Deep-Motion-Net: GNN-based volumetric liver shape reconstruction from single-view 2D projections
-
By
-
Isuru Wijesinghe
-
Michael Nix
-
Arezoo Zakeri
-
Alireza Hokmabadi
-
Bashar Al-Qaisieh
-
Ali Gooya
-
Zeike Taylor
-
May 13, 2026
-
Clinical Scorecard: Volumetric Reconstruction of Liver Shape from Single-View 2D Projections Using a GNN Approach: Deep-Motion-Net
At a Glance
| Category | Detail |
|---|
| Condition | Anatomical motion during external beam radiotherapy |
| Key Mechanisms | Deep learning-based GNN for 3D volumetric reconstruction from 2D X-ray images |
| Target Population | Patients undergoing radiotherapy requiring precise tumor targeting |
| Care Setting | Radiotherapy departments utilizing kV imaging |
Key Highlights
- Deep-Motion-Net reconstructs 3D organ shapes from single kV X-ray images.
- Utilizes graph attention networks for accurate mesh deformation.
- Incorporates projection angle information for improved feature mapping.
- Addresses limitations of existing methods with arbitrary projection angles.
- Enhances treatment precision by mitigating anatomical motion.
Guideline-Based Recommendations
Diagnosis
- Consider anatomical motion in treatment planning for radiotherapy.
Management
- Implement Deep-Motion-Net for improved organ shape reconstruction.
Monitoring & Follow-up
- Regularly assess the accuracy of 3D reconstructions during treatment.
Risks
- Unaccounted motion may lead to overdosing of organs-at-risk or underdosing of tumors.
Patient & Prescribing Data
Patients receiving hypo-fractionated radiotherapy
Deep-Motion-Net allows for precise targeting and reduced normal tissue irradiation.
Clinical Best Practices
- Utilize noninvasive imaging techniques for real-time monitoring.
- Incorporate advanced deep learning methods for anatomical reconstruction.
- Ensure continuous evaluation of treatment delivery accuracy.
Related Resources & Content
