MRI-Image based Brain Tumor Detection and Classification using CNN-KNN

A brain tumor is the topmost dangerous disease that can be a cause of death in humans. Machine learning tools are generally used for medical science brain tumor detection and classification. Although the commonly used ANN tool can classify the features, it is unable to extract features automatically, so the accuracy is low. To solve the problem this paper designed a new CNN-KNN architecture that can automatically detect and classify the various forms of tumors. The images from the dataset are resized and gray-scale transformation is used to transform RGB image to gray image in pre-processing step. Different intensity-based features are extracted through CNN in the feature extraction step. Finally, two machine learning classifiers, Softmax and KNN are used in the classier step to detect and classify the various forms of brain tumors automatically. This proposed method CNN-KNN effectively detects and classifies the various forms of the tumors with the promising accuracy of 95.7%. Among the two types of classifiers, KNN has performed the best accuracy.