A Hybrid CNN-KNN Model for MRI brain Tumor Classification

This paper proposes a hybrid model (CNNKNN) for Magneto Resonance Image (MRI) brain tumor classification, which integrates convolutional neural networks (CNNs) with K-Nearest Neighbor (KNN). The CNN model is considered to extract the features and then applied to KNN classifier to predict the classes. Experiments are conducted on an open dataset images chosen from BraTS 2015 and BraTS 2017 database for classification. An accuracy of 96.25% is the performance shown using this proposed method on the test set and proven to be better in terms of accuracy, error rate, F-1 score, sensitivity, and specificity based on experimental results.