Deep Learning Technique to Detect and Classify Brain Tumor

Abstract

Tumors can develop anywhere in the brain, it is impossible to establish a general rule for how they are structured. Their properties, such as shape, contrast, and size, have also always been ambiguous. The wide variety of humans tormented by mind tumors is increasing. These elements drive the creation of an intelligent method for dividing aberrant brain tissues using deep learning. These MRI pictures can be segmented, and the segmented version of these images can be compared to tumor cells and normal brain tissues. The findings are determined and categorized based on the comparison. Data is segmented using a probabilistic neural network and a convolution neural network. Convolution Neural Networks (CNN) containing both 7*7 and 3*3 in an overlapping manner and constructed in a cascaded architecture are used to precisely segment tumors using the picture dataset Brats13. Similar to this, cancers are identified using probabilistic neural network and compared the results with parts of the brain that are normal. In this paper, a novel CNN and PNN architecture is suggested in place of the conventional models used in computer vision and image processing technologies.