Related Experiment Video
Updated: Jun 10, 2026
07:12
Analysis of Fluorescent-Stained Lipid Droplets with 3D Reconstruction for Hepatic Steatosis Assessment
Published on: June 2, 2023
Interactive vs. automatic ultrasound image segmentation methods for staging hepatic lipidosis
Gert Weijers1, Alexander Starke, Alois Haudum
1Clinical Physics Laboratory, Department of Pediatrics, Radboud University Nijmegen Medical Center, Geert Grooteplein Zuid 10 6500 HB Nijmegen, The Netherlands. g.weijers@cukz.umcn.nl
Ultrasonic Imaging
|August 20, 2010
Summary
Automatic segmentation of ultrasound images for fatty liver grading in cows proved more effective than manual methods. This approach enhances accuracy in assessing liver fat content, offering a more efficient and reliable diagnostic tool.
Area of Science:
- Veterinary Medicine
- Medical Imaging
- Biochemistry
Background:
- Fatty liver disease (hepatic lipidosis) is a significant metabolic disorder in dairy cows, impacting animal health and productivity.
- Accurate grading of fatty liver is crucial for effective management and treatment strategies.
- Current methods for assessing liver fat content, such as biochemical analysis of biopsies, are invasive and time-consuming.
Purpose of the Study:
- To evaluate the efficacy of automatic vessel segmentation in ultrasound (US) images for grading fatty liver in dairy cows.
- To compare the performance of automatic segmentation against traditional interactive segmentation methods.
- To determine if automatic segmentation can provide results similar to or better than interactive segmentation for fatty liver assessment.
Main Methods:
- A study involving 151 postpartum dairy cows, an established animal model for human fatty liver disease.
- Acquisition of transcutaneous and intraoperative US liver images, alongside liver biopsies for biochemical triacylglycerol (TAG) analysis.
- Development and application of automatic segmentation algorithms using fixed and adaptive thresholding with speckle exclusion techniques, compared to manual segmentation.
Main Results:
- Automatic segmentation techniques significantly improved correlations between ultrasonic tissue characterization (UTC) parameters and liver TAG levels.
- An SNR-based adaptive automatic-segmentation method achieved the best performance, with R² = 0.71 and AUC = 0.94 for predicting TAG.
- The automatic method demonstrated superior accuracy and efficiency compared to subjective, time-consuming interactive segmentation.
Conclusions:
- Automatic segmentation of hepatic vessels in US images is a feasible and advantageous method for grading fatty liver disease.
- This automated approach offers improved accuracy and efficiency in assessing liver fat content, outperforming manual segmentation.
- The findings support the adoption of automatic segmentation as a valuable tool in veterinary diagnostics for fatty liver disease.
