Effective Prediction of Brain Tumor Using Machine Learning Algorithms

Medical science has been advancing with the help of technology in which Machine Learning plays a major role starting from checking a person’s vitals to predicting benign tumors in almost all parts of the body. So, these advancements have inspired us and got us interested in brain tumor prediction. Brain Tumor prediction is one of the most crucial and tedious tasks in the field of medical image processing as classification of tumors manually can result in inaccurate diagnosis and prediction. As there is much similarity between normal tissues and tumors, there is difficulty in their classification, thus making them unyielding. We have used various classifiers namely logistic regression, random forest, decision tree, and Naïve Bayes in order to predict whether a tumor is there or not. The observed accuracies for the above classifiers were not satisfactory. So, in order to predict the tumors with utmost accuracy, we have chosen a machine learning algorithm called Convolutional neural network using Keras and Tensorflow. The convolutional neural network is a deep learning technique in image classification. This is a supervised learning algorithm and feed-forward neural network with 20–30 networks used to identify patterns in raw images without any image processing. This technique has resulted in an accuracy of 90%.