Related Experiment Video
Updated: Jun 25, 2026
08:41
Novel In Vivo Micro-Computed Tomography Imaging Techniques for Assessing the Progression of Non-Alcoholic Fatty Liver Disease
Published on: March 24, 2023
Comparison and evaluation of methods for liver segmentation from CT datasets.
Tobias Heimann1, Bram van Ginneken, Martin A Styner
1Division of Medical and Biological Informatics, German Cancer Research Center, 69121 Heidelberg, Germany. t.heimann@dkfz.de
IEEE Transactions on Medical Imaging
|February 13, 2009
Summary
Interactive methods generally outperformed automatic approaches for liver segmentation in CT scans, offering more consistent quality. However, advanced automatic methods showed competitive performance, highlighting current image analysis capabilities.
Area of Science:
- Medical Imaging Analysis
- Computer-Aided Diagnosis
- Radiology
Background:
- Accurate liver segmentation from contrast-enhanced CT images is crucial for diagnosis and treatment planning.
- Evaluating the performance of diverse segmentation algorithms is essential for advancing medical image analysis.
Purpose of the Study:
- To compare the performance of 10 automatic and six interactive liver segmentation methods.
- To assess algorithm accuracy against reference segmentations and human expert variability.
- To provide insights into the strengths and limitations of current segmentation techniques.
Main Methods:
- Utilized results from the MICCAI 2007 Grand Challenge, involving 16 participating teams.
- Algorithms included statistical shape models, atlas registration, level-sets, graph-cuts, and rule-based systems.
- Evaluated methods on a common database of 20 clinical CT images with reference segmentations.
Main Results:
- Interactive methods generally achieved higher average scores and better consistency than automatic methods.
- The best automatic methods, particularly those using statistical shape models, demonstrated competitive accuracy.
- Performance was assessed using five error measures and a scoring system relative to human expert variability.
Conclusions:
- Interactive segmentation methods offer superior and more consistent results for liver segmentation.
- Advanced automatic methods show promise and can compete with interactive approaches in many cases.
- The study highlights the current state and challenges in automated medical image segmentation.
