Brain Tumor Detection Using Convolutional Neural Network

Information Science

In the field of medical image processing, brain tumor segmentation is one of the most important and challenging problems since manual classification by a person can lead to incorrect diagnosis and prediction. Furthermore, it is a frustrating chore when there is a lot of data that needs to be helped. The extraction of tumor regions from images becomes hard because brain tumors show a great degree of visual diversity and resemble normal tissues. In this work, we suggested using the fuzzy C-Means clustering approach to extract brain tumors from 2D magnetic resonance imaging (MRI). Conventional classifiers and convolutional neural networks were then used. The experimental investigation was conducted using a real-time dataset that included a variety of tumor sizes, locations, forms, as well as various picture intensities. We used six conventional classifiers in the traditional classifier section, including Support Vector Machine(SVM), Scikit-learn was used to implement K-Nearest Neighbor (KNN), Multilayer Perceptron (MLP), Logistic Regression, Naïve Bayes, and Random Forest. Then, as it performs better than the conventional ones, we switched to Convolutional Neural Networks (CNNs), which are implemented using Keras and Tensorflow. CNN achieved a really impressive accuracy of 97.87% in our work. This paper's primary goal is to differentiate between normal and aberrant pixels using statistical and texture-based criteria.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.