Related Experiment Video
Updated: Apr 13, 2026
Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images
Published on: April 13, 2013
Hybrid method for automatic initialization and segmentation of ventricular on large-scale cardiovascular magnetic
Ning Pan1,2,3, Zhi Li1,2,3, Cailu Xu1,2,3
1College of Biomedical Engineering, South-Central Minzu University, Wuhan, 430074, China.
Insights
A new deep learning algorithm automates cardiac MRI segmentation for large-scale studies. This robust method combines CNNs and Transformers for accurate analysis of cardiovascular magnetic resonance images, improving efficiency in research.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Cardiovascular Research
Background:
- Cardiovascular diseases are a leading global cause of death.
- Accurate cardiac MRI segmentation is crucial for research.
- Current methods are too slow for large datasets.
Purpose of the Study:
- To develop a fully automatic and robust algorithm for large-scale cardiac MRI segmentation.
- To improve the efficiency and accuracy of analyzing population cardiac imaging studies.
Main Methods:
- A hybrid deep learning network (CTr-HNs) integrating CNNs and Transformers with Edge Feature Guidance (EFG) for localization.
- Initial shape acquisition via alignment with a 3D-ASM surface model.
- Refinement of segmentation across all short-axis slices using complex transformations.
Main Results:
- Achieved high Dice coefficients (0.95 for LV, 0.88 for LV myocardium, 0.91 for RV).
- Demonstrated low mean contour distance (0.10 for LV) and Hausdorff distance (1.54 for LV).
- Overall Point-to-Surface error of 1.45 mm for SPASM experiments.
Conclusions:
- The proposed method enables automatic, robust, and accurate large-scale quantification from cardiac images.
- This approach significantly enhances the analysis of population cardiac imaging data.
- The algorithm offers a scalable solution for cardiovascular research.
Background:
Cardiovascular diseases are the number one cause of death globally, making cardiac magnetic resonance image segmentation a popular research topic. Existing schemas relying on manual user interaction or semi-automatic segmentation are infeasible when dealing thousands of cardiac MRI studies. Thus, we proposed a full automatic and robust algorithm for large-scale cardiac MRI segmentation by combining the advantages of deep learning localization and 3D-ASM restriction.
Material And Methods:
The proposed method comprises several key techniques: 1) a hybrid network integrating CNNs and Transformer as a encoder with the EFG (Edge feature guidance) module (named as CTr-HNs) to localize the target regions of the cardiac on MRI images, 2) initial shape acquisition by alignment of coarse segmentation contours to the initial surface model of 3D-ASM, 3) refinement of the initial shape to cover all slices of MRI in the short axis by complex transformation. The datasets used are from the UK BioBank and the CAP (Cardiac Atlas Project). In cardiac coarse segmentation experiments on MR images, Dice coefficients (Dice), mean contour distances (MCD), and mean Hausdorff distances (HD95) are used to evaluate segmentation performance. In SPASM experiments, Point-to-surface (P2S) distances, Dice score are compared between automatic results and ground truth.
Results:
The CTr-HNs from our proposed method achieves Dice coefficients (Dice), mean contour distances (MCD), and mean Hausdorff distances (HD95) of 0.95, 0.10 and 1.54 for the LV segmentation respectively, 0.88, 0.13 and 1.94 for the LV myocardium segmentation, and 0.91, 0.24 and 3.25 for the RV segmentation. The overall P2S errors from our proposed schema is 1.45 mm. For endocardium and epicardium, the Dice scores are 0.87 and 0.91 respectively.
Conclusions:
Our experimental results show that the proposed schema can automatically analyze large-scale quantification from population cardiac images with robustness and accuracy.
