A neural network to predict the knee adduction moment in patients with osteoarthritis using anatomical landmarks obtainable from 2D video analysis

被引：43

作者：

Boswell, M. A. ^{[1
]}

Uhlrich, S. D. ^{[2
,6
]}

Kidzinski, L. ^{[1
]}

Thomas, K. ^{[3
]}

Kolesar, J. A. ^{[2
,6
]}

Gold, G. E. ^{[4
]}

Beaupre, G. S. ^{[1
,6
]}

Delp, S. L. ^{[1
,2
,5
]}

机构：

[1] Stanford Univ, Dept Bioengn, Stanford, CA 94305 USA

[2] Stanford Univ, Dept Mech Engn, Stanford, CA 94305 USA

[3] Stanford Univ, Dept Biomed Data Sci, Stanford, CA 94305 USA

[4] Stanford Univ, Dept Radiol, Stanford, CA 94305 USA

[5] Stanford Univ, Dept Orthopaed Surg, Stanford, CA 94305 USA

[6] VA Palo Alto Healthcare Syst, Musculoskeletal Res Lab, Palo Alto, CA USA

来源：

OSTEOARTHRITIS AND CARTILAGE | 2021年 / 29卷 / 03期

基金：

美国国家卫生研究院; 美国国家科学基金会;

关键词：

Knee adduction moment; Osteoarthritis; Gait; Machine learning; Neural network; Video motion analysis; GAIT MODIFICATIONS; DISEASE SEVERITY; TOE-IN; STRATEGIES; REDUCE; LOAD; SELECTION; SUBJECT; PEOPLE; FORCE;

D O I：

10.1016/j.joca.2020.12.017

中图分类号：

R826.8 [整形外科学]; R782.2 [口腔颌面部整形外科学]; R726.2 [小儿整形外科学]; R62 [整形外科学（修复外科学）];

学科分类号：

摘要：

Objective: The knee adduction moment (KAM) can inform treatment of medial knee osteoarthritis; however, measuring the KAM requires an expensive gait analysis laboratory. We evaluated the feasibility of predicting the peak KAM during natural and modified walking patterns using the positions of anatomical landmarks that could be identified from video analysis. Method: Using inverse dynamics, we calculated the KAM for 86 individuals (64 with knee osteoarthritis, 22 without) walking naturally and with foot progression angle modifications. We trained a neural network to predict the peak KAM using the 3-dimensional positions of 13 anatomical landmarks measured with motion capture (3D neural network). We also trained models to predict the peak KAM using 2-dimensional subsets of the dataset to simulate 2-dimensional video analysis (frontal and sagittal plane neural networks). Model performance was evaluated on a held-out, 8-person test set that included steps from all trials. Results: The 3D neural network predicted the peak KAM for all test steps with r(2)( Murray et al., 2012) 2 = 0.78. This model predicted individuals' average peak KAM during natural walking with r(2) ( Murray et al., 2012) 2 = 0.86 and classified which 15 degrees foot progression angle modifications reduced the peak KAM with accuracy = 0.85. The frontal plane neural network predicted peak KAM with similar accuracy (r(2) ( Murray et al., 2012) 2 = 0.85) to the 3D neural network, but the sagittal plane neural network did not (r(2) ( Murray et al., 2012) 2 = 0.14). Conclusion: Using the positions of anatomical landmarks from motion capture, a neural network accurately predicted the peak KAM during natural and modified walking. This study demonstrates the feasibility of measuring the peak KAM using positions obtainable from 2D video analysis. (C) 2021 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.

引用

页码：346 / 356

页数：11

共 55 条

[1]

Aljaaf AJ, 2016, IEEE IJCNN, P5168, DOI 10.1109/IJCNN.2016.7727882

[2] Knee adduction moment and development of chronic knee pain in elders [J].