Classification of brain neoplasm from multi-modality MRI with the aid of ANFIS classifier

Abstract

In medical Image processing, the chief objective is to detect Neoplasm effectively. Neoplasm is basically a sort of abnormal excessive cell growth but when it generates a mass, it is referred as tumors. Brain tumor (BT) is a deadly disease and also it is regarded as a common sort of cancer on adults and even on children. Therefore, early recognition of the correct sort of BT is significant for devising a proper treatment chart and envisioning patients' response to the adopted treatment. Human understanding of countless medical images (Abnormal or Normal) may bring misclassification and thereby there is a requisite of the automated recognition system for classifying the BT types. This paper offers an effective framework for classifying the BT from the multi-modality Magnetic Resonance Images (MRI) by employing ANFIS (Adaptive Neuro-Fuzzy Inference System) classifier. Primarily, the input data-set undertakes the process of skull stripping. Subsequently, the resultant skull striped image undergoes preprocessing utilizing AHE (Adaptive Histogram Equalization). Subsequently, the clustering process is done by employing the Modified-Fuzzy C Means (MFCM) clustering algorithm. From the benign and malignant classes, features are extorted, and then the optimized features are attained utilizing the Adaptive Elephant Herd Optimization (AEHO) algorithm. Finally, the different sorts of BT are effectively classified by implementing the ANFIS classifier. The outcomes are examined and contrasted to the other conventional techniques to corroborate that the proposed work classifies the BT in great efficiency.