A robust multi-variability model based liver segmentation algorithm for CT-scan and MRI modalities.

Area of Science:

• Medical image analysis
• Computer-aided diagnosis
• Biomedical engineering

Background:

• Liver segmentation in medical imaging is challenging due to anatomical variability and similar tissue intensities.
• Inconsistent imaging parameters across different scanners (CT, MRI) further complicate automated segmentation.
• Existing methods often lack robust evaluation metrics for diverse datasets.

Purpose of the Study:

• To develop an automatic model-based segmentation technique for creating accurate 3D liver models.
• To evaluate the proposed method against existing semi-automatic and state-of-the-art approaches.
• To introduce a novel robustness measure for assessing liver segmentation performance across varying scales and datasets.

Main Methods:

• An automatic model-based segmentation approach was developed for 3D liver reconstruction.
• The method was validated on diverse CT and MRI datasets from multiple hospitals and public challenges.
• Performance was compared using the Dice coefficient and a newly proposed robustness metric.

Main Results:

• The automatic segmentation model achieved a mean Dice value of 90.3% on combined CT and MRI datasets.
• The method demonstrated effectiveness across various data sources and imaging parameters.
• A novel robustness measure was introduced, addressing limitations in current evaluation standards.

Conclusions:

• The proposed automatic model-based segmentation offers a robust and accurate solution for liver segmentation.
• The new robustness metric provides a more comprehensive evaluation of segmentation algorithms.
• This work advances the field of medical image analysis for liver studies.