Brain Stroke Detection using MRI Image and Logistic Regression

Brain stroke occurs because of blockage in the artery which deliver oxygenated blood to the brain. Acute Ischemic Stroke (AIS) is mostly occurred brain stroke. Early detection of brain stroke can be life saving for patient. For study of brain tumor detection and segmentation, MRI Images are mostly used in recent years. Due to MRI Images we can detect the brain tumor. A brain MRI can help to analyze bleeding, tumors, infections, damage from an injury or a stroke, or problems with the blood vessels inside the brain. Using the proposed algorithm, automatic brain stroke detection is possible in the early stage. In this research, using MRI images and logistic regression classification, technique to determine Ischemic Stroke is enhanced. Initially MRI brain images are collected. In the pre-processing gray image conversion and noise removal on the raw data is performed. After that segmentation are carried out on the stroke portion using hue, saturation, and value color threshold (HSV). To reduce the computational complexity, segmented images are converted into binary images. Features like standard deviation, mean hue, mean variation and Stroke Area are extracted from the dataset. Logistic regression classifier is used to build the machine learning model for stroke detection. Finally, the accuracy of the proposed algorithm is compared with existing algorithms. It is observed that the accuracy of the proposed algorithm is higher than existing algorithm and hence using proposed algorithm it is possible to detect the brain Stroke in early stage with higher accuracy.