Precise ablation zone segmentation on CT images after liver cancer ablation using semi-automatic CNN-based
Quoc Anh Le1, Xuan Loc Pham2, Theo van Walsum3
1AVITECH, VNU University of Engineering and Technology, Hanoi, Vietnam.
Medical Physics
|September 9, 2024
Summary
This study introduces a semi-automatic method for segmenting liver ablation zones in CT scans, improving accuracy and reducing manual correction time. The technique enhances quantitative assessment of treatment success for liver lesions.
Area of Science:
- Medical Imaging
- Artificial Intelligence in Medicine
- Oncology
Background:
- Accurate segmentation of ablation zones in contrast-enhanced computed tomography (CECT) is crucial for assessing liver lesion treatment success.
- Current fully automatic methods face challenges with accuracy and require time-consuming manual refinement.
Purpose of the Study:
- To develop a semi-automatic technique for precise liver ablation zone segmentation in CT images.
- To overcome the limitations of existing methods, improving accuracy and efficiency.
Main Methods:
- A hybrid approach combining CNN-based automatic segmentation with interactive CNN-based refinement.
- Initial coarse segmentation is followed by expert validation and localized corrections on individual slices.
- Models were trained and validated on internal datasets and tested on an external benchmark dataset.
Main Results:
- The semi-automatic method achieved high accuracy, with mean Dice similarity coefficients (DSC) of 94.0% on internal data and 87.8% on benchmark data.
- Correction time averaged 2 minutes per case, with performance comparable to manual segmentation by experts.
- The approach demonstrated state-of-the-art performance in ablation segmentation accuracy.
Conclusions:
- The proposed semi-automatic CNN-based segmentation effectively segments ablation zones, enhancing CECT's value in treatment assessment.
- The method offers a reproducible solution, with models, code, and tools publicly available for research.
