Generative Medical Intelligence
Multimodal diffusion models that transform sketches, text, and limited annotations into coherent 3D CT/MRI volumes and anatomical masks.
Scientific AI · Generative Modeling · 3D Vision
Delin An
Postdoctoral Researcher at Cornell University
Ph.D. in Computer Science and Engineering University of Notre Dame
I develop data-efficient, geometry-aware AI that turns sparse observations into reliable 3D representations for medical imaging, scientific computing, and embodied perception.
10Peer-reviewed
publications
publications
8First / co-first
author papers
author papers
43D CT / MRI
benchmarks
benchmarks
3Connected research
directions
directions
Research agenda
From sparse signals
to useful worlds.
My work brings together images, masks, and geometry to help AI systems understand structure, not just appearance. I focus on controllable methods that account for uncertainty and remain useful beyond the benchmark.
Geometry-Aware Scientific Computing
Geometry-aware learning for surface reconstruction, simulation-ready modeling, and interpretable analysis of complex scientific structures.
Embodied Perception
Perception-to-control systems that connect visual intelligence with wearable robotics, adaptive locomotion, and real physical interaction.
Selected work
Research with
real geometry.
CVPR 2026 Sketch2CT: Multimodal Diffusion for Structure-Aware 3D Medical Volume Generation
A two-stage latent-diffusion framework that turns a sketch and text prompt into structurally coherent 3D masks and CT/MRI volumes.
VIS 2025 AortaDiff: Volume-Guided Conditional Diffusion Models for Multi-Branch Aortic Surface Generation
Volume-guided conditional diffusion for robust centerline extraction and smooth, multi-branch aortic surface reconstruction.
Sci. Adv. AI-powered automated model construction for patient-specific CFD simulations of aortic flows
An end-to-end AI pipeline from medical scans to patient-specific, CFD-ready aortic models for automated flow simulation.
Paper Equal contribution · co-first author
TVCG 2025 SurfPatch: Enabling Patch Matching for Exploratory Stream Surface Visualization
Multiscale surface-patch matching and interactive visual analytics for exploring complex 3D flow structures.
ISBI 2025 Hierarchical LoG Bayesian neural network for enhanced aorta segmentation
A hierarchical Bayesian network for accurate, uncertainty-aware aorta segmentation in volumetric medical images.
WACV 2025 Sli2Vol+: Segmenting 3D medical images based on an object estimation guided correspondence flow network
Object-guided correspondence flow propagates a single annotated slice through an entire 3D volume with strong structural consistency.
AEI 2027 Using conditional diffusion models for data-driven ground motion identification of structural dynamic systems under uncertainty
Conditional diffusion models recover ground motion from sparse structural responses while accounting for uncertainty in structural dynamic systems.
Paper Equal contribution · co-first author
Academic path
Engineering roots.
Scientific AI future.
My path from mechanical engineering and wearable robotics to computer science shapes how I build AI: grounded in geometry, physical systems, and deployable outcomes.
Download full CV- 2026 — Present
Academic appointment
Cornell University
Postdoctoral Researcher · Scientific machine learning and generative modeling
- 2022 — 2026
Doctoral education
University of Notre Dame
Ph.D. in Computer Science and Engineering · GPA 3.8 / 4.0
- 2019 — 2022
Graduate education
Xi’an Jiaotong University
M.Eng. in Mechanical Engineering · Robotics and perception
- 2015 — 2019
Undergraduate education
UESTC
B.Eng. in Mechanical Engineering · Top 10%
Recognition
Honors, awards
& fellowships.
Recognition spanning scientific AI, graduate research, robotics, and academic excellence.
SAI Graduate Fellowship Pilot
Inaugural fellow and the first and only recipient selected from Notre Dame's Department of Computer Science & Engineering.
University of Notre Dame
Graduate Professional Development Award
Competitive support for graduate research, professional training, and scholarly development.
University of Notre Dame
National Robotics Honors
Third Prize in the National Graduate Robotics Contest and Grand Prize in the Robot Creativity Contest.
Xi'an Jiaotong University
Scholarships & Outstanding Graduate
Special Scholarship, First-Class Scholarship, People's Scholarships, and Outstanding Graduate recognition.
XJTU · UESTC
Now & next
Recent updates.
New paper in Advanced Engineering Informatics: conditional diffusion for ground motion identification under uncertainty. Co-first author.
Joined Cornell University as a Postdoctoral Researcher, working on scientific machine learning and generative modeling.
Successfully defended my Ph.D. dissertation at the University of Notre Dame.
Sketch2CT: Multimodal Diffusion for Structure-Aware 3D Medical Volume Generation was accepted to CVPR 2026.
Our patient-specific aortic modeling work was published in Science Advances.
AortaDiff was accepted to IEEE VIS 2025 and IEEE TVCG.
Collaborate
Let’s build AI that
understands structure.
I am open to conversations around scientific AI, generative modeling, 3D medical imaging, and embodied intelligence.
da568@cornell.edu