Semixup: In- and Out-of-Manifold Regularization for Deep Semi-Supervised Knee Osteoarthritis Severity Grading From Plain Radiographs
Open Access
- 17 August 2020
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Medical Imaging
- Vol. 39 (12) , 4346-4356
Abstract
Knee osteoarthritis (OA) is one of the highest disability factors in the world. This musculoskeletal disorder is assessed from clinical symptoms, and typically confirmed via radiographic assessment. This visual assessment done by a radiologist requires experience, and suffers from moderate to high inter-observer variability. The recent literature has shown that deep learning methods can reliably perform the OA severity assessment according to the gold standard Kellgren-Lawrence (KL) grading system. However, these methods require large amounts of labeled data, which are costly to obtain. In this study, we propose the Semixup algorithm, a semi-supervised learning (SSL) approach to leverage unlabeled data. Semixup relies on consistency regularization using in- and out-of-manifold samples, together with interpolated consistency. On an independent test set, our method significantly outperformed other state-of-the-art SSL methods in most cases. Finally, when compared to a well-tuned fully supervised baseline that yielded a balanced accuracy (BA) of 70.9 ± 0.8% on the test set, Semixup had comparable performance – BA of 71± 0.8% (p = 0.368) while requiring 6 times less labeled data. These results show that our proposed SSL method allows building fully automatic OA severity assessment tools with datasets that are available outside research settings.Keywords
All Related Versions
Funding Information
- strategic funding of the University of Oulu
- KAUTE Foundation, Finland
- Sigrid Juselius Foundation, Finland
- OAI, a public-private partnership
- National Institutes of Health, a branch of the Department of Health and Human Services, and conducted by the OAI Study Investigators (N01-AR-2-2258, N01-AR-2-2259, N01-AR-2-2260, N01-AR-2-2261, N01-AR-2-2262)
- MOST
- National Institutes of Health, a branch of the Department of Health and Human Services, and conducted by MOST study investigators (AG18820, AG18832, AG18947, AG19069)
- MerckResearch Laboratories
- Novartis Pharmaceuticals Corporation
- GlaxoSmithKline
- Pfizer, Inc.
This publication has 37 references indexed in Scilit:
- Influence of Body Mass Index (BMI) on Functional Improvements at 3 Years Following Total Knee Replacement: A Retrospective Cohort StudyPLOS ONE, 2013
- Knee replacementThe Lancet, 2012
- Obesity, physically demanding work and traumatic knee injury are major risk factors for knee osteoarthritis--a population-based study with a follow-up of 22 yearsRheumatology, 2009
- Semi-Supervised Learning (Chapelle, O. et al., Eds.; 2006) [Book reviews]IEEE Transactions on Neural Networks, 2009
- Prevalence of bone attrition on knee radiographs and MRI in a community-based cohortOsteoarthritis and Cartilage, 2008
- Atlas of individual radiographic features in osteoarthritis, revisedOsteoarthritis and Cartilage, 2007
- Pathogenesis and management of pain in osteoarthritisThe Lancet, 2005
- Modifiers of change in physical functioning in older adults with knee pain: the Observational Arthritis Study in Seniors (OASIS)Arthritis & Rheumatism, 2001
- Approximate Inference in Generalized Linear Mixed ModelsJournal of the American Statistical Association, 1993
- A Coefficient of Agreement for Nominal ScalesEducational and Psychological Measurement, 1960