Application of Image Processing in Detection of Bone Diseases Using X-rays

Abstract This study compares published algorithms for the detection of bone diseases particularly osteoporosis (which is characterized by low level of bone mineral density and porosity due to microarchitectural deterioration) with claimed accuracy on based on the author selected dataset. In this study common dataset is used to verify accuracy and performance of the published algorithms by comparing the output results published by the authors and the results gathered and compiled by this study. Features like contrast, correlation, homogeneity, entropy, energy along with standard deviation, range, skewness are calculated from Gray-Level Co-occurrence Matrix (GLCM) technique. Study also implement all algorithms published by the authors and tested with common dataset containing digital images of X-ray femur (left and right leg femur; both). The research concludes that the standard deviation, image contrast and specifically energy with entropy plays a vital role in determining the disease by performing Haralick features textural analysis on plain (Non-DEXA) radiographs.