Department of Radiology
NYU Grossman School of Medicine
Email: ys1001 at nyu dot edu
My research focuses on creating label-efficient and explainable deep learning models for medical imaging to improve clinical workflow and patient care.In particular, I am excited about exploring the following directions:
- Multimodal learning: How can we efficiently integrate information from different modalities (e.g. imaging, text, tabular data, sequential data) to better inform clinical decisions?
- Explainable AI: Can we enable computer aided diagnosis systems to explain their reasoning in a way that mimics how humans communicate?
- Human-AI collaboration: What is the optimal way to present an AI’s diagnosis to clinicians so they can collaborate effectively to maximize patient outcomes?
I received my Ph.D. at the NYU Center for Data Science, advised by Prof. Krzysztof J. Geras and Prof. Kyunghyun Cho. Prior to joining NYU, I worked at Two Sigma Investments. I hold a Bachelor’s degree in Computer Science from Rice University.
Perspective students: Starting in Fall 2026, I am looking for highly motivated Ph.D. students to join my research group. You may apply through NYU Data Science PhD (see the medical AI track) or the Biomedical Imaging & Technology PhD at the NYU Vilcek Institute. For more details, please check out my advising statement . I’m also open to collaborations with predoctoral researchers and visiting scholars. If you are interested in working with me, please drop me an email.
Our work explores exciting frontiers in AI and medical imaging, including:
- Developing multi-modal deep learning models that integrate medical imaging (across multiple modalities) with clinical data such as EHRs and lab results.
- Designing counterfactual deep learning approaches to optimize diagnostic and treatment pathways.
- Building AI systems for video analysis, with applications such as understanding infant psychology development.
news
| Oct 15, 2026 | Our pancreatic cyst LLM paper was named a semifinalist for AuntMinnie’s 2026 Scientific Paper of the Year! See the semifinalist list and AJR’s announcement. |
|---|---|
| Oct 7, 2026 | Our study on vascular involvement reporting in pancreatic cancer was published in Abdominal Radiology, examining the completeness of routine CT reports and their association with margin-positive resection. |
| Sep 28, 2026 | Our study evaluating pancreatic cyst guidelines for long-term malignancy risk prediction was published in the Journal of the American College of Radiology! Using LLM-extracted radiology features from 13,269 patients, we compared ACR, Kyoto, and European guidelines to inform more individualized surveillance. |
| Sep 1, 2026 | I received the 2026 Milstein Pilot Project Fund. Grateful for this support for our research! |
| Aug 25, 2026 | Our study on pancreatic cyst prevalence and reporting on MRI was published in Abdominal Radiology, revealing how often cysts go unreported and which factors influence their detection in clinical reports. |
| Aug 12, 2026 | Our NYU-DRP breast cancer risk model was published in the American Journal of Roentgenology! The model uses longitudinal 3D mammograms to predict five-year risk. Read the coverage in Yahoo, Health Tech World, and NYU Langone’s press release. |
| Aug 6, 2026 | Our paper rethinking imaging-based IPMN subtype classification was published in Abdominal Radiology, questioning whether a distinct mixed-type category adds value for malignancy risk stratification. |
| Jul 28, 2026 | Our study of incidental findings during pancreatic cyst surveillance was published in Abdominal Radiology, exploring their clinical relevance and implications for more focused MRI protocols. |
| Jul 1, 2026 | Our research received support from the Biswas Family Foundation. Thank you for investing in our work! |
| Jun 15, 2026 | Our team GoGene won NYU Langone’s inaugural DREAMx Challenge for predicting type 2 diabetes and coronary artery disease risk from genomic data! Congratulations to Jiajian Ma and Wushuang Rui. Read the announcement. |
selected publications
2026
-
Leveraging Fine-Tuned Large Language Models for Interpretable Pancreatic Cystic Lesion Feature Extraction and Risk CategorizationAmerican Journal of Roentgenology, 2026
2025
-
Evaluating Generative AI as an Educational Tool for Radiology Resident Report DraftingJournal of the American College of Radiology, 2025
2024
-
Burextract-llama: An llm for clinical concept extraction in breast ultrasound reportsIn Proceedings of the 1st International Workshop on Multimedia Computing for Health and Medicine, 2024
