Work

Magnetic resonance imaging semi-quantitative evaluation of cartilage damage in patients with knee osteoarthritis

Year: 2026

Type: article

Abstract: This study aimed to develop and validate an improved U-Net model for optimizing knee MRI reconstruction and used these optimized images to perform semi-quantitative assessments of cartilage damage severity in patients with knee osteoarthritis (KOA), as well as to explore the association between cartilage damage and subchondral bone marrow edema. The improved U-shaped network (U-net) was utilized for optimizing the MRI images, and the network reconstruction time, structural similarity index measure (SSIM), and peak signal-to-noise ratio (PSNR) were analyzed. 108 patients with KOA who underwent magnetic resonance examination were included as the research objects. After the patients were divided into normal, mild, and severe groups, their grades of semi-quantitative evaluation of cartilage damage were analyzed. The SSIM of the improved U-net was significantly higher than that of the original U-net, and the difference between the reconstructed images and the original images was highly reduced. MRI could show the types of cartilage damage in the knee joint. An MRI system with the improved U-net model improved the quality of MRI images of KOA patients. The improved U-net achieved a mean PSNR of 42.7 dB and SSIM of 0.94 at a 20 % sampling rate, representing a 15 % and 10 % improvement over the standard U-net, respectively. After the improvement, the reconstruction time of the U-net became shorter. There was a statistically significant difference in the probability of cartilage injury grades I–IV occurring among subchondral bone marrow edema grades 1–3 ( Z = 5.098, P < 0.05). The semi-quantitative MRI assessment facilitated by the improved U-net shows promise for accurately displaying and grading articular cartilage damage in KOA. Further validation against arthroscopic findings and in multi-center cohorts is warranted before clinical adoption can be considered. In conclusion, the improved U-net provided knee MRI reconstructions with superior quantitative quality and speed. Using these images, semi-quantitative assessment revealed a significant association between the severity of subchondral bone marrow edema and cartilage injury grade, demonstrating the method's potential for structured image evaluation in KOA.

Source: Journal of Radiation Research and Applied Sciences

DOI: 10.1016/j.jrras.2026.102173