BRAIN TUMOUR DETECTION USING DEEP LEARNING

Abstract

When it comes to medical imaging, brain tumour segmentation is a particularly challenging and crucial endeavour. This is because manual categorization can lead to inaccurate prognosis and diagnosis. It's also a lot of effort when there's a lot of data to work with. In photographs, it's difficult to tell the difference between a brain tumour and normal tissue because both have a wide variety of appearances and are so similar. Fuzzy C-Means was used in this study to remove brain tumours from two-dimensional MRI images, which was then followed by classical classifiers and also cnn. There were tumours of various sizes, shapes and intensities included in the dataset used in the study. The scikit-learn library's classic classifier used SVM, KNN, MLP, LR, Naive Bayes, and Random Forest as well as other well-known machine learning algorithms. Convolutional Neural Networks (CNN) were then examined, which are better at predicting outcomes than standard neural networks developed with Keras or Tensorflow. Our analysis found that CNN's accuracy percentage was 97.87%. The primary goal of this investigation is to use statistical and textural data to discriminate between normal and aberrant pixels.