Related Experiment Video
Updated: Jan 1, 2026
12:09
Patient-specific Modeling of the Heart: Estimation of Ventricular Fiber Orientations
Published on: January 8, 2013
14.0K
Localised delineation uncertainty for iterative atlas selection in automatic cardiac segmentation.
Robert Finnegan1,2,3, Ebbe Lorenzen4,5, Jason Dowling6,7
1Institute of Medical Physics, School of Physics, University of Sydney, Sydney, Australia.
Physics in Medicine and Biology
|December 24, 2019
Summary
This study introduces an improved automatic heart segmentation method for radiotherapy planning. The novel iterative atlas selection procedure successfully eliminated all segmentation errors, enhancing cardiac safety during cancer treatment.
Area of Science:
- Medical Physics
- Radiotherapy
- Medical Imaging
Background:
- Thoracic radiotherapy poses risks to the heart, with studies linking mean heart dose to increased cardiovascular disease.
- Accurate delineation of cardiac structures is crucial for personalized dose estimation in radiotherapy.
- Automatic segmentation tools are vital for consistent organ-at-risk delineation in large retrospective studies.
Purpose of the Study:
- To develop a robust automatic segmentation method for cardiac structures in thoracic radiotherapy.
- To improve the reliability of multi-atlas based segmentation by addressing contour uncertainties.
- To enhance the accuracy of heart and left anterior descending coronary artery (LADCA) segmentation.
Main Methods:
- Implemented an iterative atlas selection procedure within a multi-atlas segmentation framework.
- Utilized two independent datasets of planning computed tomography (CT) images from Danish and Australian breast cancer patients.
- Employed a cross-validation strategy for segmentation performance assessment using Dice Similarity Coefficient (DSC), Mean Surface-to-Surface Distance (MASD), and Hausdorff Distance (HD).
Main Results:
- The iterative atlas selection procedure successfully removed all segmentation errors.
- The refined segmentation achieved a Dice Similarity Coefficient (DSC) of [Formula: see text].
- Excellent quantitative results were obtained for Mean Surface-to-Surface Distance (MASD) ([Formula: see text] mm) and Hausdorff Distance (HD) ([Formula: see text] mm).
Conclusions:
- The proposed iterative atlas selection method significantly enhances the reliability of automatic cardiac segmentation.
- This robust method is critical for improving the safety of thoracic radiotherapy by ensuring accurate delineation of the heart and LADCA.
- The findings support the clinical application of advanced automatic segmentation techniques for personalized radiation oncology.
