Diverse Convolutional Neural Network Models for Feature Extraction from Brain Tumor Images
G. Malleswari, A. Srinivasa Reddy
Abstract
Medical imaging is important for both the doctor's precise diagnosis and the patient's subsequent treatment. Through several investigations, different algorithms have been incorporated into medical imaging. Medical picture feature extraction is investigated on a sizable amount of data to produce processing outputs that help doctors diagnose cases more precisely. To begin, the local binary pattern features will be extracted from the tumour image using rotation invariance. The rotation and movement of the image are altered, but its stability with respect to the coordinate system is maintained. By accurately describing the textural attributes of the tumour picture's surface layer, the approach increases the robustness of the image region description. The test results show how well the CNN algorithm retrieves features from photos of tumours. The techniques used by DenseNet, Xception, and Binary pattern CNN are compared in this work. With additional information, the CNN approach can precisely extract characteristics from tumour CT scans.