Multi-Feature Analysis for Automated Brain Stroke Classification Using Weighted Gaussian Naive Bayes Classifier

In today's world, brain stroke is considered as a life-threatening disease provoked by undesirable blockage among the arteries feeding the human brain. The timely diagnosis of this brain stroke detection in Magnetic Resonance Imaging (MRI) images increases the patient's survival rate. However, automated detection plays a significant challenge owing to the complexity of the shape, dimension of size and the location of stroke lesions. In this paper, a novel optimized fuzzy level segmentation algorithm is proposed to detect the ischemic stroke lesions. After segmentation, the multi-textural features are extracted to form a feature set. These features are given as input to the proposed weighted Gaussian Naive Bayes classifier to discriminate normal and abnormal stroke lesion classes. The experimental result manifests that the proposed methodology achieves a higher accuracy as compared with the existing state-of-the-art techniques. The proposed classifier discriminates normal and abnormal classes effciently and attains 99.32% of accuracy, 96.87% of sensitivity and 98.82% of F1 measure.

Automated summary

In today's world, brain stroke is a life-threatening disease caused by blockage in the brain's arteries. Timely diagnosis through MRI images is crucial for increasing survival rates, but automated detection faces challenges due to the complexity of stroke lesions. This research proposes an optimized fuzzy segmentation algorithm and achieves higher accuracy compared to existing techniques.