Related Experiment Video
Updated: Dec 8, 2025
06:48
Automated Segmentation of Cortical Grey Matter from T1-Weighted MRI Images
Published on: January 7, 2019
9.3K
A Quality Control System for Automated Prostate Segmentation on T2-Weighted MRI
Mohammed R S Sunoqrot1, Kirsten M Selnæs1,2, Elise Sandsmark2
1Department of Circulation and Medical Imaging, NTNU-Norwegian University of Science and Technology, 7030 Trondheim, Norway.
Diagnostics (Basel, Switzerland)
|September 23, 2020
Summary
This study developed an automated quality control system for prostate segmentation in MRI scans. The system uses radiomics features to accurately assess segmentation quality, improving computer-aided diagnosis efficiency.
Area of Science:
- Radiology
- Medical Imaging
- Artificial Intelligence
Background:
- Computer-aided detection and diagnosis (CAD) systems offer improved efficiency for magnetic resonance imaging (MRI) analysis.
- Automated prostate segmentation is vital for CAD in prostate cancer detection but requires visual inspection for quality assurance.
- Current methods necessitate manual review to identify inaccuracies in automated prostate segmentations.
Purpose of the Study:
- To develop a fully automated quality control (QC) system for prostate segmentation using T2-weighted MRI.
- To establish a reliable method for evaluating the quality of automated prostate segmentations without manual intervention.
Main Methods:
- Four deep learning methods were employed for prostate segmentation across 585 patients.
- Radiomics features (first order, shape, textural) were extracted from segmented prostates.
- A linear regression model using LASSO regularization was trained on radiomics features to predict a reference quality score (QS).
Main Results:
- The automated QC system achieved a mean absolute error of 5.47 ± 6.33 in estimating segmentation quality (scale 0-100).
- A strong correlation (rho = 0.70) was observed between the estimated and reference QSs.
- The model demonstrated generalizability on an independent testing dataset.
Conclusions:
- An automated QC system for prostate segmentation based on T2-weighted MRI has been successfully developed.
- This system shows potential for enhancing the evaluation of automated prostate segmentation quality in CAD systems.
- The findings suggest a promising approach to improve the robustness and efficiency of radiological reading.
