Osteoporosis identification based on the validated trabecular area on digital dental radiographic images

Abstract Research for identifying osteoporosis using dental radiographic images is increasing rapidly. Subjects data from various regions and countries have been used by many researchers. This indicates that osteoporosis has become a widespread disease that should be studied more deeply. A method for osteoporosis identification based on the validated trabecular area present on digital dental radiographic images is proposed in this paper. Digital dental radiographic images of subjects were first prepared. This study performs a sequence of morphological operations to obtain the region of interest (RoI) from the validated trabecular area on the images. The validated area is then evaluated using dice similarity method. Bone mineral density is measured using dual X-ray absorptiometry at two sites to assess the presence of osteoporosis. We propose four statistical features, namely deviation, entropy, homogeneity, and correlation, which are extracted from the RoIs. These four features are obtained through feature extraction followed by feature selection using the C4.5 feature selection method. Thereafter, multilayer perceptron is used to predict the presence of osteoporosis by statistical texture analysis. The average dice similarity coefficient for all of RoIs achieves an index of 0.8924. Multilayer perceptron classifier is an appropriate method in our proposed work, which achieves an accuracy of 87.87%. This research shows that the proposed method using the sequence of morphological operations achieves high similarity for forming validated trabecular area and the four statistical features achieves a good performance for osteoporosis identification.