Coronary X-ray angiography segmentation using Artificial Intelligence: a multicentric validation study of a deep learning model
See where this sits in the topic map →Summary AI-generated
- TL;DR
- We validate a deep learning model for the automatic segmentation of coronary X-ray angiography images using a new multicentric dataset.
- Problem
- Validating previously developed deep learning models on new, multi-center datasets is essential to ensure their reliability and accuracy in clinical applications.
- Method
- The study retrospectively evaluated 123 regions of interest from four centers. Images containing coronary lesions were analyzed using validated software for Quantitative Coronary Analysis and then segmented by the AI model, measuring lesion diameters, pixel overlap, and a global segmentation score.
- Results
- The AI model achieved high accuracy, showing no significant differences in lesion diameter, percentage diameter stenosis, and distal border diameter compared to original images, alongside a Dice Score of 94.8% and a median global segmentation score of 92.
- Contributions
- A multicentric validation study demonstrating that the deep learning model maintains accurate coronary angiography segmentation across diverse data.
- Limitations
- Not specified in the abstract.
- Takeaways
- The AI model accurately segments coronary X-ray angiographies across multiple performance metrics on a multicentric dataset, paving the way for future clinical research.
- Applications
- Automated coronary angiography segmentation and quantitative coronary analysis in clinical settings.
- Topics
- Artificial Intelligence, Deep Learning, Medical Image Segmentation, Cardiovascular Imaging
- For industry
- Healthcare and clinical decision support
- Why it matters
- Advances automated image analysis tools that could support clinicians in evaluating coronary artery disease.
Abstract
INTRODUCTION: We previously developed an artificial intelligence (AI) model for automatic coronary angiography (CAG) segmentation, using deep learning. To validate this approach, the model was applied to a new dataset and results are reported. METHODS: Retrospective selection of patients undergoing CAG and percutaneous coronary intervention or invasive physiology assessment over a one month period from four centers. A single frame was selected from images containing a lesion with a 50-99% stenosis (visual estimation). Automatic Quantitative Coronary Analysis (QCA) was performed with a validated software. Images were then segmented by the AI model. Lesion diameters, area overlap [based on true positive (TP) and true negative (TN) pixels] and a global segmentation score (GSS - 0 -100 points) - previously developed and published - were measured. RESULTS: 123 regions of interest from 117 images across 90 patients were included. There were no significant differences between lesion diameter, percentage diameter stenosis and distal border diameter between the original/segmented images. There was a statistically significant albeit minor difference [0,19 mm (0,09-0,28)] regarding proximal border diameter. Overlap accuracy ((TP + TN)/(TP + TN + FP + FN)), sensitivity (TP / (TP + FN)) and Dice Score (2TP / (2TP + FN + FP)) between original/segmented images was 99,9%, 95,1% and 94,8%, respectively. The GSS was 92 (87-96), similar to the previously obtained value in the training dataset. CONCLUSION: the AI model was capable of accurate CAG segmentation across multiple performance metrics, when applied to a multicentric validation dataset. This paves the way for future research on its clinical uses.