Official websites use .gov

A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS

A lock ( 🔒 ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Need Help

Physics-Informed Machine Learning For Medical Image Analysis

Award Number: DP2EB041544
ORGANIZATION: NATIONAL INSTITUTE Of BIOMEDICAL IMAGING & BIOENGINEERING
OPDIV: NIH
AWARD CLASS: DISCRETIONARY
AWARD ACTIVITY TYPE: SCIENTIFIC/HEALTH RESEARCH (INCLUDES SURVEYS)
PERIOD OF PERFORMANCE START DATE: 09/01/2026
PERIOD OF PERFORMANCE END DATE: 08/31/2031

Group Awards By:

View Award Description

Physics-Informed Machine Learning For Medical Image Analysis - PROJECT SUMMARY Despite major advances in artificial intelligence (AI), I believe that the clinical utility of machine learning (ML) systems for medical image analysis remains limited. While foundation models such as Large Language Models (LLMs) and Vision-Language Models (VLMs) have demonstrated exceptional performance in natural image and language domains, their real-world performance in clinical imaging tasks continues to fall short, especially in settings with limited data, complex clinical context, or the need for causal reasoning. In my view, a central limitation is that these models are typically trained on large, static image-report pairs without incorporating the underlying physical and anatomical knowledge that human experts rely on. In this project, I propose a fundamentally new approach to medical image understanding by embedding physical and anatomical priors into the training of AI systems. My strategy is twofold: First, I will build a generative simulation environment that models dynamic human anatomy and physiology and generates realistic synthetic imaging data, such as radiographs and CT scans, using known tissue properties and forward imaging models. Second, I will develop vision-language-based causal reasoning models that can explain imaging findings based on underlying physical and physiological mechanisms. These two components complement one another: the simulation engine will provide structured, controllable, and dynamic visual inputs, while the reasoning models will learn to associate visual content with causal explanations using curated image- text sources. I will apply and evaluate these approaches in two representative use cases: general radiograph analysis and body CT interpretation. I will leverage public datasets and anatomical atlases alongside synthetic data generated through my simulation pipeline. I will assess performance through expert radiologist evaluations and benchmark tasks measuring realism, generalization, robustness, and data efficiency. My goal is to develop the next generation of medical AI systems that do not simply detect patterns but understand the physical principles that give rise to them. This work will open new paths to explainability, more efficient training, and improved clinical relevance. Given its interdisciplinary nature and its potential to shape foundational approaches in medical AI, I believe this project is uniquely suited for the NIH New Innovator Award.


Issue Date FY	Funding FY	Legal Entity Name	Legal Entity Address	Legal Entity City	Legal Entity State	Legal Entity Zip Code	Legal Entity COUNTY	Legal Entity COUNTRY	Assistance Listing	Award Code	Budget Year	Action Date	Action Type	Action Amount

Issue Date FY: 2026 ( Subtotal = $731,500 )
2026	2026	THE LELAND STANFORD JUNIOR UNIVERSITY	450 JANE STANFORD WAY	STANFORD	CA	94305	SANTA CLARA	USA	Trans-NIH Research Support	000	1	9/2/2026	NEW	$731,500
														Subtotal = $731,500

Grand Total All Awards = $731,500

Top

All Categories

About

Search

Reports

Data Submission

Award Information

Physics-Informed Machine Learning For Medical Image Analysis

Award Number: DP2EB041544

ORGANIZATION: NATIONAL INSTITUTE Of BIOMEDICAL IMAGING & BIOENGINEERING

OPDIV: NIH

AWARD CLASS: DISCRETIONARY

AWARD ACTIVITY TYPE: SCIENTIFIC/HEALTH RESEARCH (INCLUDES SURVEYS)

PERIOD OF PERFORMANCE START DATE: 09/01/2026

PERIOD OF PERFORMANCE END DATE: 08/31/2031

Federal Websites

Department of Health & Human Services

HHS Operating Divisions

HHS Staff Divisions

Download A Document Viewer