Gary H Chang1, David T Felson2,3, Shangran Qiu1, Ali Guermazi4, Terence D Capellini5,6, Vijaya B Kolachalama7,8,9,10. 1. Section of Computational Biomedicine, Department of Medicine, Boston University School of Medicine, 72 E. Concord Street, Evans 636, Boston, MA, 02118, USA. 2. Section of Rheumatology, Department of Medicine, Boston University School of Medicine, Boston, MA, 02118, USA. 3. Centre for Epidemiology, University of Manchester and the NIHR Manchester BRC, Manchester University, NHS Trust, Manchester, UK. 4. Department of Radiology, Boston University School of Medicine, Boston, MA, 02118, USA. 5. Department of Human Evolutionary Biology, Harvard University, Cambridge, MA, 02138, USA. 6. Broad Institute of MIT and Harvard, Cambridge, MA, 02142, USA. 7. Section of Computational Biomedicine, Department of Medicine, Boston University School of Medicine, 72 E. Concord Street, Evans 636, Boston, MA, 02118, USA. vkola@bu.edu. 8. Whitaker Cardiovascular Institute, Boston University School of Medicine, Boston, MA, 02118, USA. vkola@bu.edu. 9. Hariri Institute for Computing and Computational Science and Engineering, Boston University, Boston, MA, 02215, USA. vkola@bu.edu. 10. Boston University Alzheimer's Disease Center, Boston, MA, 02118, USA. vkola@bu.edu.
Abstract
OBJECTIVES: It remains difficult to characterize the source of pain in knee joints either using radiographs or magnetic resonance imaging (MRI). We sought to determine if advanced machine learning methods such as deep neural networks could distinguish knees with pain from those without it and identify the structural features that are associated with knee pain. METHODS: We constructed a convolutional Siamese network to associate MRI scans obtained on subjects from the Osteoarthritis Initiative (OAI) with frequent unilateral knee pain comparing the knee with frequent pain to the contralateral knee without pain. The Siamese network architecture enabled pairwise learning of information from two-dimensional (2D) sagittal intermediate-weighted turbo spin echo slices obtained from similar locations on both knees. Class activation mapping (CAM) was utilized to create saliency maps, which highlighted the regions most associated with knee pain. The MRI scans and the CAMs of each subject were reviewed by an expert radiologist to identify the presence of abnormalities within the model-predicted regions of high association. RESULTS: Using 10-fold cross-validation, our model achieved an area under curve (AUC) value of 0.808. When individuals whose knee WOMAC pain scores were not discordant were excluded, model performance increased to 0.853. The radiologist review revealed that about 86% of the cases that were predicted correctly had effusion-synovitis within the regions that were most associated with pain. CONCLUSIONS: This study demonstrates a proof of principle that deep learning can be applied to assess knee pain from MRI scans. KEY POINTS: • Our article is the first to leverage a deep learning framework to associate MR images of the knee with knee pain. • We developed a convolutional Siamese network that had the ability to fuse information from multiple two-dimensional (2D) MRI slices from the knee with pain and the contralateral knee of the same individual without pain to predict unilateral knee pain. • Our model achieved an area under curve (AUC) value of 0.808. When individuals who had WOMAC pain scores that were not discordant for knees (pain discordance < 3) were excluded, model performance increased to 0.853.
OBJECTIVES: It remains difficult to characterize the source of pain in knee joints either using radiographs or magnetic resonance imaging (MRI). We sought to determine if advanced machine learning methods such as deep neural networks could distinguish knees with pain from those without it and identify the structural features that are associated with knee pain. METHODS: We constructed a convolutional Siamese network to associate MRI scans obtained on subjects from the Osteoarthritis Initiative (OAI) with frequent unilateral knee pain comparing the knee with frequent pain to the contralateral knee without pain. The Siamese network architecture enabled pairwise learning of information from two-dimensional (2D) sagittal intermediate-weighted turbo spin echo slices obtained from similar locations on both knees. Class activation mapping (CAM) was utilized to create saliency maps, which highlighted the regions most associated with knee pain. The MRI scans and the CAMs of each subject were reviewed by an expert radiologist to identify the presence of abnormalities within the model-predicted regions of high association. RESULTS: Using 10-fold cross-validation, our model achieved an area under curve (AUC) value of 0.808. When individuals whose knee WOMACpain scores were not discordant were excluded, model performance increased to 0.853. The radiologist review revealed that about 86% of the cases that were predicted correctly had effusion-synovitis within the regions that were most associated with pain. CONCLUSIONS: This study demonstrates a proof of principle that deep learning can be applied to assess knee pain from MRI scans. KEY POINTS: • Our article is the first to leverage a deep learning framework to associate MR images of the knee with knee pain. • We developed a convolutional Siamese network that had the ability to fuse information from multiple two-dimensional (2D) MRI slices from the knee with pain and the contralateral knee of the same individual without pain to predict unilateral knee pain. • Our model achieved an area under curve (AUC) value of 0.808. When individuals who had WOMACpain scores that were not discordant for knees (pain discordance < 3) were excluded, model performance increased to 0.853.
Entities:
Keywords:
Knee; Machine learning; Magnetic resonance imaging; Osteoarthritis; Pain
Authors: M K Javaid; A Kiran; A Guermazi; C K Kwoh; S Zaim; L Carbone; T Harris; C E McCulloch; N K Arden; N E Lane; D Felson; M Nevitt Journal: Arthritis Rheum Date: 2012-10
Authors: Theodoros Dimitroulas; Rui V Duarte; Asis Behura; George D Kitas; Jon H Raphael Journal: Semin Arthritis Rheum Date: 2014-05-14 Impact factor: 5.532
Authors: J Cibere; E C Sayre; A Guermazi; S Nicolaou; J A Kopec; J M Esdaile; A Thorne; J Singer; H Wong Journal: Osteoarthritis Cartilage Date: 2011-02-15 Impact factor: 6.576
Authors: Erlangga Yusuf; Marion C Kortekaas; Iain Watt; Tom W J Huizinga; Margreet Kloppenburg Journal: Ann Rheum Dis Date: 2010-09-09 Impact factor: 19.103
Authors: M F Sowers; C Hayes; D Jamadar; D Capul; L Lachance; M Jannausch; G Welch Journal: Osteoarthritis Cartilage Date: 2003-06 Impact factor: 6.576
Authors: M Englund; J Niu; A Guermazi; F W Roemer; D J Hunter; J A Lynch; C E Lewis; J Torner; M C Nevitt; Y Q Zhang; D T Felson Journal: Arthritis Rheum Date: 2007-12
Authors: Kai Roman Laukamp; Frank Thiele; Georgy Shakirin; David Zopfs; Andrea Faymonville; Marco Timmer; David Maintz; Michael Perkuhn; Jan Borggrefe Journal: Eur Radiol Date: 2018-06-25 Impact factor: 5.315
Authors: Gary H Chang; Lisa K Park; Nina A Le; Ray S Jhun; Tejus Surendran; Joseph Lai; Hojoon Seo; Nuwapa Promchotichai; Grace Yoon; Jonathan Scalera; Terence D Capellini; David T Felson; Vijaya B Kolachalama Journal: Arthritis Rheumatol Date: 2021-10-29 Impact factor: 10.995
Authors: Yi Zheng; Clarissa A Cassol; Saemi Jung; Divya Veerapaneni; Vipul C Chitalia; Kevin Y M Ren; Shubha S Bellur; Peter Boor; Laura M Barisoni; Sushrut S Waikar; Margrit Betke; Vijaya B Kolachalama Journal: Am J Pathol Date: 2021-05-23 Impact factor: 5.770