Related Experiment Videos
Quantitative ultrasonic detection and classification of diffuse liver disease. Comparison with human observer
B S Garra1, M F Insana, T H Shawker
1Diagnostic Radiology Department, National Institutes of Health, Bethesda, Maryland.
Investigative Radiology
|March 1, 1989
Summary
A new ultrasound system accurately detects diffuse liver diseases like chronic hepatitis and Gaucher's disease. This multiparameter tissue characterization significantly outperforms human observers in diagnosis.
Area of Science:
- Medical Imaging
- Biophysics
- Diagnostic Ultrasound
Background:
- Diffuse liver diseases present diagnostic challenges.
- Traditional B-scan ultrasound interpretation can be subjective.
- Objective tissue characterization is needed for improved diagnosis.
Purpose of the Study:
- To develop and evaluate a multiparameter ultrasonic tissue characterization system.
- To assess the system's performance in diagnosing various diffuse liver diseases.
- To compare the system's diagnostic accuracy against human observers.
Main Methods:
- Developed a system using four tissue characterization parameters derived from B-scan image statistics.
- Evaluated system performance using receiver operating characteristic (ROC) analysis.
- Compared system performance with experienced human observers.
Main Results:
- Achieved an area under the ROC curve (Az) of 0.88 for chronic hepatitis detection, significantly better than human observers (Az = 0.64).
- System demonstrated high accuracy for Gaucher's disease (Az = 0.92) and differentiating hepatitis from Gaucher's (Az = 0.84).
- Achieved Az values of 0.80 for primary biliary cirrhosis and 0.94 for glycogen storage disease.
Conclusions:
- Multiparameter ultrasonic tissue characterization significantly enhances ultrasound's utility for diffuse liver disease evaluation.
- The developed system shows superior diagnostic performance compared to human interpretation.
- This technology offers a promising, objective approach to liver disease diagnosis.