LAUSR

Menisci are structures that directly affect movement, so early detection of meniscus tears also helps to prevent progressive knee disorders such as osteoarthritis. Manual segmentation of the menisci and diagnosis of the meniscal tear is a costly process in terms of time and effort for a radiologist. The aim of this study is to automatically determine the location and the type of meniscal tears that are important in the diagnosis and effective treatment of this problem. For this purpose, 29 different MR images, which were provided by Osteoarthritis Initiative (OAI), were used in the study. This study proposes a novel three-stage (preprocessing, segmentation and classification) method for fully automated classification from MR images, and shows the performance of each stage separately. At the preprocessing step, the most compact rectangular windows for the menisci were obtained from MR slices. At the segmentation step, the menisci were segmented using fuzzy clustering methods. In order to classify the segmented images and to determine meniscus tears, three different classifiers were used. The method first decides whether there are tears on menisci; if this is the case then, determines the place and type of the tears. There are no studies that classify the meniscus tears according to their types up to now in the literature. The experimental results indicate that the automated process can be completed within a time range of 3 to 4 min with a high classification performance. Hence, the suggested computer-aided diagnosis (CAD) system can be used as a decision support system for the diagnosis of meniscal tears by radiologists.