High-quality annotations for deep learning enabled plaque analysis in SCAPIS cardiac computed tomography angiography
Erika Fagman,
Jennifer Alvén,
Johan Westerbergh,
Pieter Kitslaar,
Michael Kercsik,
Kerstin Cederlund,
Olov Duvernoy,
Jan Engvall,
Isabel Gonçalves,
Hanna Markstad,
Ellen Ostenfeld,
Göran Bergström,
Ola Hjelmgren
Affiliations
Erika Fagman
Department of Radiology, Institute of Clinical Sciences, University of Gothenburg, Sweden; Department of Radiology, Sahlgrenska University Hospital, Gothenburg, Sweden
Jennifer Alvén
Department of Molecular and Clinical Medicine, Institute of Medicine, University of Gothenburg, Sweden; Computer Vision and Medical Image Analysis, Department of Electrical Engineering, Chalmers University of Technology, Gothenburg, Sweden
Johan Westerbergh
Uppsala Clinical Research Center, Uppsala University, Uppsala, Sweden
Pieter Kitslaar
Medis Medical Imaging Systems BV, Leiden, the Netherlands
Michael Kercsik
Department of Molecular and Clinical Medicine, Institute of Medicine, University of Gothenburg, Sweden; Department of Radiology, Alingsås Hospital, Alingsås, Sweden
Kerstin Cederlund
Department of Clinical Science, Intervention and Technology, Karolinska Institutet, Stockholm, Sweden
Olov Duvernoy
Section of Radiology, Department of Surgical Sciences, Uppsala University, Sweden
Jan Engvall
Department of Clinical Physiology and Department of Health, Medicine and Caring Sciences, Linkoping University, Linkoping, Sweden; CMIV – Center for Medical Image Science and Visualization, Linkoping University, Linkoping, Sweden
Isabel Gonçalves
Department of Cardiology, Skane University Hospital, Lund, Sweden; Cardiovascular Research Translational Studies, Clinical Sciences Malmö, Lund University, Sweden
Hanna Markstad
Cardiovascular Research Translational Studies, Clinical Sciences Malmö, Lund University, Sweden; Department of Clinical Sciences Lund, Diagnostic Radiology, Lund University, Skane University Hospital, Lund, Sweden
Ellen Ostenfeld
Department of Clinical Sciences Lund, Clinical Physiology, Lund University, Skane University Hospital, Lund, Sweden
Göran Bergström
Department of Molecular and Clinical Medicine, Institute of Medicine, University of Gothenburg, Sweden; Department of Clinical Physiology, Sahlgrenska University Hospital, Gothenburg, Sweden
Ola Hjelmgren
Department of Molecular and Clinical Medicine, Institute of Medicine, University of Gothenburg, Sweden; Department of Clinical Physiology, Sahlgrenska University Hospital, Gothenburg, Sweden; Corresponding author. Department of Clinical Physiology, Sahlgrenska University Hospital, SE-413 45, Gothenburg, Sweden.
Background: Plaque analysis with coronary computed tomography angiography (CCTA) is a promising tool to identify high risk of future coronary events. The analysis process is time-consuming, and requires highly trained readers. Deep learning models have proved to excel at similar tasks, however, training these models requires large sets of expert-annotated training data. The aims of this study were to generate a large, high-quality annotated CCTA dataset derived from Swedish CArdioPulmonary BioImage Study (SCAPIS), report the reproducibility of the annotation core lab and describe the plaque characteristics and their association with established risk factors. Methods and results: The coronary artery tree was manually segmented using semi-automatic software by four primary and one senior secondary reader. A randomly selected sample of 469 subjects, all with coronary plaques and stratified for cardiovascular risk using the Systematic Coronary Risk Evaluation (SCORE), were analyzed. The reproducibility study (n = 78) showed an agreement for plaque detection of 0.91 (0.84–0.97). The mean percentage difference for plaque volumes was −0.6% the mean absolute percentage difference 19.4% (CV 13.7%, ICC 0.94). There was a positive correlation between SCORE and total plaque volume (rho = 0.30, p < 0.001) and total low attenuation plaque volume (rho = 0.29, p < 0.001). Conclusions: We have generated a CCTA dataset with high-quality plaque annotations showing good reproducibility and an expected correlation between plaque features and cardiovascular risk. The stratified data sampling has enriched high-risk plaques making the data well suited as training, validation and test data for a fully automatic analysis tool based on deep learning.
