A physics-informed foundation model for quantitative diffusion MRI

2026-05-29 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Health & Medical Research · Depth: Expert, quick

Summary

A physics-informed generative microstructure network (PIGMENT) has been developed to enable reliable quantitative diffusion MRI mapping of human brain microstructure, even with sparse or heterogeneous data. Trained on 11375 scans from multiple sites, vendors, and field strengths, PIGMENT learns a universal generative prior and adapts it zero-shot to individual participant data. This foundation model successfully recovered meaningful quantitative maps for tensor, kurtosis, and NODDI models across external datasets from five independent centers, outperforming conventional fitting methods in unreliable conditions. PIGMENT effectively extracts data from extremely sparse acquisitions, supports downstream tractography and structural connectivity mapping, and preserves submillimeter cortical microarchitectural patterns and early-childhood white matter developmental trajectories from 10-fold accelerated scans. It also enables reliable quantitative tensor mapping on cost-efficient low-field systems and the extraction of tumor-related biomarkers using ultra-fast clinical protocols.

Key takeaway

For research scientists and clinicians working with diffusion MRI, PIGMENT offers a robust solution to overcome data sparsity and heterogeneity challenges. You can now reliably obtain quantitative microstructure maps from previously unusable sparse acquisitions, including 10-fold accelerated scans or data from low-field systems. This enables broader application of dMRI in clinical settings and research, allowing you to extract meaningful biomarkers and support advanced analyses like tractography, even with ultra-fast protocols.

Key insights

PIGMENT uses a physics-informed generative prior to enable reliable quantitative diffusion MRI from sparse, heterogeneous data.

Principles

• Generative priors can overcome data sparsity in medical imaging.
• Physics-informed models enhance biological validity and robustness.
• Foundation models can generalize across diverse MRI acquisition protocols.

Method

PIGMENT learns a universal generative prior of human brain microstructure from a large dataset, then adapts it zero-shot to individual participant data to recover subject-specific maps.

In practice

• Recover quantitative maps from 10-fold accelerated diffusion MRI scans.
• Enable reliable tensor mapping on cost-efficient low-field MRI systems.
• Extract tumor-related biomarkers using ultra-fast clinical protocols.

Topics

• Diffusion MRI
• Foundation Models
• Brain Microstructure
• Generative Networks
• Neuroimaging
• Quantitative MRI

Best for: AI Scientist, Research Scientist, Domain Expert

Related on AIssential

• GloResNet: A lightweight 3D CNN with global topological features for preterm brain injury prediction — Lightweight 3D CNN with global topological features predicts preterm brain injury effectively.
• AI for Interoperability in Health Care: Philips’s Carla Goulart Peron — AI in healthcare augments clinicians to expand access and improve diagnostics, with interoperability identified as the most impactful global capability.
• China has approved the world’s first invasive brain-computer chip—here’s what’s next — China's NEO BCI, less invasive and government-backed, is the first globally approved for commercial use, offering hope for paralysis patients.
• ClinEnv: An Interactive Multi-Stage Long Horizon EHR Environment for Agents — ClinEnv reveals LLMs struggle with sequential, uncertain clinical decisions and information gathering in a multi-stage environment.
• Quantitative Movement Testing: Measuring Patient Movements from a Single Smartphone Video — QMT uses smartphone video and deep learning for accessible, accurate 3D kinematic biomarker extraction in clinical settings.
• Predicting the risk of colorectal anastomotic leak based on preoperative mapping of the blood supply of the bowel — An AI-driven system predicts colorectal anastomotic leak risk using CT imaging and deep learning for enhanced surgical planning.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.