Related Experiment Video
Updated: Apr 30, 2026
02:09
Multi-modal Pulmonary Imaging: Using Complementary Information from CT and Hyperpolarized 129Xe MRI to Evaluate Lung Structure-Function
Published on: April 12, 2024
1.3K
Shape-Constrained Multi-Atlas Segmentation of Spleen in CT
Zhoubing Xu1, Bo Li2, Swetasudha Panda1
1Electrical Engineering, Vanderbilt University, Nashville, TN, USA 37235.
Summary
This study introduces a novel shape-constrained multi-atlas method for spleen segmentation, significantly improving accuracy. The new approach enhances spleen labeling by integrating shape models, outperforming existing techniques.
Area of Science:
- Medical Imaging
- Computer-Aided Diagnosis
- Computational Anatomy
Background:
- Spleen segmentation in CT scans is complex due to anatomical variability.
- Multi-atlas segmentation methods can be hindered by registration inaccuracies.
- Existing methods often implicitly use shape information from atlases.
Purpose of the Study:
- To develop a shape-constrained multi-atlas framework for improved spleen segmentation.
- To integrate a level set shape model into label fusion for spleen segmentation.
- To enhance the robustness of spleen segmentation in clinical CT data.
Main Methods:
- A novel framework combining label fusion with a level set shape model was developed.
- Atlas-to-target registrations were adapted to define spleen boundary constraints.
- Shape priors were derived from projected fusion estimates and integrated into the segmentation process.
Main Results:
- The proposed shape-constrained method significantly improved spleen segmentation accuracy.
- Dice Similarity Coefficient (DSC) increased by 0.06 compared to the LWV method.
- Symmetric mean surface distance decreased by 4.01 mm and symmetric Hausdorff surface distance by 23.21 mm.
Conclusions:
- Integrating shape priors into multi-atlas segmentation enhances spleen labeling accuracy.
- The novel framework offers a statistically significant improvement over traditional methods.
- This approach provides a more robust solution for spleen segmentation in clinical CT imaging.
