Related Experiment Video
Updated: May 25, 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
A chance-constrained programming level set method for longitudinal segmentation of lung tumors in CT
Youssef Rouchdy1, Isabelle Bloch
1Telecom ParisTech, CNRSLTCI, Paris, France. youssef.rouchdy@gmail.com
Summary
This study introduces a novel Chance-Constrained Programming (CCP) level set method for tracking lung tumors in CT scans. This new approach effectively handles significant shape changes, outperforming existing methods in accuracy and coherence with manual segmentation.
Area of Science:
- Medical Imaging
- Computational Anatomy
- Radiology
Background:
- Longitudinal tracking of lung tumors in computed tomography (CT) is crucial for monitoring treatment response and disease progression.
- Existing methods like registration-based and segmentation-based approaches have limitations in handling the significant shape and volume variations of lung tumors over time.
- Traditional level set methods struggle with tumors exhibiting random and large shape changes due to their reliance on rigid shape priors.
Purpose of the Study:
- To present a novel stochastic level set method for accurate longitudinal tracking of lung tumors in CT images.
- To address the limitations of existing methods by integrating the advantages of registration and segmentation approaches.
- To introduce a new probabilistic framework, Chance-Constrained Programming (CCP), for flexible tumor shape modeling.
Main Methods:
- Development of a novel Chance-Constrained Programming (CCP) level set model.
- Incorporation of a flexible prior that allows for a specified probability of constraint violation to accommodate variable tumor shapes.
- Computation of chance constraints using user-defined points or segmented reference images, enabling adaptability to different scenarios.
- Implementation of a numerical scheme for approximating the model's solution and application to lung tumor tracking in CT datasets.
Main Results:
- The proposed CCP level set model successfully tracks lung tumors in CT, demonstrating robustness to significant shape and appearance changes.
- Comparison with a Bayesian level set approach showed superior performance of the CCP model.
- The CCP level set model achieved higher coherence with manual segmentations, indicating improved accuracy.
Conclusions:
- The novel CCP level set method offers a significant advancement in longitudinal lung tumor tracking.
- This probabilistic framework provides a flexible and accurate approach for handling tumors with highly variable shapes.
- The CCP level set model represents a more coherent and reliable tool for clinical applications compared to existing methods.
