Meeting Banner
Abstract #2970

3D Texture Analysis of 3D DESS Cartilage Images for Prediction of Knee Osteoarthritis

Daniel Uher1, Ari Väärälä1, Antti Isosalo1, Victor Casula1,2, and Miika T. Nieminen1,2,3
1Research Unit of Medical Imaging, Physics and Technology, University of Oulu, Oulu, Finland, 2Medical Research Center, University of Oulu and Oulu University Hospital, Oulu, Finland, 3Department of Diagnostic Radiology, Oulu University Hospital, Oulu, Finland

In this study, a gray level co-occurrence matrix (GLCM) based 3D Texture Analysis method was utilized for early prediction of knee osteoarthritis using 3D DESS images. Twenty subjects were extracted from the Osteoarthritis Initiative (baseline) with Kellgren-Lawrence (KL) score = 0 at baseline. Ten of the selected subjects developed the disease and showed KL ≥ 2 at the 36-month visit. Knee DESS images were analyzed using various quantization schemes and three machine learning models were trained based on the output GLCM features. Naïve Bayes model trained on tibial features showed the highest accuracy (86.8%) for OA onset after 36 months.

This abstract and the presentation materials are available to members only; a login is required.

Join Here