Background: Hand osteoarthritis (OA) is a common condition with a lifetime risk of symptomatic hand OA of 40% [1]. Hand OA is a heterogeneous disease with ultrasound findings as osteophytes, joint effusion, synovial hypertrophy, inflammation, and joint space narrowing [2]. Artificial intelligence (AI) models assessing cartilage thickness and joint inflammation have been developed [3-4]. However, within the field of AI models for osteophyte assessment we did not find any published work. A semiquantitative grading system from 0 to 3 called the EULAR-OMERACT grading system (EOGS) is validated to describe the severity of osteophytes in hand OA [5-7].

Objectives: The aim of this study was to develop an AI model able to perform both segmentation of hand joint ultrasound images for osteophytes, bone, and synovium, and perform osteophyte severity scoring following the EOGS for osteophytes.

Methods: One hundred sixty patients with hand pain or reduced hand function were included during planned outpatient visits in a hospital section of rheumatology. Ultrasound images of the metacarpophalangeal (MCP), proximal interphalangeal (PIP), distal interphalangeal (DIP), and first carpometacarpal (CMC1) joints were manually segmented for bone, synovium and osteophytes and scored from 0 to 3 according to the EOGS for osteophytes. The dataset was divided into a training, validation, and test set. An AI model was trained on the training set to perform identification of bone, synovium, and osteophytes. Based on the image segmentation and grading by an experienced rheumatologist, the AI model was trained to classify the severity of osteophytes according to EOGS. Percent Exact Agreement (PEA) and Percent Close Agreement (PCA) were assessed on individual joints and overall. PCA allows a difference of one EOGS grade between doctor assessment and AI.

Results: A total of 4615 ultrasound images were used for AI development and testing. The results of the AI model on the test set for the MCP joints was a PEA of 76% and PCA of 97%; for PIP, a PEA of 70% and PCA of 97%; for DIP, a PEA of 59% and PCA of 94%, and CMC a PEA of 50% and PCA of 82%. Combining all joints, we found a PEA between AI and doctor assessments of 68% and a PCA of 95%.

Conclusion: We demonstrate a PEA between AI and experts slightly higher than what have previously been shown between experts [5, 6]. PEA between experts was 54.2% and 61%, respectively, while PEA in this study was 68.1% in the test set. This suggests that the developed AI model is a success as a proof of concept. In the future, a unifying AI model combining assessment of multiple hand joint conditions would be a marked improvement. This could open up for a much more detailed understanding of the very heterogenous disease hand OA and how these factors interact, and change over time.

REFERENCES: [1] Qin, J., et al., Lifetime Risk of Symptomatic Hand Osteoarthritis: The Johnston County Osteoarthritis Project. Arthritis Rheumatol, 2017. 69 (6): p. 1204-1212.

[2] Kloppenburg, M. and W.Y. Kwok, Hand osteoarthritis--a heterogeneous disorder. Nat Rev Rheumatol, 2011. 8 (1): p. 22-31.

[3] Mathiessen, A., et al., Ultrasonography of Inflammatory and Structural Lesions in Hand Osteoarthritis : An Outcome Measures in Rheumatology Agreement and Reliability Study. Arthritis Care & Research, 2022. 74 (12): p. 2005-2012.

[4] Christensen, A.B.H., et al., Applying cascaded convolutional neural network design further enhances automatic scoring of arthritis disease activity on ultrasound images from rheumatoid arthritis patients. Ann Rheum Dis, 2020. 79 (9): p. 1189-1193.

[5] Keen, H.I., et al., The development of a preliminary ultrasonographic scoring system for features of hand osteoarthritis. Ann Rheum Dis, 2008. 67 (5): p. 651-5.

[6] Hammer, H.B., et al., Global ultrasound assessment of structural lesions in osteoarthritis: a reliability study by the OMERACT ultrasonography group on scoring cartilage and osteophytes in finger joints. Ann Rheum Dis, 2016. 75 (2): p. 402-7.

[7] Fiorentino, M.C., et al., A deep-learning framework for metacarpal-head cartilage-thickness estimation in ultrasound rheumatological images. Comput Biol Med, 2022. 141 : p. 105117.

Acknowledgements: NIL.

Disclosure of Interests: Benjamin Schultz Overgaard: None declared, Anders Bossel Holst Christensen Full time employee at ROPCA., Lene Terslev Janssen, Novartis, Pfizer, eli Lilly, Thiusius Rajeeth Savarimuthu Cofounder of ROPCA ApS., Søren Andreas Just Cofounder of ROPCA ApS.