Enhanced Identification of Brain Tumors Based on 3-Dimensional Hybridseqnet-CNN with RNN

Abstract

Brain tumors pose a significant threat to human health, necessitating accurate and timely diagnosis for effective treatment. The identification technique is difficult since brain tissues in both healthy and pathological states resemble one another. It is particularly challenging to automatically identify medical images when conducting clinical examinations of brain tumors and earlier patient care. Rapid decision-making is made possible by computerized medical imaging, which also helps doctors provide patients with necessary care. Brain tumors can take diverse shapes and arise in various regions across the brain. Machine learning models face a huge hurdle in effectively identifying and classifying these various tumor types and locations. The data samples used to train our suggested 3DHybridseqnet-CNNRNN approach are gathered from the BraTS2021 database. Blur reduction is first performed in raw data using a standardized median filter (SMF) for picture de-noise and quality enhancement. The modified 3D adaptable threshold (M3DT) approach is subsequently employed to segment the data. The features retrieved using the wavelet-based Local Binary Pattern (W-LBP) technique are further processed using the whale optimization algorithm (WOA) to yield optimized feature subsets. Their importance, however, has to be determined and validated, as the abstract of this study does not provide definitive data. The developed 3DHybridseqnet-CNNRNN model is then employed for improved brain tumor identification. The simulated results (using the Python tool) established that the recommended techniques improved identification consequences, as measured by precision, accuracy, recall, and f-measure metrics, are greater than the existing methods. With a combination of improved identification accuracy (98%), medical professionals may be able to make more precise diagnoses and save lives. Therefore, our results on improved tumor detection show very good efficiency when compared with comparable methods.