Enhanced Brain Tumor Classification through Gamma Correction in Deep Learning

Classification of brain tumors is a problem in computer-aided diagnosis (CAD). This study classifies three classes of brain tumors: gliomas, meningiomas, and pituitary tumors. Image enhancement is useful for improving the quality of images to be recognized by Computer-Aided Diagnosis (CAD) systems. Gamma correction is one spatial method aimed at manipulating contrast. This method operates with a spatial approach and has relatively low computational time but yields satisfactory results in certain cases. This research compares Gamma Correction with Convolutional Neural Network (CNN) in the classification of brain tumor types. The CNN method without Gamma Correction achieves an accuracy of 86.52%, precision of 83.63%, sensitivity of 86.11%, and specificity of 93.27%. The application of Gamma Correction at 1.5 results in improved performance with an accuracy of 88.80%, precision of 86.49%, sensitivity of 88.06%, and specificity of 94.50%. Meanwhile, Gamma Correction at 0.5 shows an accuracy of 88.59%, precision of 87.59%, sensitivity of 86.68%, and specificity of 94.17%. Overall, the implementation of Gamma Correction in the classification of brain tumor types successfully enhances the CNN classification performance in terms of precision, sensitivity, and specificity compared to without its use.