A transformer-based deep learning model for predicting the use of atherectomy devices during PCI from preprocedural cardiac computed tomography angiography.
Authors
Affiliations (17)
Affiliations (17)
- Department of Cardiovascular Medicine, Kyushu University Hospital, 3-1-1 Maidashi, Higashi-ku, Fukuoka 812-8582, Japan.
- Department of Advanced Information Technology, Kyushu University, Fukuoka, Japan.
- Department of Cardiovascular Medicine, Tokai University Hachioji Hospital, Tokyo, Japan.
- Department of Cardiovascular Medicine, New Tokyo Hospital, Matsudo, Japan.
- Department of Cardiology, Omuta City Hospital, Omuta, Japan.
- Department of Cardiovascular Medicine, Fukuoka Wajiro Hospital, Fukuoka, Japan.
- Department of Cardiology, Higashi Takarazuka Satoh Hospital, Takarazuka, Japan.
- Department of Cardiology, International University of Health and Welfare Narita Hospital, Chiba, Japan.
- Department of Cardiology, Kita-Harima Medical Center, Ono, Japan.
- Department of Cardiovascular Medicine, Miyazaki Prefectural Nobeoka Hospital, Nobeoka, Japan.
- Department of Cardiology, Ageo Central General Hospital, Ageo, Japan.
- Department of Cardiology, Tokai University Hospital, Isehara, Japan.
- Department of Cardiology, Nozaki Tokushukai Hospital, Daito, Japan.
- Department of Cardiology, Hirakata City Hospital, Hirakata, Japan.
- Department of Cardiology, Fukui Prefectural Hospital, Fukui, Japan.
- Department of Cardiology, Higashikani Hospital, Kani, Japan.
- Cardiovascular Center, Steel Memorial Yawata Hospital, Kitakyushu, Japan.
Abstract
Severe coronary calcification complicates percutaneous coronary intervention (PCI), often necessitating atherectomy-based debulking devices that increase procedural complexity and risk. We aimed to develop and externally validate a Transformer-based deep learning model using preprocedural cardiac computed tomography angiography (CCTA) to predict atherectomy device use during PCI and compare its performance with that of blinded human readers, including general cardiologists and PCI experts. This retrospective multicentre study included 498 calcified coronary lesions from 419 patients who underwent PCI after preprocedural CCTA. Data from 12 institutions were used for training and validation with four-fold cross-validation, and data from two independent institutions (97 lesions) served as an external test cohort. The model combined a convolutional neural network feature extractor with Transformer-based multiple instance learning, and clinical variables were evaluated for incremental value. In the external test cohort, the CCTA-only model achieved an area under the receiver operating characteristic curve (AUC) of 0.788 (95% confidence interval [CI], 0.691-0.873) for predicting atherectomy device use. Adding clinical variables resulted in an AUC of 0.805 (95% CI, 0.713-0.886), indicating limited incremental value. In the reader comparison, the model showed higher accuracy than general cardiologists (0.742 vs. 0.619; <i>P</i> = 0.029) and numerically similar accuracy to PCI experts (0.742 vs. 0.670; <i>P</i> = 0.230). A Transformer-based deep learning model using preprocedural CCTA showed clinically relevant performance for predicting atherectomy device use during PCI in an independent external test cohort. These findings suggest that the model may support lesion assessment and procedural planning.