Reliability and accuracy of segmentation of mandibular condyles from different three-dimensional imaging modalities: a systematic review.

Abstract

To critically synthesize the literature surrounding segmentation of the mandibular condyle using three-dimensional imaging modalities. Specifically, analyzing the reliability and accuracy of methods used for three-dimensional condyle segmentation. Three electronic databases were searched for studies reporting the reliability and accuracy of various methods used to segment mandibular condyles from three-dimensional imaging modalities. Two authors independently reviewed articles for eligibility and data extraction. Nine studies fulfilled the inclusion criteria. Eight studies assessed the condylar segmentation from CBCT images and limited studies were available on non-CBCT three-dimensional imaging modalities. Threshold-based volume segmentation, manual segmentation, and semi-automatic segmentation techniques were presented. Threshold-based volume segmentation reported higher accuracy when completed by an experienced technician compared to clinicians. Adequate reliability and accuracy were observed in manual segmentation. Although adequate reliability was reported in semi-automatic segmentation, data on its accuracy were lacking. A definitive conclusion with regards to which current technique is most reliable and accurate to efficiently segment the mandibular condyle cannot be made with the currently available evidence. This is especially true in terms of non-CBCT imaging modalities with very limited literature available.