Journal of Cardiovascular Disease Research
Deep Learning-Based Brain Tumor Classification Using Convolutional Neural Networks
D. Mahammad Rafi, Macha Mahipal Reddy, Kunduru Ashwini
JCDR. 2020: 2230-2238
Abstract
Deep Learning has emerged as the cutting-edge trend in the field of machine learning, garnering significant attention from researchers in recent years. This powerful machine learning tool has found widespread applications in solving complex problems that demand high accuracy and sensitivity, particularly in the medical domain. Brain tumors, being one of the most prevalent and aggressive malignant diseases, pose a significant threat with a short life expectancy when diagnosed at an advanced stage. Thus, accurate brain tumor classification is a crucial step after tumor detection to formulate effective treatment plans. In this study, we employ Convolutional Neural Networks (CNN), a prominent deep learning architecture, to classify a dataset of 3064 T1-weighted contrast-enhanced brain MR images into three categories (Glioma, Meningioma, and Pituitary Tumor). The proposed CNN classifier demonstrates exceptional performance, achieving an accuracy of 98.93% and sensitivity of 98.18% for cropped lesions. Furthermore, uncropped lesions yield results with 99% accuracy and 98.52% sensitivity, while segmented lesion images exhibit an accuracy of 97.62% and sensitivity of 97.40%.
» PDF