Feature extraction with mixture gaussian for stroke classification

The stroke remains the second leading cause of death in the world. The stroke diagnosis is usually obtained by neuroimaging analysis, and among the main techniques, Computed Tomography (CT) is the most used. A quick diagnosis of stroke can generally contribute positively to the patient's recovery. CAD systems that analyze CT scan images become extremely important when diagnosis speed is a relevant factor for patient recovery. This work presents a new feature extractor using gaussian mixtures, called Mixture Gaussian Analysis of Brain Tissue Density (MGABTD). The MGABTD achieved accuracy and f1-score of 99.9%. The results demonstrated the effectiveness of the method in extracting features used to determine whether a CT scan is normal or shows an ischemic or hemorrhagic stroke.