End-to-end deep-learning model for the detection of coronary artery stenosis on coronary CT images
Purpose We examined whether end-to-end deep-learning models could detect moderate (≥50%) or severe (≥70%) stenosis in the left anterior descending artery (LAD), right coronary artery (RCA) or left circumflex artery (LCX) in iodine contrast-enhanced ECG-gated coronary CT angiography (CCTA) scans.Meth...
Saved in:
| Main Authors: | , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
BMJ Publishing Group
2025-01-01
|
| Series: | Open Heart |
| Online Access: | https://openheart.bmj.com/content/12/1/e002998.full |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850079805203021824 |
|---|---|
| author | Truls Råmunddal Naveed Sattar Erik Andersson Kristofer Skoglund Araz Rawshani Göran Bergström Deepak L Bhatt Elmir Omerovic Oskar Angerås Bjorn Redfors Jan Borén Petur Petursson Aidin Rawshani Lukas Hilgendorf Vibha Gupta Gustav Smith Carlo Pirazzi |
| author_facet | Truls Råmunddal Naveed Sattar Erik Andersson Kristofer Skoglund Araz Rawshani Göran Bergström Deepak L Bhatt Elmir Omerovic Oskar Angerås Bjorn Redfors Jan Borén Petur Petursson Aidin Rawshani Lukas Hilgendorf Vibha Gupta Gustav Smith Carlo Pirazzi |
| author_sort | Truls Råmunddal |
| collection | DOAJ |
| description | Purpose We examined whether end-to-end deep-learning models could detect moderate (≥50%) or severe (≥70%) stenosis in the left anterior descending artery (LAD), right coronary artery (RCA) or left circumflex artery (LCX) in iodine contrast-enhanced ECG-gated coronary CT angiography (CCTA) scans.Methods From a database of 6293 CCTA scans, we used pre-existing curved multiplanar reformations (CMR) images of the LAD, RCA and LCX arteries to create end-to-end deep-learning models for the detection of moderate or severe stenoses. We preprocessed the images by exploiting domain knowledge and employed a transfer learning approach using EfficientNet, ResNet, DenseNet and Inception-ResNet, with a class-weighted strategy optimised through cross-validation. Heatmaps were generated to indicate critical areas identified by the models, aiding clinicians in understanding the model’s decision-making process.Results Among the 900 CMR cases, 279 involved the LAD artery, 259 the RCA artery and 253 the LCX artery. EfficientNet models outperformed others, with EfficientNetB3 and EfficientNetB0 demonstrating the highest accuracy for LAD, EfficientNetB2 for RCA and EfficientNetB0 for LCX. The area under the curve for receiver operating characteristic (AUROC) reached 0.95 for moderate and 0.94 for severe stenosis in the LAD. For the RCA, the AUROC was 0.92 for both moderate and severe stenosis detection. The LCX achieved an AUROC of 0.88 for the detection of moderate stenoses, though the calibration curve exhibited significant overestimation. Calibration curves matched probabilities for the LAD but showed discrepancies for the RCA. Heatmap visualisations confirmed the models’ precision in delineating stenotic lesions. Decision curve analysis and net reclassification index assessments reinforced the efficacy of EfficientNet models, confirming their superior diagnostic capabilities.Conclusion Our end-to-end deep-learning model demonstrates, for the LAD artery, excellent discriminatory ability and calibration during internal validation, despite a small dataset used to train the network. The model reliably produces precise, highly interpretable images. |
| format | Article |
| id | doaj-art-2fd91659629249acaa11744460e7c5b0 |
| institution | DOAJ |
| issn | 2053-3624 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | BMJ Publishing Group |
| record_format | Article |
| series | Open Heart |
| spelling | doaj-art-2fd91659629249acaa11744460e7c5b02025-08-20T02:45:07ZengBMJ Publishing GroupOpen Heart2053-36242025-01-0112110.1136/openhrt-2024-002998End-to-end deep-learning model for the detection of coronary artery stenosis on coronary CT imagesTruls Råmunddal0Naveed Sattar1Erik Andersson2Kristofer Skoglund3Araz Rawshani4Göran Bergström5Deepak L Bhatt6Elmir Omerovic7Oskar Angerås8Bjorn Redfors9Jan Borén10Petur Petursson11Aidin Rawshani12Lukas Hilgendorf13Vibha Gupta14Gustav Smith15Carlo Pirazzi16Department of Cardiology, Sahlgrenska universitetssjukhuset, Goteborg, SwedenInstitute of Cardiovascular and Medical Sciences, University of Glasgow, Glasgow, UKDepartment of Molecular and Clinical Medicine, University of Gothenburg Institute of Medicine, Gothenburg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg Institute of Medicine, Gothenburg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg Institute of Medicine, Gothenburg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg Institute of Medicine, Gothenburg, SwedenIcahn School of Medicine at Mount Sinai, New York, New York, USADepartment of Molecular and Clinical Medicine, University of Gothenburg Institute of Medicine, Gothenburg, SwedenDepartment of Cardiology, Sahlgrenska universitetssjukhuset, Goteborg, SwedenDepartment of Cardiology, Sahlgrenska University Hospital, Gothenburg, 41345, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg Institute of Medicine, Gothenburg, SwedenDepartment of Cardiology, Sahlgrenska universitetssjukhuset, Goteborg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg Institute of Medicine, Gothenburg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg