Diagnosis of Osteoporosis from radiographs using Image processing and Deep learning techniques
A low cost prescreening tool for early diagnosis of osteoporosis using metacarpal radiogrammetry and texture analysis is developed and validated using sample data of Indian and Swiss population. Third metacarpal bone shaft and distal radius are automatically segmented from hand and wrist radiographs using marker-controlled watershed segmentation and intensity profile. Significant features are selected using statistical analysis and trained using various classifiers to classify healthy subjects and those with low bone mass. The accuracy of the developed prescreening tool can be further improved by using deep learning techniques in combination with handcrafted texture features.
A low cost technique to measure cortical volume of metacarpal bone shaft using 3D reconstruction multi-view hand radiographs is also developed. The 3D reconstruction is done iteratively by 2D registration of projection contours of a template model and X-ray image contours and corresponding deformation of the template model to reconstruct the patient-specific bone model.
Recent publications related to the work:
Areeckal, Anu Shaju, "Diagnosis of Osteoporosis from radiographs using Image processing and Deep learning techniques" (2022). Technical Collection. 22.