Related Experiment Video
Updated: May 31, 2026
08:05
Lung CT Segmentation to Identify Consolidations and Ground Glass Areas for Quantitative Assesment of SARS-CoV Pneumonia
Published on: December 19, 2020
Surface-region context in optimal multi-object graph-based segmentation: robust delineation of pulmonary tumors
Qi Song1, Mingqing Chen, Junjie Bai
1Department of Electrical & Computer Engineering, University of Iowa, Iowa City, IA 52242, USA. qi-song@uiowa.edu
Information Processing in Medical Imaging : Proceedings of the ... Conference
|July 19, 2011
Summary
This study introduces a novel graph-based method for segmenting interacting objects in medical images. The approach significantly improves lung tumor delineation accuracy compared to conventional methods.
Area of Science:
- Medical Image Analysis
- Computer Vision
- Computational Imaging
Background:
- Multi-object segmentation with mutual interaction is complex in medical imaging.
- Terrain-like surfaces pose challenges for accurate segmentation.
- Existing methods struggle with simultaneous segmentation of interacting objects.
Purpose of the Study:
- To develop a novel solution for segmenting mutually interacting objects and terrain-like surfaces.
- To incorporate context information for simultaneous multi-object segmentation.
- To achieve a globally optimal segmentation solution efficiently.
Main Methods:
- A novel graph model incorporating object-surface interaction via weighted inter-graph arcs.
- Simultaneous segmentation of multiple objects using context information.
- Solving the segmentation problem as a single maximum flow problem in polynomial time.
Main Results:
- The method achieved highly accurate lung tumor segmentations in megavoltage cone-beam CT images.
- Significantly improved segmentation accuracy compared to conventional graph-cut methods (p < 0.001).
- Enhanced Dice coefficient from 0.76 +/- 0.10 to 0.84 +/- 0.05 for pulmonary tumor segmentation.
Conclusions:
- The novel graph-based approach effectively addresses multi-object segmentation with mutual interaction.
- The method offers a robust and accurate solution for medical image segmentation, particularly for lung tumors.
- This technique provides a significant advancement over conventional graph-cut methods in terms of accuracy and performance.
