Integrating Diverse Approaches in Medical Image Analysis: PAA-CNN and Feature Extraction Fusion for Classification of Psychological Disorders using Anatomical Scans

A psychological disorder is a condition that impacts a persons behavior. Due to the contemporary way of life, a large number of individuals suffer from disorders like stress, depression, and other similar ones. These might turn into severe issues that would significantly impact a persons quality of life. We present a sample framework that uses an Anatomical scan captured along with fMRI. Anatomical scans were used to extract characteristics, which were then utilized to classify using a random forest classifier. In a follow-up experiment, CNN is applied to features obtained from the piecewise aggregate approximation method for multi-class classification of psychological disorders. This method performs noticeably better than the conventional feature extraction techniques, and with this approach, obtained an accuracy of up to 79%. Combining several approaches may boost the classification and prediction accuracy of medical data.