Convolutional Neural Network (CNN) Prediction on Meningioma, Glioma with Tensorflow

Abstract

Brain tumors can significantly affect a patient's life in a variety of ways. Classification of brain tumors is also important. Artificial intelligence (AI) techniques such as machine learning and deep learning can be very beneficial to physicians to classify tumors based on various parameters. In this study, the dataset is comprised of two distinct components which were prepared specifically for testing and training purposes, respectively. TensorFlow software library was used to utilize of Convolutional Neural Network (CNN). Since the most suitable weight values to solve the problem in deep learning are calculated step by step, the performance in the first epochs was low and unstable compared to the progressive values, and the performance increased as the number of epochs increased. However, after a certain step, the learning status of our model decreased considerably. The accuracy of the created model was observed to reach 0,90. As a result, as stated in its intended use, a mechanism that helps physicians and uses time efficiently has been successfully developed. In order to obtain more efficient results, the data set used in the study can be expanded, allowing deep learning models to work more effectively.