Automated Detection of Osteoporosis Using X-Ray Image Based on Optimized Recurrent Neural Network and Context Encoder Network

Abstract

Loss of bone mineral density (BMD) is the primary cause of osteoporosis, a prevalent skeletal condition. In the healthcare industry, inefficiencies are typically caused by a degree of variability resulting from manual and semi-automated processes used in medical data analysis. Therefore, the automatic prediction is considered as essential in order to detect whether the patient is affected by osteoporosis or not. However, the primary issue with existing techniques is imprecise disease diagnosis at early stages, as well as a lack of processing approaches. In order to overcome this issue, deep learning-based approach is developed. Initially, the x-ray images are collected and pre-processed using frost filter, reformed histogram equalization in order to improve the image resolution. Then the pre-processed image was segmented to obtain the region of interest (ROI) required for prediction. Values for these metrics will be better and suited for early osteoporosis diagnosis to improve the living standard of people.