Related Experiment Video
Updated: Apr 20, 2026
07:53
Author Spotlight: Advancing 3D Modeling for Enhanced Diagnosis and Treatment of Pulmonary Nodules in Early-Stage Lung Cancer
Published on: October 13, 2023
2.3K
Robust Initialization of Active Shape Models for Lung Segmentation in CT Scans: A Feature-Based Atlas Approach
Gurman Gill1, Matthew Toews2, Reinhard R Beichel3
1Department of Electrical and Computer Engineering, The University of Iowa, Iowa City, IA 52242, USA ; The Iowa Institute for Biomedical Imaging, The University of Iowa, Iowa City, IA 52242, USA.
International Journal of Biomedical Imaging
|November 18, 2014
Summary
This study introduces a new method for initializing active shape models (ASM) for 3D lung segmentation in CT scans, significantly improving accuracy and overcoming initialization challenges.
Area of Science:
- Medical Imaging
- Computer Vision
- Radiology
Background:
- Model-based segmentation, like Active Shape Models (ASM), requires close initialization.
- Initialization proximity is a critical limitation for ASM in medical image segmentation.
- Accurate 3D lung segmentation in CT scans is essential for diagnosis and treatment planning.
Purpose of the Study:
- To develop and evaluate a novel initialization method for ASM in 3D lung segmentation.
- To improve the robustness and accuracy of ASM segmentation by addressing initialization issues.
- To demonstrate the generalizability of the proposed initialization technique.
Main Methods:
- Constructed a lung atlas with representative features and an average lung shape.
- Computed affine transforms based on feature matching for pose parameter estimation.
- Applied the initialized ASM to 3D lung segmentation in CT scans.
Main Results:
- Achieved an average Dice coefficient of 0.746 ± 0.068 for initialization.
- Attained an average Dice coefficient of 0.974 ± 0.017 for subsequent segmentation.
- Demonstrated statistically significant improvements over four other methods with a mean absolute surface distance error of 0.948 ± 1.537 mm.
Conclusions:
- The proposed method provides a robust and accurate initialization for ASM in 3D lung segmentation.
- The novel initialization approach significantly enhances segmentation performance compared to existing methods.
- The technique is generalizable to other ASM-based segmentation applications.
