Related Experiment Video
Updated: Jun 23, 2025
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
HaN-Seg: The head and neck organ-at-risk CT and MR segmentation challenge
Gašper Podobnik1, Bulat Ibragimov2, Elias Tappeiner3
1University of Ljubljana, Faculty Electrical Engineering, Tržaška cesta 25, Ljubljana 1000, Slovenia.
Insights
The Head and Neck Organ-at-Risk CT and MR Segmentation Challenge advanced auto-segmentation for radiation therapy planning. Top methods surpassed human accuracy in segmenting organs-at-risk using multi-modal imaging.
Area of Science:
- Medical imaging analysis
- Radiation oncology
- Artificial intelligence in healthcare
Background:
- Accurate segmentation of organs-at-risk (OARs) in the head and neck (HaN) region is crucial for radiation treatment (RT) planning.
- Exploiting multi-modal imaging data, such as computed tomography (CT) and magnetic resonance (MR) imaging, can improve segmentation accuracy.
- The development of automated segmentation methods is essential to streamline RT planning workflows.
Purpose of the Study:
- To promote the development of auto-segmentation methods for HaN radiation treatment planning.
- To leverage information from both CT and MR imaging modalities for improved segmentation.
- To establish a benchmark for multi-modal image segmentation in a clinical context.
Main Methods:
- Organized the HaN-Seg challenge, tasking participants with automatically segmenting 30 OARs in HaN CT and MR images.
- Provided 42 training cases with reference OAR delineations and 14 withheld test cases.
- Performance evaluated using Dice Similarity Coefficient (DSC) and 95-percentile Hausdorff distance (HD95), with statistical ranking via Wilcoxon signed-rank test.
Main Results:
- Seven teams submitted methods, all utilizing U-Net based architectures.
- The top-performing team achieved a DSC of 76.9% and HD95 of 3.5 mm.
- The winning method combined rigid MR to CT registration with multi-modal concatenation at the network input.
Conclusions:
- The challenge successfully simulated a real-world clinical scenario with non-registered, multi-modal images.
- Top-performing methods surpassed inter-observer agreement, demonstrating significant advancements in automated segmentation.
- The publicly available dataset and challenge results provide a valuable benchmark for future research in paired multi-modal image segmentation.
Background And Purpose:
To promote the development of auto-segmentation methods for head and neck (HaN) radiation treatment (RT) planning that exploit the information of computed tomography (CT) and magnetic resonance (MR) imaging modalities, we organized HaN-Seg: The Head and Neck Organ-at-Risk CT and MR Segmentation Challenge.
Materials And Methods:
The challenge task was to automatically segment 30 organs-at-risk (OARs) of the HaN region in 14 withheld test cases given the availability of 42 publicly available training cases. Each case consisted of one contrast-enhanced CT and one T1-weighted MR image of the HaN region of the same patient, with up to 30 corresponding reference OAR delineation masks. The performance was evaluated in terms of the Dice similarity coefficient (DSC) and 95-percentile Hausdorff distance (HD95), and statistical ranking was applied for each metric by pairwise comparison of the submitted methods using the Wilcoxon signed-rank test.
Results:
While 23 teams registered for the challenge, only seven submitted their methods for the final phase. The top-performing team achieved a DSC of 76.9 % and a HD95 of 3.5 mm. All participating teams utilized architectures based on U-Net, with the winning team leveraging rigid MR to CT registration combined with network entry-level concatenation of both modalities.
Conclusion:
This challenge simulated a real-world clinical scenario by providing non-registered MR and CT images with varying fields-of-view and voxel sizes. Remarkably, the top-performing teams achieved segmentation performance surpassing the inter-observer agreement on the same dataset. These results set a benchmark for future research on this publicly available dataset and on paired multi-modal image segmentation in general.
