA identified the 20 most informative features from the initial set. Upon selecting the optimal features, the SVM algorithm was trained on the dataset. The experimental results yielded promising outcomes, with the trained SVM model achieving an accuracy of 90.42%. This suggests that the system effectively classified the majority of samples as either healthy or abnormal based on the identified brain abnormalities. These findings offer valuable insights into the application of supervised learning and MRI data analysis for investigating and comprehending BD.