Abstract:
In the recent years, Cervical Osteophytes/ Neck Bone Spurs have been observed much frequently in elderly people. According to the medical survey reports, people of age more than 60 years are more often to suffer from this abnormality. Due to the aging factor, it’s a very common phenomenon that the bone growth of the cervical vertebrae may build up new bone cells resulting in bone projections called Bone Spurs or Osteophytes. Thus, there had been a great interest in clinical research to detect and analyze the shape of cervical vertebrae for effective cervical spine assessment. This thesis proposes a novel approach for segmentation of cervical vertebras from a video fluoroscopic data using a variational level sets technique. In the past, level sets have been widely used for the segmentation of target objects along with their shape information. In most methods the level set curve evolves towards the boundary of the object using either region based information or edge based information. However, in many scenarios region based information is non-existent due to low intensity and poor contrast, so the evolution of level set from an arbitrary initial curve becomes very challenging task just like in our case. Thus, we propose a novel segmentation method that uses local intensity clustering property as shape estimate of the initial curve and then this curve evolve towards the nearest boundary edge. Experimental results shows that our proposed method is more robust to initialization and showed more accurate results than other level set approaches.
