Selective block based approach for neoplasm detection from T2-weighted brain MRIs

Abstract

A realistic challenge in neuroanatomy is to assist radiologists to detect the brain neoplasm at an early stage. This paper presents a fast and accurate Computer Aided Diagnosis (CAD) system based on selective block based approach for neoplasm (tumor) detection from T2-weighted brain MR images. The salient contribution of the presented work lies in a fast discrimination using selective block based approach. Local binary patterns are considered as features, which are trained by support vector machine. The experiments are performed on the dataset of 100 patients, in which 55 patients reported with brain tumor and rest as normal. The proposed CAD system achieves 99.67% accuracy with 100% sensitivity. The comparative studies on the same dataset report the outperformance of proposed CAD system by comparison with some of the existing system.