This study presents a comprehensive radiomics-based approach for the classification of intracranial hemorrhage subtypes using machine learning techniques. We analyzed medical imaging data from 33 patients, extracting 52 quantitative radiomics features including first-order statistics, texture features, and advanced imaging biomarkers. The study employed multiple machine learning classifiers to differentiate between hemorrhage subtypes, with a particular focus on feature importance and statistical relationships. Our results demonstrate the potential of radiomics features in characterizing hemorrhage patterns, achieving classification accuracy up to 60% using logistic regression. The findings suggest that radiomics analysis, combined with machine learning, can provide valuable insights into hemorrhage subtype classification, potentially aiding in clinical decision-making and treatment planning.