Analysis of Mineral Density in Bone Using Deep Learning and Smart Tracking System

Arthritis is the world one of the most worst immuno deficiency disease. It mainly invades joints of the bone in the body of humans. This deficiency analysis obtained from the images of the joints manually, is a chronic process and it requires experts to analyze the images obtain from the scan periodically which ultimately leads to delay in time and increase in cost. In order to analyze the problem effectively it is required to identify the mineral density of the bone BMD (Bone Mineral Density), a prime factor for identifying the disease in the bone and the risk level of fracture. The purpose of the work is to automate the process of analyzing density of the mineral present in the bone thermal images. The proposed detection model involves phases such as Image pre-processing, segmenting the portion of interest, extracting the features from the region segmented and classifying the abnormality. Initially the input thermal images is processed and filtered using two steps namely, de-noising the image using Anisotropic diffusion filter followed by enhancing technique using Contrast Limited adaptive histogram equalization-CLAHE. Next to image enhancement is segmentation, fuzzy C means is adopted for segmenting the affected portion. Once the portion is segmented, first order gray level features such a mean, median, energy, correlation entropy, variance and area are extracted and classified as normal or abnormal using deep neural network. Finally the output is fed to the controller and communicated remotely to the patients using GSM module.