Comparison among Supervised Classifiers for Classification of Brain Tumor

To augment the survival rate of tumor patients, an early detection of tumor is immensely significant. Supervised classifiers such as Convolutional Neural Network (CNN), Artificial Neural Network (ANN) and Support Vector Machine (SVM) can aid the radiologists to improve the diagnostic accuracy in concise time. In this paper, a comparison based on features is done by incorporating those supervised classifiers using same data sets to find out the condition of brain tumor. At first, normal and tumor affected brain is classified with accuracy of 97.18%, 95.80%, and 94.4% by applying CNN, ANN, and SVM classifiers respectively. Then benign (non-cancerous) and malignant (cancerous) tumors are also detected from those classifiers with accuracy of 92%, 88%, and 84% respectively. Finally, area of brain tumor is enumerated to get the condition of tumor. For CNN classifier False Discovery Rate (FDR) for sound and tumor affected brain is 0% and for benign and malignant tumor is 9.10% which are less than both ANN and SVM classifiers. Matthews Correlation Coefficient (MCC) and F1 score for CNN classifier is higher than both ANN and SVM classifiers. For both classifications CNN classifier has provided better result than ANN and SVM classifiers for same amount of data set.