Segmentation of lumbar intervertebral discs from spine MR images

Degeneration of the intervertebral disc is a very common issue these days and results in protracted and persistent back pain, and finally in most of the spine surgical procedures. MRI is the modality of preference, for diagnosing these disc degeneration issues. In this paper, both normal and degenerated lumbar intervertebral discs (T2-weighted sagittal MR images of the spine) are being considered for a fast and robust unsupervised watershed segmentation algorithm. This task is very much challenging because of: partial volume effects and overlapping gray-level values between neighboring tissue classes. In order to overcome these problems preprocessing has been done in which normalization of the MRI images is done by performing CLAHE operation followed by de-noising by morphological opening and filling of the holes and then applying segmentation algorithm to the MR spine images. Also we have implemented a level set segmentation for the same data base. Level set segmentation is a robust technique but very much time consuming compared to the watershed algorithm used.