Optimized Deep Learning for Human age Estimation

Abstract

Background/ Introduction: Human age assessment plays a crucial part in diagnosing genetic problems, and development abnormalities in children. They are also used in several applications, like forensic investigation and criminal scenes. Generally, human age is estimated from X-ray images of bones, and the manual estimation of human age is highly subjective, as they depend on the medical experience of the professionals. Further, they are prone to error and are laborious, thereby requiring automated Bone Age Assessment (BAA) for determining human age with high accurateness.
 Materials and methods: This work presents a novel approach using Deep Learning (DL) and optimization considering the X-ray images of hand. Here, human age is estimated using the Deep Residual Network (DRN), whose parameters are trained using the proposed Beluga whale lion optimization (BWLO) algorithm. Further, several processes, like pre-processing, Region of Interest (RoI) extraction, Image augmentation, and feature extraction are used to enhance the accuracy of the estimation model.
 Results and conclusion: The BWLO_DRN is examined for its superiority considering metrics, like accuracy, Sensitivity, Specificity, Positive Predictive Value (PPV) and Negative Predictive Value (NPV).