Artificial intelligence can be used in the identification and classification of shoulder osteoarthritis and avascular necrosis on plain radiographs: a training study of 7,139 radiograph sets.

Publisher: Medical Journals Sweden AB Country of Publication: Sweden NLM ID: 101231512 Publication Model: Electronic Cited Medium: Internet ISSN: 1745-3682 (Electronic) Linking ISSN: 17453674 NLM ISO Abbreviation: Acta Orthop Subsets: MEDLINE

Publication: 2022- : Uppsala, Sweden : Medical Journals Sweden AB
Original Publication: Basingstoke, Hampshire, UK : Taylor & Francis, c2005-

Osteoarthritis*/diagnostic imaging ; Osteoarthritis*/classification ; Osteonecrosis*/diagnostic imaging ; Osteonecrosis*/classification ; Shoulder Joint*/diagnostic imaging ; Radiography*; Humans ; Male ; Artificial Intelligence ; Female ; Deep Learning ; Middle Aged ; Aged ; Sensitivity and Specificity ; Adult

Background and Purpose: Knowledge concerning the use AI models for the classification of glenohumeral osteoarthritis (GHOA) and avascular necrosis (AVN) of the humeral head is lacking. We aimed to analyze how a deep learning (DL) model trained to identify and grade GHOA on plain radiographs performs. Our secondary aim was to train a DL model to identify and grade AVN on plain radiographs.
Patients and Methods: A modified ResNet-type network was trained on a dataset of radiographic shoulder examinations from a large tertiary hospital. A total of 7,139 radiographs were included. The dataset included various projections of the shoulder, and the network was trained using stochastic gradient descent. Performance evaluation metrics, area under the receiver operating characteristic curve (AUC), sensitivity, and specificity were used to assess the network's performance for each outcome.
Results: The network demonstrated AUC values ranging from 0.73 to 0.93 for GHOA classification and > 0.90 for all AVN classification classes. The network exhibited lower AUC for mild cases compared with definitive cases of GHOA. When none and mild grades were combined, the AUC increased, suggesting difficulties in distinguishing between these 2 grades.
Conclusion: We found that a DL model can be trained to identify and grade GHOA on plain radiographs. Furthermore, we show that a DL model can identify and grade AVN on plain radiographs. The network performed well, particularly for definitive cases of GHOA and any level of AVN. However, challenges remain in distinguishing between none and mild GHOA grades.

Acta Orthop Scand. 2003 Apr;74(2):186-9. (PMID: 12807327)
Nature. 2015 May 28;521(7553):436-44. (PMID: 26017442)
Radiology. 2020 Apr;295(1):136-145. (PMID: 32013791)
Radiol Artif Intell. 2020 Mar 25;2(2):e200029. (PMID: 33937821)
Curr Opin Rheumatol. 2005 Sep;17(5):634-40. (PMID: 16093845)
Caspian J Intern Med. 2013 Spring;4(2):627-35. (PMID: 24009950)
J Am Acad Orthop Surg. 2009 Jun;17(6):345-55. (PMID: 19474444)
Semin Musculoskelet Radiol. 2015 Jul;19(3):307-18. (PMID: 26021591)
J Clin Med. 2020 Oct 18;9(10):. (PMID: 33080993)
Knee Surg Sports Traumatol Arthrosc. 2017 Oct;25(10):3270-3278. (PMID: 27198139)
J Bone Joint Surg Br. 1976 Aug;58(3):313-7. (PMID: 956247)
J Bone Joint Surg Am. 1998 Jun;80(6):841-52. (PMID: 9655102)
Cancer. 1950 Jan;3(1):32-5. (PMID: 15405679)
Acta Orthop. 2024 Mar 21;95:152-156. (PMID: 38597205)
Phys Med Rehabil Clin N Am. 2004 May;15(2):373-406. (PMID: 15145423)
Skeletal Radiol. 2022 Feb;51(2):355-362. (PMID: 33611622)
J Shoulder Elbow Surg. 2013 Aug;22(8):1063-7. (PMID: 23375877)
Disabil Rehabil. 2017 Aug;39(16):1674-1682. (PMID: 27416338)
AJR Am J Roentgenol. 2019 Jul;213(1):155-162. (PMID: 30917021)
J Bone Joint Surg Am. 1983 Apr;65(4):456-60. (PMID: 6833319)
J Clin Med. 2022 May 08;11(9):. (PMID: 35566766)

Date Created: 20240617 Date Completed: 20240617 Latest Revision: 20240823

20250114

PMC11182033

10.2340/17453674.2024.40905

38884536