Segmentation of Temporomandibular Condyle in MR images using Marker-Controlled Watershed and Random Walks

Abstract

One of the most complex joints in the human body is the temporomandibular joint (TMJ), which connects the mandible and the skull's temporal bone. The two main structures of TMJ are mandibular condyles and the articular disc. Around 28% of the population is affected by TMJ dysfunction. The leading cause of TMJ dysfunction is due to internal derangement of the condyle and disc. In current scenarios, magnetic resonance imaging (MRI) is an imaging technique used in diagnosing TMJ dysfunction. The clinician must visually investigate the derangement of condyle with a disc which can lead to subjective error. This study focuses on segmenting the TMJ condyle using two image processing techniques like marker-controlled watershed segmentation and Random walks. These techniques segment the TMJ from MR images in sagittal orientation, even when the MR images are corrupted due to noise.