MRI brain image analysis using deep learning techniques and multi-class support vector machine

In recent times, an identification and classification of brain tumour become more essential to save human life. Brain tumour detection is considered most challenging problem and many researchers are finding optimized solution for early diagnosis. It occurs because of the irrepressible growth of cells in the brain and classified as malignant and benign tumour. In this research work, an automatic brain tumour detection system using CNN with Softmax and CNN with Multiclass SVM (M-SVM). It was clearly comprehend that the correct learning procedures and matching must yield perfect results. A database of the medical image was complex to divide. Classifying and identifying brain tumour a novel learning procedure, the combination of CNN and M-SVM were used to classify the input MRI Brin image is tumour or non-tumour. This Proposed method evaluated by the fig share dataset and proves the proposed method produced high accuracy. Evaluation and testing of the process used 5 fold validation process with Harvard, Radiopaedia and Figshare dataset. The proposed methods evaluated using Figshare dataset and classifier produced classification accuracy of 98.9% of CNN with Softmax and produced an accuracy of 99.2% of CNN with M-SVM.