| Abstract [eng] |
Temporomandibular joint is joint is the most complicated joints in the human body. Temporomandibular joint is composed of several parts, which interacts with each other and performs a rotary motion, which allows the lower jaw movement. Tempormandibular joint disorders are the second most frequent causes of oral and facial pain. The main causes of joint dysfunction is associated with joint disc pathologies due to the jaw joint rotation function is disrupted. The diagnosis of temporomandibular joint disorders is complex, because the pain is often mistakenly associated with the dental, mandible pain. Today it is possible to obtain images of internal organs without surgical intervention thanks to technological advances in non-invasive diagnostics. NMR tomography images are considered to be the gold standard of temporomandibular joint images. Based on this data, it is possible to perform segmentation and get joint model, which helps to diagnose a pathology. The main problem that joint is so small, thats why the segmentation must be carried out with high precision, and small noise gives high distortion level. The main goals of this work are to reduce noise and carry out joint‘s segmentation.The results allow to conclude that filters designed in frequency domain give the best results in noise reduction and make the sharpest boundary between an object and a background. Poor segmentation results allow to conclude that the Otsu method is not suitable for segmentation. |
