Classification of Brain Tumor using Deep Learning with RMSPROP Optimization

Accurate classification of tumors in the brain is directly proportional to the precision level in diagnosis and efficient designing of treatment. In the current study, the performances of Convolutional Neural Networks optimized using three different algorithms of optimization: RMSprop, Adadelta, Adagrad &SGD are examined for classification of brain tumors that contain diverse types and varying features. The experimental results show that the best accuracy in the classification was obtained with RMSprop optimization with a score of 98%, where it was outperforming Adadelta at 88%, Adagrad &SGD at 75% with comparatively worse scores. RMSprop, with adaptive adjustment of the learning rates, can ensure faster convergence to optimal values. These results reveal the capabilities of RMSprop in making medical imaging diagnostics more accurate and provide a robust tool to aid neurology and oncology practitioners. In conclusion, this paper highlights how the resolution of computational issues coupled with the advancement of diagnostic precision through deep learning may bring a paradigm shift in medical imaging and treatment planning.