Related Experiment Video
Updated: Aug 29, 2025
08:17
Dynamic Lung Tumor Tracking for Stereotactic Ablative Body Radiation Therapy
Published on: June 7, 2015
15.8K
A Unified 3D Framework for Organs-at-Risk Localization and Segmentation for Radiation Therapy Planning
Summary
This study introduces a unified 3D pipeline for automatic organ-at-risk (OAR) localization and segmentation in CT scans, improving radiation therapy planning by reducing manual delineation errors and leveraging 3D context for accurate results.
Area of Science:
- Medical Image Analysis
- Radiotherapy Planning
- Computational Anatomy
Background:
- Manual delineation of organs-at-risk (OAR) in CT scans is time-consuming, error-prone, and subject to inter-observer variability.
- Accurate OAR segmentation is crucial for effective radiation therapy planning, optimizing tumor targeting while sparing healthy tissues.
Purpose of the Study:
- To develop and validate a unified 3D pipeline for automated OAR localization and segmentation in CT images.
- To enhance the exploitation of 3D contextual information for improved segmentation accuracy.
Main Methods:
- A 3D multi-variate regression network was employed to predict organ centroids and bounding boxes.
- Subsequent 3D organ-specific segmentation networks generated a multi-organ segmentation map, utilizing predicted locations.
Main Results:
- The proposed unified 3D pipeline achieved a high overall Dice score of 0.9260 ± 0.18% on the challenging VISCERAL dataset.
- The framework effectively utilized 3D context for localization and segmentation of multiple OARs in CT scans.
Conclusions:
- The developed unified 3D pipeline offers an automated and accurate solution for OAR localization and segmentation.
- This approach has the potential to significantly streamline the medical workflow for radiation therapy planning.
