Related Experiment Video
Updated: May 1, 2026
Lung CT Segmentation to Identify Consolidations and Ground Glass Areas for Quantitative Assesment of SARS-CoV Pneumonia
Published on: December 19, 2020
Deep Learning Segmentation of Ascites on Abdominal CT Scans for Automatic Volume Quantification
Benjamin Hou1, Sungwon Lee1, Jung-Min Lee1
1From the Imaging Biomarkers and Computer-Aided Diagnosis Laboratory, Department of Radiology and Imaging Sciences, Clinical Center, National Institutes of Health, 10 Center Dr, Bldg 10, Rm 1C224, Bethesda, MD 20892-1182 (B.H., R.M.S.); Department of Radiology, The Catholic University of Korea, Seoul St. Mary's Hospital, Seoul, Korea (S.L.); Women's Malignancies Branch, National Cancer Institute, National Institutes of Health, Bethesda, Md (J.M.L.); Liver Diseases Branch, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, Md (C.K.); Ping An Technology, Shenzhen, China (J.X.); and Department of Radiology, University of Wisconsin School of Medicine & Public Health, Madison, Wis (P.J.P.).
A deep learning method accurately detects and quantifies ascites in liver cirrhosis and ovarian cancer patients using CT scans. This automated segmentation tool shows strong agreement with expert assessments, improving diagnostic capabilities.
Area of Science:
- Medical Imaging and Artificial Intelligence
- Radiology and Oncology
Background:
- Ascites, the accumulation of fluid in the peritoneal cavity, is a common complication in patients with liver cirrhosis and a sign of advanced ovarian cancer.
- Accurate detection and volume quantification of ascites are crucial for diagnosis, treatment planning, and monitoring disease progression.
- Manual segmentation of ascites on CT scans is time-consuming and subject to inter-observer variability.
Purpose of the Study:
- To evaluate the performance of an automated deep learning method for detecting and quantifying ascites volume in patients with liver cirrhosis and ovarian cancer.
- To assess the accuracy and reliability of the deep learning model compared to expert radiologist assessments.
Main Methods:
- A retrospective study utilizing contrast-enhanced and noncontrast abdominal-pelvic CT scans from two institutions (NIH and UofW).
- A deep learning model, trained on The Cancer Genome Atlas Ovarian Cancer dataset, was tested on internal (NIH-LC, NIH-OV) and external (UofW-LC) datasets.
- Performance was measured using F1/Dice coefficients, standard deviations, 95% confidence intervals, and median volume estimation errors, with correlation analysis (r² values) against expert assessments.
Main Results:
- The deep learning model achieved high F1/Dice scores across all test sets: 85.5% ± 6.1 (NIH-LC), 82.6% ± 15.3 (NIH-OV), and 83.0% ± 10.7 (UofW-LC).
- Median volume estimation errors were 19.6% (NIH-LC), 5.3% (NIH-OV), and 9.7% (UofW-LC), indicating good accuracy.
- The model demonstrated strong agreement with expert assessments, with r² values of 0.79, 0.98, and 0.97 across the test sets.
Conclusions:
- The proposed deep learning method effectively segments and quantifies ascites volume in patients with liver cirrhosis and ovarian cancer.
- The automated approach demonstrates performance comparable to expert radiologists, offering a reliable tool for clinical use.
- This deep learning technique has the potential to streamline ascites assessment, improving efficiency and consistency in radiological practice.
Related Concept Videos
01:10Computed Tomography
The technique was invented in the 1970s and is based on the principle that as X-rays pass through the body, they are absorbed or reflected at different levels. In the technique, a patient lies on a motorized platform while a computerized axial tomography (CAT) scanner rotates...
01:28Imaging Studies for Cardiovascular System V: CT
01:27Imaging Studies III: Computed Tomography