Integrating Radiomics and Neural Networks for Knee Osteoarthritis Incidence Prediction

Shengfa Li; Peihua Cao; Jia Li; Tianyu Chen; Ping Luo; Guangfeng Ruan; Yan Zhang; Xiaoshuai Wang; Weiyu Han; Zhaohua Zhu; Qin Dang; Qianyi Wang; Mengdi Zhang; Qiushun Bai; Zhiyi Chai; Hao Yang; Haowei Chen; Mingze Tang; Arafat Akbar; Alexander Tack; David J Hunter; Changhai Ding

doi:10.1002/art.42915

Integrating Radiomics and Neural Networks for Knee Osteoarthritis Incidence Prediction

Arthritis Rheumatol. 2024 May 15. doi: 10.1002/art.42915. Online ahead of print.

Authors

Shengfa Li^#^{1

2

3}, Peihua Cao^#¹, Jia Li⁴, Tianyu Chen⁵, Ping Luo^{1

6}, Guangfeng Ruan⁷, Yan Zhang¹, Xiaoshuai Wang¹, Weiyu Han², Zhaohua Zhu¹, Qin Dang¹, Qianyi Wang¹, Mengdi Zhang^{1

2}, Qiushun Bai⁸, Zhiyi Chai^{1

2}, Hao Yang^{1

2}, Haowei Chen^{1

2}, Mingze Tang^{1

2}, Arafat Akbar^{1

2}, Alexander Tack⁹, David J Hunter^{1

10}, Changhai Ding^{1

7

11

12}

Affiliations

¹ Clinical Research Centre, Zhujiang Hospital, Southern Medical University, Guangzhou, Guangdong, China.
² Department of Orthopedics, Zhujiang Hospital, Southern Medical University, Guangzhou, Guangdong, China.
³ Department of Orthopaedics, The Third People's Hospital of Chengdu, Affiliated Hospital of Southwest Jiaotong University, The Second Affiliated Chengdu Hospital of Chongqing Medical University, Chengdu, Sichuan, China.
⁴ Division of Orthopaedic Surgery, Department of Orthopedics, Nanfang Hospital, Southern Medical University, Guangzhou, Guangdong, China.
⁵ Department of Orthopedics, The Third Affiliated Hospital of Southern Medical University, Guangzhou, China.
⁶ Department of Spinal Surgery, The Fourth Hospital of Changsha, Changsha Hospital of Hunan Normal University, Changsha, Hunan, China.
⁷ Clinical Research Centre, Guangzhou First People's Hospital, School of Medicine, South China University of Technology, Guangzhou, China.
⁸ School of Biomedical Engineering, Southern Medical University, Guangzhou, China.
⁹ Zuse Institute Berlin, Berlin, Germany.
¹⁰ Department of Rheumatology, Royal North Shore Hospital and Sydney Musculoskeletal Health, Kolling Institute, University of Sydney, Australia.
¹¹ Menzies Institute for Medical Research, University of Tasmania, Hobart, Tasmania, Australia.
¹² Guangdong Provincial Key Laboratory of Bone and Joint Degenerative Diseases, Southern Medical University, Guangzhou, Guangdong, China.

^# Contributed equally.

PMID: 38751101
DOI: 10.1002/art.42915

Abstract

Objective: Accurately predicting knee osteoarthritis (KOA) is essential for early detection and personalized treatment. We aimed to develop and test an MRI-based Joint Space Radiomic Model (JS-RM) to predict radiographic KOA incidence through neural networks by integrating meniscus and femorotibial cartilage radiomic features.

Methods: In the Osteoarthritis Initiative cohort, knees without radiographic KOA at baseline but at high risk for radiographic KOA were included. Case knees developed radiographic KOA whereas control knees did not over 4-year. We randomly split the knees into development and test cohorts (D/T=8/2) and extracted features from baseline 3D-DESS-sequence MRI. Model performance was evaluated using an area under the receiver operating characteristic curve (AUC), sensitivity, and specificity in both cohorts. Nine resident surgeons performed the reader experiment without/with the JS-RM aid.

Results: Our study included 549 knees in the development cohort (275 cases vs. 274 controls) and 137 knees in the test cohort (68 cases vs. 69 controls). In the test cohort, JS-RM had a favorable accuracy for predicting the radiographic KOA incidence with an AUC of 0.931 (95%CI: 0.876-0.963), a sensitivity of 84.4% (95%CI: 83.9%-84.9%), and a specificity of 85.6% (95%CI: 85.2%-86.0%). The mean specificity and sensitivity of resident surgeons through MRI reading in predicting radiographic KOA incidence were increased from 0.474 (95%CI: 0.333-0.614) and 0.586 (95%CI: 0.429-0.743) without the assistance of JS-RM to 0.874 (95%CI: 0.847-0.901) and 0.812 (95%CI: 0.742-0.881) with JS-RM assistance, respectively (p<.001).

Conclusion: JS-RM integrating the features of the meniscus and cartilage showed improved predictive values in radiographic KOA incidence.