Fully automated MR liver volumetry using watershed segmentation coupled with active contouring
Hieu Trung Huynh1, Ngoc Le-Trong2,3, Pham The Bao3,4
1Faculty of Information Technology, Industrial University of Ho Chi Minh City, Ho Chi Minh City, Vietnam. hthieu@ieee.org.
International Journal of Computer Assisted Radiology and Surgery
|November 23, 2016
Summary
This study introduces a fully automated liver volumetry scheme using MR images, achieving results comparable to manual measurements. The automated system significantly reduces radiologist time, improving efficiency in liver volume assessment.
Area of Science:
- Medical Imaging
- Radiology
- Image Analysis
Background:
- Accurate liver volume measurement is crucial for clinical decisions.
- Manual volumetry is time-consuming and prone to inter-observer variability.
- Automated methods are needed to improve efficiency and consistency.
Purpose of the Study:
- To develop a fully automated scheme for liver volume measurement in abdominal MR images.
- To eliminate the need for user input or interaction in the volumetry process.
Main Methods:
- A three-stage automated scheme: preprocessing, rough liver shape generation, and liver extraction.
- Utilized watershed segmentation, thresholding, morphological operations, and active contour models.
- Compared automated volumes against a radiologist's manual measurements as the gold standard.
Main Results:
- Excellent agreement between automated and manual liver volumes (Intra-class correlation coefficient = 0.94).
- Average processing time of 8.4 minutes per case.
- Demonstrated high accuracy and reliability across 27 cases from multiple medical centers.
Conclusions:
- A fully automated liver volumetry scheme in MR imaging was successfully developed.
- The system requires no user interaction and performs comparably to manual methods.
- Significant time savings for radiologists (24.7 minutes per case) were achieved.
