An Interactive System for Detection and Analysis of Mandibular Cystic Lesions

Mandibular cystic lesion is a common disease in cranio-maxillofacial surgery with a high recurrence rate. Regular follow-up, early detection and diagnosis are very important in preventing the relapse and giving an active and effective treatment timely. In clinical, the patients after the surgery will be followed-up for 3 - 24 months, and the cone beam computed tomography (CBCT) images will be collected by the physician to observe the morphological changes of cysts. However, the cystic lesion is a small volume in CBCT images, the observation of the lesions is laborious and it will take a certain amount of time to find and analyze the recovery of the cystic lesion at different period. This paper presents an interactive system for automatically detecting and analyzing the cystic lesions of mandible from CBCT images. Firstly, the system integrated an improved mutual information(MI) based image registration method to register the CBCT images collected at different time. Then, an automatic procedure is designed to detect the lesion area, and the features are calculated for doctors to make a better diagnosis. Experimental results demonstrate that our system performs well and has great potential value in clinical practice.