An Advanced Feature Fusion and Subject- Feature Fusion with Contrast Enhancement Model for Osteoporosis Detection in Femur Bone

Abstract

Osteoporosis causes the mineral density of bones to decrease, the bones become more porous and fragile, which increases the risk of fracture. Dual Energy X-ray Absorptiometry (DEXA) detects bone mineral density (BMD) effectively, it is the most widely used method for diagnosing osteoporosis. Despite DEXA's efficacy in determining BMD, some disadvantages of the technology included size, expense, and limited availability. To overcome these issues, the medical image based osteoporosis diagnosis was done. Yet those models also have some impacts like poor feature extraction and low contrast. The best way to diagnose osteoporosis was analyzed both BMD and images at a time. This merging model provided a better outcome but the linkage of both images and subject values was most complicated and take too much of time. In proposed work, a fusion strategy based detection approach was designed to predict osteoporosis in femur bone. The proposed model have three stages namely feature extraction, feature fusion and subject-feature fusion. The collected X ray images and its subject record were collected and split separately for accurate prediction. Pre-processing and augmentation process were done to improve the image information. Then, extract the images using two different methods and fused both features. Further, the subject records were fused with its appropriate features to detect the osteoporosis disease appropriately using a deep learning approach. The proposed model provides 97% accuracy with 7% false positive rate and compared to another traditional models. The suggested approach detects osteoporosis effectively so it was well suitable for real-time applications.