A Novel Method for Detecting Bone Contours in Hand Radiographic Images

Rheumatoid arthritis (RA) is a disorder that causes pain and swelling in the hand and feet joints. From X-ray images, the exact eroded part of bone may not be visible. To overcome this, an Adaptive Snake Algorithm without edge detection is implemented to detect bone contours from hand radiographs. Without prior knowledge and locating initial contours for each bone specifically can be evaluated from both inside and outside regions. Morphological operations are done for binary images to get a better result. Techniques used for Active contour model are Neumann boundary condition and Gradient method (Curvature Center method). Subsequently, another algorithm named Multiple Kernel Fuzzy C-Means (MKFC) is also implemented to detect the eroded part of the bone.