Related Experiment Video
Updated: Jun 24, 2026
Identifying Coronary Artery Calcification on Non-gated Computed Tomography Scans
Published on: August 28, 2018
Improving coronary artery segmentation with self-supervised learning and automated pericoronary adipose tissue
Justin N Kim1, Yingnan Song1, Hao Wu1
1Case Western Reserve University, Department of Biomedical Engineering, Cleveland, Ohio, United States.
Insights
Self-supervised learning (SSL) improves coronary artery segmentation on coronary computed tomography angiography (CCTA) scans. This AI approach enhances diagnostic accuracy for coronary artery disease (CAD) and pericoronary adipose tissue (PCAT) analysis.
Area of Science:
- Cardiovascular Imaging and AI
- Medical Image Analysis
- Machine Learning in Healthcare
Background:
- Coronary artery disease (CAD) is a major global health concern, with coronary computed tomography angiography (CCTA) being vital for diagnosis.
- Pericoronary adipose tissue (PCAT) characteristics, measured by Hounsfield units (HU), are associated with cardiovascular risk.
- Limited annotated data poses a challenge for developing accurate segmentation models in CCTA analysis.
Purpose of the Study:
- To enhance the accuracy and generalizability of coronary artery segmentation in CCTA using a self-supervised learning (SSL) framework.
- To develop an automated algorithm for segmenting pericoronary adipose tissue (PCAT) for cardiovascular risk assessment.
- To address the limitations of small annotated datasets in medical image analysis for CAD diagnosis.
Main Methods:
- Employed a self-supervised pretraining followed by supervised fine-tuning strategy for coronary artery segmentation.
- Investigated the impact of varying pretraining data volumes on SSL model performance and data efficiency.
- Developed an automated PCAT segmentation pipeline involving centerline extraction, coronary identification, and landmark detection.
Main Results:
- Achieved significant improvements in coronary artery segmentation accuracy, reaching Dice scores up to 0.787 post-pretraining.
- Demonstrated near-perfect performance in automated PCAT segmentation with R-squared values of 0.9998 for LAD and RCA.
- Showcased enhanced model generalizability on external datasets, leading to improved overall segmentation accuracy.
Conclusions:
- Self-supervised learning (SSL) holds significant potential for advancing CCTA image analysis and improving CAD diagnostics.
- The developed SSL approach offers robust automated segmentation for both coronary arteries and PCAT.
- These advancements provide promising tools for enhanced cardiovascular care and risk stratification.
Purpose:
Coronary artery disease (CAD) is a leading cause of morbidity and mortality worldwide, with coronary computed tomography angiography (CCTA) playing a crucial role in its diagnosis. The mean Hounsfield unit (HU) of pericoronary adipose tissue (PCAT) is linked to cardiovascular risk. We utilized a self-supervised learning framework (SSL) to improve the accuracy and generalizability of coronary artery segmentation on CCTA volumes while addressing the limitations of small-annotated datasets.
Approach:
We utilized self-supervised pretraining followed by supervised fine-tuning to segment coronary arteries. To evaluate the data efficiency of SSL, we varied the number of CCTA volumes used during pretraining. In addition, we developed an automated PCAT segmentation algorithm utilizing centerline extraction, spatial-geometric coronary identification, and landmark detection. We evaluated our method on a multi-institutional dataset by assessing coronary artery and PCAT segmentation accuracy via Dice scores and comparing mean PCAT HU values with the ground truth.
Results:
Our approach significantly improved coronary artery segmentation, achieving Dice scores up to 0.787 after self-supervised pretraining. The automated PCAT segmentation achieved near-perfect performance, with -squared values of 0.9998 for both the left anterior descending artery and the right coronary artery indicating excellent agreement between predicted and actual mean PCAT HU values. Self-supervised pretraining notably enhanced model generalizability on external datasets, improving overall segmentation accuracy.
Conclusions:
We demonstrate the potential of SSL to advance CCTA image analysis, enabling more accurate CAD diagnostics. Our findings highlight the robustness of SSL for automated coronary artery and PCAT segmentation, offering promising advancements in cardiovascular care.
More Related Videos
13:07Optical Coherence Tomography Based Biomechanical Fluid-Structure Interaction Analysis of Coronary Atherosclerosis Progression
Published on: January 15, 2022
06:57Author Spotlight: Advancing Cardiovascular Imaging - Introducing the Spatially Weighted Calcium Score for Early Disease Detection
Published on: September 22, 2023
Related Concept Videos
01:21Imaging Studies for Cardiovascular System IV: CMRI
01:28Imaging Studies for Cardiovascular System V: CT
01:25Imaging Studies for Cardiovascular System VI: Calcium -Scoring CT