Meeting Banner
Abstract #3981

Efficient and Effective Control Adapters for Diffusion Plug-and-Play models: A DWI Application

Bingyu Xin1, Radu Miron2, Mahmoud Mostapha3, Nirmal Janardhanan3, Omar Darwish4, Till Huelnhagen4, Tobias Wuerfl4, David Grodzki4, Rainer Schneider4, and Mariappan S. Nadar3
1Rutgers, The State University of New Jersey, NJ, USA, New Jersey, NJ, United States, 2Siemens Industry Software România, Brasov, Romania, Brasov, Romania, 3Siemens Healthineers, Princeton, NJ, USA, Princeton, NJ, United States, 4Siemens Healthineers AG, Erlangen, Germany, Erlangen, Germany

Synopsis

Keywords: AI Diffusion Models, Diffusion Reconstruction, Diffusion Models, Image Reconstruction, ControlNet

Motivation: High-b value DWI suffer from low SNR. Existing deep learning methods for acquisition speed-up are sensitive to protocol changes. A powerful model robust to different clinic scenarios would be ideal.

Goal(s): We aim to use plug-and-play techniques with a diffusion prior with spatial conditioning.

Approach: We first train a powerful diffusion prior on a diverse dataset. An adapter is trained for high-b value DWI reconstruction using the low-b value DWI as condition.

Results: Based on our evaluation, the usage of adapters improves upon the existing plug-and-play with diffusion methods, offering faster convergence.

Impact: The proposed PnP method offers a flexible solution for using a pre-trained diffusion prior in a flexible framework for image reconstruction. Using the adapter can improve certain scenarios, only if needed, without the need for re-training the prior.

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