Institute of Medicine, Gothenburg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg Institute of Medicine, Gothenburg, SwedenDepartment of Molecular and Clinical Medicine, University of Gothenburg Institute of Medicine, Gothenburg, SwedenDepartment of Cardiology, Sahlgrenska universitetssjukhuset, Goteborg, SwedenPurpose We examined whether end-to-end deep-learning models could detect moderate (≥50%) or severe (≥70%) stenosis in the left anterior descending artery (LAD), right coronary artery (RCA) or left circumflex artery (LCX) in iodine contrast-enhanced ECG-gated coronary CT angiography (CCTA) scans.Methods From a database of 6293 CCTA scans, we used pre-existing curved multiplanar reformations (CMR) images of the LAD, RCA and LCX arteries to create end-to-end deep-learning models for the detection of moderate or severe stenoses. We preprocessed the images by exploiting domain knowledge and employed a transfer learning approach using EfficientNet, ResNet, DenseNet and Inception-ResNet, with a class-weighted strategy optimised through cross-validation. Heatmaps were generated to indicate critical areas identified by the models, aiding clinicians in understanding the model’s decision-making process.Results Among the 900 CMR cases, 279 involved the LAD artery, 259 the RCA artery and 253 the LCX artery. EfficientNet models outperformed others, with EfficientNetB3 and EfficientNetB0 demonstrating the highest accuracy for LAD, EfficientNetB2 for RCA and EfficientNetB0 for LCX. The area under the curve for receiver operating characteristic (AUROC) reached 0.95 for moderate and 0.94 for severe stenosis in the LAD. For the RCA, the AUROC was 0.92 for both moderate and severe stenosis detection. The LCX achieved an AUROC of 0.88 for the detection of moderate stenoses, though the calibration curve exhibited significant overestimation. Calibration curves matched probabilities for the LAD but showed discrepancies for the RCA. Heatmap visualisations confirmed the models’ precision in delineating stenotic lesions. Decision curve analysis and net reclassification index assessments reinforced the efficacy of EfficientNet models, confirming their superior diagnostic capabilities.Conclusion Our end-to-end deep-learning model demonstrates, for the LAD artery, excellent discriminatory ability and calibration during internal validation, despite a small dataset used to train the network. The model reliably produces precise, highly interpretable images.https://openheart.bmj.com/content/12/1/e002998.full |
| spellingShingle | Truls Råmunddal Naveed Sattar Erik Andersson Kristofer Skoglund Araz Rawshani Göran Bergström Deepak L Bhatt Elmir Omerovic Oskar Angerås Bjorn Redfors Jan Borén Petur Petursson Aidin Rawshani Lukas Hilgendorf Vibha Gupta Gustav Smith Carlo Pirazzi End-to-end deep-learning model for the detection of coronary artery stenosis on coronary CT images Open Heart |
| title | End-to-end deep-learning model for the detection of coronary artery stenosis on coronary CT images |
| title_full | End-to-end deep-learning model for the detection of coronary artery stenosis on coronary CT images |
| title_fullStr | End-to-end deep-learning model for the detection of coronary artery stenosis on coronary CT images |
| title_full_unstemmed | End-to-end deep-learning model for the detection of coronary artery stenosis on coronary CT images |
| title_short | End-to-end deep-learning model for the detection of coronary artery stenosis on coronary CT images |
| title_sort | end to end deep learning model for the detection of coronary artery stenosis on coronary ct images |
| url | https://openheart.bmj.com/content/12/1/e002998.full |
| work_keys_str_mv | AT trulsramunddal endtoenddeeplearningmodelforthedetectionofcoronaryarterystenosisoncoronaryctimages AT naveedsattar endtoenddeeplearningmodelforthedetectionofcoronaryarterystenosisoncoronaryctimages AT erikandersson endtoenddeeplearningmodelforthedetectionofcoronaryarterystenosisoncoronaryctimages AT kristoferskoglund endtoenddeeplearningmodelforthedetectionofcoronaryarterystenosisoncoronaryctimages AT arazrawshani endtoenddeeplearningmodelforthedetectionofcoronaryarterystenosisoncoronaryctimages AT goranbergstrom endtoenddeeplearningmodelforthedetectionofcoronaryarterystenosisoncoronaryctimages AT deepaklbhatt endtoenddeeplearningmodelforthedetectionofcoronaryarterystenosisoncoronaryctimages AT elmiromerovic endtoenddeeplearningmodelforthedetectionofcoronaryarterystenosisoncoronaryctimages AT oskarangeras endtoenddeeplearningmodelforthedetectionofcoronaryarterystenosisoncoronaryctimages AT bjornredfors endtoenddeeplearningmodelforthedetectionofcoronaryarterystenosisoncoronaryctimages AT janboren endtoenddeeplearningmodelforthedetectionofcoronaryarterystenosisoncoronaryctimages AT peturpetursson endtoenddeeplearningmodelforthedetectionofcoronaryarterystenosisoncoronaryctimages AT aidinrawshani endtoenddeeplearningmodelforthedetectionofcoronaryarterystenosisoncoronaryctimages AT lukashilgendorf endtoenddeeplearningmodelforthedetectionofcoronaryarterystenosisoncoronaryctimages AT vibhagupta endtoenddeeplearningmodelforthedetectionofcoronaryarterystenosisoncoronaryctimages AT gustavsmith endtoenddeeplearningmodelforthedetectionofcoronaryarterystenosisoncoronaryctimages AT carlopirazzi endtoenddeeplearningmodelforthedetectionofcoronaryarterystenosisoncoronaryctimages |