Transforming clinical diagnostics: A deep learning approach for biometric face recognition using multistage regression capsule networks and optimized grey wolf algorithm

Face recognition (FR) technology is increasingly being used in clinical diagnostics and customized medicine, in addition to typical security applications. However, reliably identifying patients based on face traits in big and heterogeneous datasets remains a major difficulty. The study proposed a novel framework, the Multistage Regression Capsule Network with Modified Grey Wolf Optimization (MRCN-MGWO), to improve the accuracy and efficiency of patient identification in healthcare settings. The MRCN-MGWO model uses deep learning to increase diagnostic accuracy by assessing facial features and using specialized preparation techniques for medical photos. Facial images are first denoised using a median filter (MF) before being enhanced with contour-based image enhancement (EIC) to improve clarity. The Multistage Regression Capsule Network (MRCN) generates robust feature vectors to detect distinct facial patterns, whereas the Modified Grey Wolf Optimization (MGWO) approach optimizes the weights and biases of a stacked autoencoder (SAE). The MRCN-MGWO architecture is tested on the benchmark FEI dataset and shows promise for accurate patient identification in a variety of clinical settings by outperforming current face recognition techniques. As a result, the proposed MRCN-MGWO model improves precision and increases the rate of facial recognition while maintaining high accuracy.