Machine Vision Based Analysis of the Trabecular Bone Structure in frequency domain for Osteoporosis

Osteoporosis is a disorder which affects skeletal and is generally found in elders and women. In this methodology a novel algorithm has been developed which basically detects the radiographic image of the musculoskeletal bone. It predicts and helps in identifying whether the bone is healthy or osteoporotic which is usually difficult to analyze as these radiographs are dark and do not give the information unless it is exposed to some brightness. Here a method has been developed for feature selection in frequency domain which proves to be more reliable and accurate than spatial domain. The hybrid combination of wavelet and Gabor features are fed wrapper algorithm for best features evaluation followed by feed forward neural nets. The proposed method helps to discriminate the osteoporotic patient from that of normal subject by comparing frequency domain features of their 2-D radiographic images to test the performance of the proposed system. The experimental results achieved gives the accuracy of more than 95%.