A Dominant SM-GN-RNN and CR-FLS Based Brain Tumor Recognition Along With Survival Rate Prediction

Abstract A promising framework called smooth maximum unit Gaussian batch normalization recurrent neural network (RNN) (SM-GN-RNN) and cube root fuzzy logic system (Cr-FLS)-based brain tumor (BT) classification and survival rate prediction is proposed in this paper. First, various individuals' brain functional magnetic resonance imaging (fMRI) is collected and preprocessed. Next, active contour technique is utilized in segmenting tumor from preprocessed images. After that, tumor volume is calculated from segmented images. Similarly, necrotic area is identified from the segmented images. Lastly, malignant tumor and tumor volume are subjected to Cr-FLS, which predicts the survival rate of BT individuals. Thus, experimental findings stated that proposed methodology is eye-opening framework for reducing mortality rate due to BT.