Machine learning for abdominal aortic calcification assessment from bone density machine-derived lateral spine imagesResearch in context
Naeha Sharif,
Syed Zulqarnain Gilani,
David Suter,
Siobhan Reid,
Pawel Szulc,
Douglas Kimelman,
Barret A. Monchka,
Mohammad Jafari Jozani,
Jonathan M. Hodgson,
Marc Sim,
Kun Zhu,
Nicholas C. Harvey,
Douglas P. Kiel,
Richard L. Prince,
John T. Schousboe,
William D. Leslie,
Joshua R. Lewis
Affiliations
Naeha Sharif
Nutrition & Health Innovation Research Institute, Edith Cowan University, Perth, Australia; Centre for AI&ML, School of Science, Edith Cowan University, Perth, Australia; Department of Computer Science and Software Engineering, The University of Western Australia, Perth, Australia
Syed Zulqarnain Gilani
Nutrition & Health Innovation Research Institute, Edith Cowan University, Perth, Australia; Centre for AI&ML, School of Science, Edith Cowan University, Perth, Australia
David Suter
Nutrition & Health Innovation Research Institute, Edith Cowan University, Perth, Australia; Centre for AI&ML, School of Science, Edith Cowan University, Perth, Australia
Siobhan Reid
Department of Electrical and Computer Engineering, University of Manitoba, Winnipeg, Canada
Pawel Szulc
INSERM UMR 1033, University of Lyon, Hospices Civils de Lyon, Lyon, France
Douglas Kimelman
Department of Radiology, Rady Faculty of Health Sciences, University of Manitoba, Winnipeg, Canada
Barret A. Monchka
George and Fay Yee Centre for Healthcare Innovation, University of Manitoba, Winnipeg, Canada
Mohammad Jafari Jozani
Department of Statistics, University of Manitoba, Winnipeg, Canada
Jonathan M. Hodgson
Nutrition & Health Innovation Research Institute, Edith Cowan University, Perth, Australia; Medical School, The University of Western Australia, Perth, Australia
Marc Sim
Nutrition & Health Innovation Research Institute, Edith Cowan University, Perth, Australia; Medical School, The University of Western Australia, Perth, Australia
Kun Zhu
Medical School, The University of Western Australia, Perth, Australia; Department of Endocrinology and Diabetes, Sir Charles Gairdner Hospital, Perth, Australia
Nicholas C. Harvey
MRC Lifecourse Epidemiology Centre, University of Southampton, Southampton, United Kingdom; NIHR Southampton Biomedical Research Centre, University of Southampton and University Hospital Southampton NHS Foundation Trust, Southampton, United Kingdom
Douglas P. Kiel
Hinda and Arthur Marcus Institute for Aging Research, Hebrew Senior Life, Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA
Richard L. Prince
Nutrition & Health Innovation Research Institute, Edith Cowan University, Perth, Australia; Medical School, The University of Western Australia, Perth, Australia
John T. Schousboe
Park Nicollet Clinic and HealthPartners Institute, HealthPartners, Minneapolis, USA; Division of Health Policy and Management, University of Minnesota, Minneapolis, USA
William D. Leslie
Departments of Medicine and Radiology, University of Manitoba, Winnipeg, Canada
Joshua R. Lewis
Nutrition & Health Innovation Research Institute, Edith Cowan University, Perth, Australia; Medical School, The University of Western Australia, Perth, Australia; Centre for Kidney Research, Children's Hospital at Westmead School of Public Health, Sydney Medical School, the University of Sydney, Sydney, Australia; Corresponding author. Nutrition & Health Innovation Research Institute, Edith Cowan University, Perth, Australia.
Summary: Background: Lateral spine images for vertebral fracture assessment can be easily obtained on modern bone density machines. Abdominal aortic calcification (AAC) can be scored on these images by trained imaging specialists to assess cardiovascular disease risk. However, this process is laborious and requires careful training. Methods: Training and testing of model performance of the convolutional neural network (CNN) algorithm for automated AAC-24 scoring utilised 5012 lateral spine images (2 manufacturers, 4 models of bone density machines), with trained imaging specialist AAC scores. Validation occurred in a registry-based cohort study of 8565 older men and women with images captured as part of routine clinical practice for fracture risk assessment. Cox proportional hazards models were used to estimate the association between machine-learning AAC (ML-AAC-24) scores with future incident Major Adverse Cardiovascular Events (MACE) that including death, hospitalised acute myocardial infarction or ischemic cerebrovascular disease ascertained from linked healthcare data. Findings: The average intraclass correlation coefficient between imaging specialist and ML-AAC-24 scores for 5012 images was 0.84 (95% CI 0.83, 0.84) with classification accuracy of 80% for established AAC groups. During a mean follow-up 4 years in the registry-based cohort, MACE outcomes were reported in 1177 people (13.7%). With increasing ML-AAC-24 scores there was an increasing proportion of people with MACE (low 7.9%, moderate 14.5%, high 21.2%), as well as individual MACE components (all p-trend <0.001). After multivariable adjustment, moderate and high ML-AAC-24 groups remained significantly associated with MACE (HR 1.54, 95% CI 1.31–1.80 & HR 2.06, 95% CI 1.75–2.42, respectively), compared to those with low ML-AAC-24. Interpretation: The ML-AAC-24 scores had substantial levels of agreement with trained imaging specialists, and was associated with a substantial gradient of risk for cardiovascular events in a real-world setting. This approach could be readily implemented into these clinical settings to improve identification of people at high CVD risk. Funding: The study was supported by a National Health and Medical Research Council of Australia Ideas grant and the Rady Innovation Fund, Rady Faculty of Health Sciences, University of Manitoba.
