A CNN-Based Computer-Aided Diagnosis System for Brain Tumor Identification and Classification in MRI Images

Abstract

Brain tumor detection and diagnosis are critical tasks in medical image analysis, as early identification significantly improves treatment planning and patient survival rates. Magnetic Resonance Imaging (MRI) is widely used for brain tumor diagnosis due to its superior soft-tissue contrast and non-invasive nature. However, manual examination of MRI images by radiologists is time-consuming and susceptible to inter-observer variability, necessitating the development of automated and reliable diagnostic systems. In this work, an intelligent algorithm is developed for automatic detection and diagnosis of brain tumors using magnetic resonance images. The proposed framework incorporates image preprocessing, adaptive contrast enhancement, skull stripping, feature extraction, and deep convolutional neural network (CNN)-based classification for accurate tumor identification. The segmented tumor regions are further analyzed to distinguish different tumor categories and improve diagnostic performance. Experimental evaluation was conducted using benchmark MRI datasets, and the proposed model achieved a classification accuracy of 98.24%, sensitivity of 97.85%, specificity of 98.61%, precision of 97.94%, and F1-score of 98.02%. Comparative analysis demonstrated that the proposed algorithm outperformed conventional machine learning and existing CNN models in terms of accuracy and computational efficiency. The developed framework provides a reliable and efficient computer-aided diagnosis tool for early brain tumor detection and can assist radiologists in clinical decision-making and personalized treatment planning.