EFFECTIVE MACHINE LEARNING TECHNIQUES FOR BRAIN PATHOLOGY CLASSIFICATION ON MR IMAGES Ruaa M. Mahmood, Nehad T.A. Ramaha, and Ismail R. Karas
Dept. of Computer Engineering, Karabuk University, Demir Celik Campus, 78050 Karabuk/Turkey.
Abstract
Since a brain tumor is essentially a collection of aberrant tissues, it is crucial to classify tumors of the brain using MRI before beginning therapy. Tumor segmentation and classification from brain MRI scans are well-known to be challenging and important endeavors. It has the potential used in diagnostics, preoperative planning, and postoperative evaluations. Furthermore, it^s crucial to get accurate measurements of the tumor^s location on an MRI of the brain. The development of machine learning models and other technologies will let radiologists detect malignancies without having to cut into patients. Pre-processing, skull stripping, and tumor segmentation are the steps in the process of detecting a brain tumor and measurement (size and form). After a certain period of time, CNN models get overfit because of the large number of training images used to train them. That^s why we now have deep CNN that uses transfer learning. CNN-based Relu architecture and SVM with fused retrieved features via HOG and LPB are used to classify brain MRI tumors (glioma or meningioma). The methods^ efficacy is measured by precision, recall, F-measure, and accuracy. The modified systems showed that the accuracy of SVM with combined LBP with HOG is 97% and modified CNN of 98%.
