RETRACTED ARTICLE: Anticipating brain tumor classification and severity levels: employing the CDF-BILSTM model approach

In adults, Brain Tumor (BT) is the deadliest disease. Thus, in the treatment of BTs, accurate detection and classification are essential. Many approaches are developed, but these previous approaches still have defects in accuracy and reliability and they only predict the tumor or tumor localization. So, a cumulative density function based bi-directional long short term memory network (CDF-BiLSTM) centered BT classification and severity level prediction is proposed. Primarily, by employing Suppressed Sobel Operator-Covariance Speckle Reducing Anisotropic Diffusion (SSO-CSRAD) techniques, noise in the input MRI image was removed. Next, by deploying Neighbourhood Function-based Texton Map Generation (NF-TMG) techniques, the tumor location was detected. Then, the tumor was segmented by using Gradient Operator-based Balloon Snake (GO-BS). Afterward, the features are extracted; also, the optimal features are selected by deploying Secant Wild Geese Migration Optimization (SWGMO). Then, the type of tumor was detected by utilizing CDF-BiLSTM. In the end, for each class, the severity prediction was determined. The experimental analysis was done, where the proposed model’s efficiency is depicted.