Quantification of knee cartilage using automated cartilage segmentation prototype software : Clinical validation on volunteers

Manual cartilage segmentation is a time-consuming post-processing procedure, especially for clinical doctors. Automatic cartilage segmentation frees doctors from tedious computer work. Although there have been various automatic segmentation algorithms, accurate automatic segmentation is still a challenge. In this study, we validated the automatic cartilage segmentation results of the knee joint using a 3D high-resolution DESS sequence. We found that the location of cartilage subregions, and the hydrarthrosis and cartilage degeneration may influence the accuracy of the segmentation. In order to derive more accurate results, a manual finetuning of the automatic segmentation was done. Automatic segmentation still saved considerable time.