Related Experiment Video
Updated: Jul 19, 2025
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
Targeted Training Reduces Search Errors but Not Classification Errors for Hepatic Metastasis Detection at
Scott S Hsieh1, Akitoshi Inoue2, Mariana Yalon2
1Department of Radiology, Mayo Clinic, 200 First St. SW, Rochester, MN 55905 (S.S.H., A.I., M.Y., H.G., P.S.P., J.L.F., S.L., L.Y., C.H.McC., J.G.F.); Department of General Internal Medicine, Mayo Clinic, 200 First St. SW, Rochester, MN 55905 (S.S.H.).
A training program improved the detection of liver metastases by reducing search errors in contrast-enhanced abdominal CT scans. However, the training did not significantly decrease classification errors or benefit all radiologist subgroups.
Area of Science:
- Radiology
- Medical Imaging
- Hepatobiliary System
Background:
- Hepatic metastases detection in contrast-enhanced abdominal CT is crucial but prone to errors.
- Errors occur in both lesion detection (search) and malignancy classification.
- Improved methods are needed to enhance diagnostic accuracy.
Purpose of the Study:
- To evaluate a training program designed to reduce search and classification errors in detecting hepatic metastases.
- To assess the impact of targeted training on radiologist performance in interpreting contrast-enhanced abdominal CT scans.
Main Methods:
- A single-group prospective pretest-posttest study involving 31 radiologists.
- Readers interpreted 40 CT exams with 91 liver metastases under eye tracking.
- Training focused on increasing interpretation time, using specific window settings, and employing coronal reformations, alongside classification practice.
Main Results:
- Search errors decreased from 11% to 8% (P=.01) after training.
- No significant reduction in classification errors was observed (P=.97).
- Jackknife free-response receiver operator characteristic (JAFROC) analysis showed no significant improvement.
Conclusions:
- Targeted training effectively reduced search errors in hepatic metastases detection on CT.
- Classification errors were not significantly impacted by the training program.
- Training benefits were not uniform across all radiologist subgroups, with abdominal subspecialists showing no improvement.
