A study on quantitative assessment of osteoporosis based on texture analysis

A quantitative assessment on osteoporosis by using texture analysis has been developed on the basis of coherence mapping, two-dimensional power spectral estimation with modulated transfer function correction and fractal features analysis on trabecular pattern of bone X-ray images. Three objective features of trabecular pattern are selected to form a feature set that can be used in automatic classification for two grades of bone:normal bone and the bone with a medium degree of osteoporosis. This set of features is extracted from each of 16 sample bone X-ray images. With the use of Euclidean norm in the feature space, a center of cluster of each grade of bone is defined as the prototype. For each bone X-ray image, the distance from its features to the two prototypes in feature space is computed to provide an objective judgement on the degree of osteoporosis. The machine classification of 10 bone X-ray images based on the developed prototype is given, and the result indicates that the proposed method is potntially applicable to the objective diagnosis of osteoporosis.