Automated stenosis estimation of coronary angiographies using end-to-end learning

Christian Kim Eschen,Karina Banasik,Anders Bjorholm Dahl,Piotr J Chmura,Peter Bruun-Rasmussen,Frants Pedersen,Lars Koeber,Thomas Engstrom,Morten Bottcher,Simon Winther,Alex Horby Christensen,Henning Bundgaard,Soren Brunak

DOI: https://doi.org/10.1101/2024.10.08.24315070

2024-10-09

Abstract:Background The initial evaluation of coronary stenosis during coronary angiography is typically performed by visual assessment. The visual assessment of coronary angiographies has limited accuracy compared to quantitative methods like fractional flow reserve and quantitative coronary angiography. Quantitative methods are also more time-consuming and costly. Objectives To test whether applying deep-learning-based image analysis to coronary angiographies might yield a faster and more accurate stenosis estimation than visual assessment. Methods We developed deep learning models for predicting coronary artery stenosis using 332,582 multi-frame x-ray images (cine loops) from 19,414 patients undergoing coronary angiography. The curated dataset for model development included 13,840 patients, with 62,165 cine loops of the left coronary artery and 31,161 cine loops of the right coronary artery. Results For identification of significant coronary stenosis (visual assessment of diameter stenosis >70%), our model obtained a receiver operator characteristic (ROC) area under the curve (ROC-AUC) of 0.903 (95% CI: 0.900-0.906) on the internal test set with 5,056 patients. The performance was evaluated on an external test set with 608 patients against visual assessment, 3D quantitative coronary angiography, and fractional flow reserve (≤ 0.80), obtaining ROC AUC values of 0.833 (95% CI: 0.814-0.852), 0.798 (95% CI: 0.741-0.842, and 0.780 (95% CI: 0.743-0.817), respectively. Conclusion For assessment of coronary stenosis during invasive coronary angiography a deep-learning-based model showed promising results for predicting visual assessment (ROC AUC of 0.903). Compared to previous work, our approach demonstrates performance increase, includes all 16 segments, does not exclude revascularized patients, is externally tested, and is simpler using fewer steps and fewer models.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to use deep - learning methods to automatically assess the degree of coronary artery stenosis during coronary angiography, in order to obtain faster and more accurate results than visual assessment. Specifically, the goals of the paper are:

1. **Improve assessment accuracy**: Traditional coronary artery stenosis assessment mainly relies on doctors' visual judgment, which has high observer variability and low accuracy. By developing an image - analysis model based on deep learning, the paper aims to provide a more objective and accurate stenosis - assessment method.

2. **Reduce time and cost**: Quantitative methods such as quantitative coronary angiography (QCA) and fractional flow reserve (FFR) are highly accurate but time - consuming and costly. The method proposed in the paper aims to reduce the assessment time and cost through an automated process.

3. **Cover all coronary artery segments**: Existing similar studies usually focus on only some coronary artery segments, while the method in this paper can assess all 16 coronary artery segments and does not exclude patients who have undergone revascularization before.

4. **External validation**: The paper not only validates the model performance on the internal test set but also on the external test set from different hospitals to ensure the generalization ability of the model.

In summary, the main goal of this paper is to use deep - learning technology to develop a clinical tool that can quickly and accurately assess the degree of coronary artery stenosis, thereby improving the diagnosis and treatment decisions of cardiovascular diseases.