Meeting Banner
Abstract #3494

Unpaired Multimodal Brain MRI Harmonization with Image Style-Guided Diffusion Model

Mengqi Wu1,2, Yongheng Sun2, Pew-Thian Yap2, Hongtu Zhu3, and Mingxia Liu2
1Joint Department of Biomedical Engineering, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States, 2Department of Radiology and Biomedical Research Imaging Center (BRIC), University of North Carolina at Chapel Hill, Chapel Hill, NC, United States, 3Department of Biostatistics and Biomedical Research Imaging Center (BRIC), University of North Carolina at Chapel Hill, Chapel Hill, NC, United States

Synopsis

Keywords: AI Diffusion Models, Machine Learning/Artificial Intelligence, MRI Harmonization, CLIP Model, Diffusion Model

Motivation: Multi-site MRI data often exhibit non-biological variations due to differences in scanner vendors, field strength, and protocols, hindering downstream image analysis.

Goal(s): We aim to develop a novel framework for brain MRI harmonization that does not require paired training data from traveling subjects, and can harmonize multiple MR sequences (T1, T2, T2-FLAIR).

Approach: Our framework first employs a class-conditioned diffusion model as a coarse harmonizer to unify multi-site MRIs into a common domain, eliminating site-specific image style characteristics. In the second stage, we fine-tune this model to translate the coarsely harmonized MRIs into a specified target style, guided by a style extractor.

Impact: By eliminating non-biological imaging variations from various acquisition sites, our framework allows researchers to utilize multi-site data more effectively, facilitating more robust and generalizable analysis. This will enable large-scale multisite longitudinal studies and increase usable data to improve statistical power.

How to access this content:

For one year after publication, abstracts and videos are only open to registrants of this annual meeting. Registrants should use their existing login information. Non-registrant access can be purchased via the ISMRM E-Library.

After one year, current ISMRM & ISMRT members get free access to both the abstracts and videos. Non-members and non-registrants must purchase access via the ISMRM E-Library.

After two years, the meeting proceedings (abstracts) are opened to the public and require no login information. Videos remain behind password for access by members, registrants and E-Library customers.

Click here for more information on becoming a member.

Keywords