ECCV 2026 · Workshop
Artificial Intelligence for Medical 3D Vision
Medical 3D vision lies at the intersection of artificial intelligence, computer vision, computer graphics, and medical imaging. It is becoming increasingly crucial for diagnosis, surgical planning, simulation, and personalised healthcare.
Recent advances in deep learning, generative models, neural rendering, and vision-language approaches have enabled significant progress in 3D reconstruction, 3D generation, novel view synthesis, and the immersive visualization of medical data, including XR-supported applications. This also extends to neuroscience, with approaches that reconstruct 3D representations from neural data.
The goal of this workshop is to highlight innovative, AI-driven techniques for 3D vision in the medical field and encourage discussion about the unique challenges in this field, including data scarcity, noise, anatomical variability, robustness, interpretability, and clinical reliability.
This workshop aims not only to serve as a venue for presenting work in this area but also to build a community and share information in this new field.
We invite original, previously unpublished work addressing, but not limited to, the following topics:
All deadlines at 23:59 Anywhere on Earth (AoE).
University of Modena and Reggio Emilia, Italy
Tenure-Track Assistant Professor at the University of Modena and Reggio Emilia, where he is part of the AImagelab research group. He earned his B.Sc. and M.Sc. degrees (cum laude) in Computer Engineering and completed his Ph.D. in ICT at the same university.
His research spans image processing, algorithm optimization, and medical imaging, including optimization of binary image processing on CPU/GPU, skin lesion segmentation for melanoma detection, CBCT maxillofacial segmentation, and whole-slide image analysis. He contributed to the H2020 DeepHealth project (co-leading the European Computer Vision Library, ECVL) and has been involved in the H2020 DECIDER project since 2022. Federico chairs the IAPR Technical Committee (TC22) on Reproducible Research and is a member of IEEE, MICCAI, and CVPL. He serves as Associate Editor for Pattern Recognition and Simulation & Game.
INRIA, France
Computer Vision PhD Student at INRIA, France, working with Prof. Adnane Boukhayma and Prof. Eric Marchand. His research focuses on deep learning for 3D reconstruction (creating virtual twins) and 3D generative AI.
He has worked on improving generalizable implicit reconstruction models and implicit representation learning from sparse inputs (point clouds, multi-view images). He is currently exploring how to improve the controllability of video generation for 3D-oriented tasks.
GAP Lab, Università Politecnica delle Marche, Italy
University of Bologna, Italy
University of Cambridge, UK
Weill Cornell Medicine, USA
University Medicine Halle (Saale), Germany
University of Macerata, Italy
University of Bologna, Italy
University of Macerata, Italy
GAP Lab, Università Politecnica delle Marche, Italy
Harvard Medical School, USA
September 9, 2026
8:30 AM - 12:30 AM
| Time | Session | Details |
|---|---|---|
| 08:30 – 08:35 | Opening | Opening & Welcome Remarks |
| 08:35 – 09:05 | Keynote Speaker | Federico Bolelli |
| 09:05 – 09:55 | Oral Session | Oral Session 1 |
| 09:05 – 09:15 | Generating 3D baby faces from limited data | |
| 09:15 – 09:25 | Constraint-Aware Self-Supervised Pretraining for Label-Efficient Left Ventricle Segmentation in SPECT MPI | |
| 09:25 – 09:35 | Evidence-Based Text-Conditioned 3D CT Synthesis for Ovarian Cancer | |
| 09:35 – 09:45 | Toward CT-Free 3D Cephalometric Landmark Estimation Using MediaPipe Face Landmarker | |
| 09:45 – 09:55 | Structural-Functional Integration for Enhanced Detection of Subtle Epileptogenic Lesions: A Region-Localized, Clinically-Informed Multimodal EEG-MRI Framework | |
| 09:55 – 10:30 | Spotlight | Spotlight Presentations |
| 09:55 – 09:58 | MedPCFM: Improving Medical Point Cloud Completion by Integrating Point Transformers and Flow Matching | |
| 09:58 – 10:01 | Jolia: Concept-Level Vision-Language Alignment for 3D CT Contrastive Learning | |
| 10:01 – 10:04 | Curia-2: Scaling Self-Supervised Learning for Radiology Foundation Models | |
| 10:04 – 10:07 | Curia-MAE: Multi-Modal Multi-Anatomy MAE Pre-Training for 3D Medical Image Segmentation | |
| 10:07 – 10:10 | YOLO-PVC: 2D-to-3D Consolidation of Slice-wise Detections for Volumetric Liver Tumor Localization in MRI | |
| 10:10 – 10:13 | How Close Is Gaussian Splatting to the Clinic? | |
| 10:13 – 10:16 | Surface-to-Skeleton 3D Cephalometry: Estimating Hidden Skeletal Landmarks from CT-Derived External Soft-Tissue Surfaces | |
| 10:16 – 10:19 | Do 3D Medical Foundation Models See Through MRI Artifacts? A Controlled Study of Representation Robustness | |
| 10:19 – 10:22 | A novel deformation-field framework for data augmentation of pathological fetal brain MRI | |
| 10:22 – 10:25 | Rendering Novel Views of MRI Using 3D Gaussian Splatting | |
| 10:25 – 10:28 | TractoGraphVLM: A Unified Vision-Language Framework for White Matter Tractography | |
| 10:30 – 11:00 | Break & Poster | Coffee Break & Poster Session |
| 11:00 – 11:30 | Keynote Speaker | Amine Ouasfi |
| 11:30 – 12:20 | Oral Session | Oral Session 2 |
| 11:30 – 11:40 | PRISM: Learning Priors for Robust Slice-to-Volume Registration in Ill-Posed Fetal MRI | |
| 11:40 – 11:50 | Learning Dense 2D-3D Correspondence for X-ray-to-CT Registration of Knee Bones | |
| 11:50 – 12:00 | Confidence-Aware Teacher–Student Distillation for 3D Medical Segmentation | |
| 12:00 – 12:10 | Memory-Efficient Skip-Connection-Free 3D Autoencoder for Whole-Brain MRI Reconstruction | |
| 12:10 – 12:20 | What Breaks Under Colonoscopy Artifacts? A Matched Benchmark of Appearance, Geometry, and Registrability in Gaussian-Splatting 3D Reconstruction | |
| 12:20 – 12:30 | Closing | Closing Remarks |