In this paper, we develop a customized version of a deep CNN to classify the when identifying MR brain images. Both the computational complexity and the accuracy of the proposed method are assessed. The results of the experiment showed that both the TPP and TPN had significantly increased. As part of the proposed approach, there will be no changes made to the weights in the fully connected layer. The significant reduction in computational complexity has made the technique viable for widespread use. Therefore, the purpose of this paper is to propose an improved alternative to CNN. Compared to the standard CNN approach, the proposed method yields about 3% better outcomes

Deep Convolutional Neural Network, True Positive Rate, True Negative Rate, Image, Classification