Phy-Diff: Physics-Guided Hourglass Diffusion Model for Diffusion MRI Synthesis

TY - GEN

T1 - Phy-Diff

T2 - 27th International Conference on Medical Image Computing and Computer-Assisted Intervention

AU - Zhang, Juanhua

AU - Yan, Ruodan

AU - Perelli, Alessandro

AU - Chen, Xi

AU - Li, Chao

N1 - Conference code: 27th

PY - 2024/10/4

Y1 - 2024/10/4

N2 - Diffusion MRI (dMRI) is an important neuroimaging technique with high acquisition costs. Deep learning approaches have been used to enhance dMRI and predict diffusion biomarkers through undersampled dMRI. To generate more comprehensive raw dMRI, generative adversarial network based methods are proposed to include b-values and b-vectors as conditions, but they are limited by unstable training and less desirable diversity. The emerging diffusion model (DM) promises to improve generative performance. However, it remains challenging to include essential information in conditioning DM for more relevant generation, i.e., the physical principles of dMRI and white matter tract structures. In this study, we propose a physics-guided diffusion model to generate high-quality dMRI. Our model introduces the physical principles of dMRI in the noise evolution in the diffusion process and introduces a query-based conditional mapping within the diffusion model. In addition, to enhance the anatomical fine details of the generation, we introduce the XTRACT atlas as a prior of white matter tracts by adopting an adapter technique. Our experiment results show that our method outperforms other state-of-the-art methods and has the potential to advance dMRI enhancement.

AB - Diffusion MRI (dMRI) is an important neuroimaging technique with high acquisition costs. Deep learning approaches have been used to enhance dMRI and predict diffusion biomarkers through undersampled dMRI. To generate more comprehensive raw dMRI, generative adversarial network based methods are proposed to include b-values and b-vectors as conditions, but they are limited by unstable training and less desirable diversity. The emerging diffusion model (DM) promises to improve generative performance. However, it remains challenging to include essential information in conditioning DM for more relevant generation, i.e., the physical principles of dMRI and white matter tract structures. In this study, we propose a physics-guided diffusion model to generate high-quality dMRI. Our model introduces the physical principles of dMRI in the noise evolution in the diffusion process and introduces a query-based conditional mapping within the diffusion model. In addition, to enhance the anatomical fine details of the generation, we introduce the XTRACT atlas as a prior of white matter tracts by adopting an adapter technique. Our experiment results show that our method outperforms other state-of-the-art methods and has the potential to advance dMRI enhancement.

KW - Diffusion MRI

KW - Hourglass diffusion model

KW - Image synthesis

KW - Physics informed deep learning

UR - https://www.scopus.com/pages/publications/85206463417

UR - https://arxiv.org/abs/2406.03002

U2 - 10.1007/978-3-031-72069-7_33

DO - 10.1007/978-3-031-72069-7_33

M3 - Conference contribution

AN - SCOPUS:85206463417

SN - 9783031720826

T3 - Lecture Notes in Computer Science

SP - 345

EP - 355

BT - Medical Image Computing and Computer Assisted Intervention

A2 - Linguraru, Marius George

A2 - Dou, Qi

A2 - Feragen, Aasa

A2 - Giannarou, Stamatia

A2 - Glocker, Ben

A2 - Lekadir, Karim

A2 - Schnabel, Julia A.

PB - Springer Nature Switzerland AG

CY - Switzerland

Y2 - 6 October 2024 through 10 October 2024

ER -