An innovative framework for brain tumor detection and types classification using ScLe2 LU–DNN and GaPL-MOA

Worldwide, Brain Tumor (BT) is a major health risk for both adults and children; also, it causes severe impairment of organ function and even death. Therefore, most of the clinical researchers focused on providing the warrant efforts for the timely detection of BT. Nevertheless, the conventional systems were inefficient owing to the heterogeneous nature of the tumor and imbalanced samples. Hence, this paper proposes a novel ScLe 2 LU–DNN and GaPL-MOA-based effective BT detection and Types Classification (TC). Firstly, the brain MRI image undergoes pre-processing. Afterward, the pre-processed data is balanced using Ch-SMOTE. Next, the balanced data is grouped; then, the patches are extracted. Furthermore, the contrast of the extracted patches is enhanced. Subsequently, the tumor is segmented using the WBo-Ro. Furthermore, the feature extraction is done; further, the optimal features are identified using GaPL-MOA. Lastly, the features being selected are given as input to the ScLe 2 LU–DNN, which proficiently classifies various types of BT. Thus, the simulation outcomes confirmed that the research methodology attained more impressive outcomes. The devised ScLe2LU-DNN framework attained a classification accuracy of 98.78%, outperforming standard deep learning methodologies (CNN, RNN, ANN) with an average accuracy of 88.26%. The WBo-Ro segmentation algorithm minimized segmentation errors by achieving a Dice Score (DS) of 0.08771 and a Mean Absolute Error (MAE) of 0.0139, thus improving tumor boundary detection.The GaPL-MOA optimization was found to possess superior feature selection, with increased classification accuracy by 7.8% from ACO and WOA.