DisC-Diff: Disentangled Conditional Diffusion Model for Multi-contrast MRI Super-Resolution

TY - GEN

T1 - DisC-Diff

T2 - 26th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2023

AU - Mao, Ye

AU - Jiang, Lan

AU - Chen, Xi

AU - Li, Chao

N1 - Conference code: 26

PY - 2023/10/1

Y1 - 2023/10/1

N2 - Multi-contrast magnetic resonance imaging (MRI) is the most common management tool used to characterize neurological disorders based on brain tissue contrasts. However, acquiring high-resolution MRI scans is time-consuming and infeasible under specific conditions. Hence, multi-contrast super-resolution methods have been developed to improve the quality of low-resolution contrasts by leveraging complementary information from multi-contrast MRI. Current deep learning-based super-resolution methods have limitations in estimating restoration uncertainty and avoiding mode collapse. Although the diffusion model has emerged as a promising approach for image enhancement, capturing complex interactions between multiple conditions introduced by multi-contrast MRI super-resolution remains a challenge for clinical applications. In this paper, we propose a disentangled conditional diffusion model, DisC-Diff, for multi-contrast brain MRI super-resolution. It utilizes the sampling-based generation and simple objective function of diffusion models to estimate uncertainty in restorations effectively and ensure a stable optimization process. Moreover, DisC-Diff leverages a disentangled multi-stream network to fully exploit complementary information from multi-contrast MRI, improving model interpretation under multiple conditions of multi-contrast inputs. We validated the effectiveness of DisC-Diff on two datasets: the IXI dataset, which contains 578 normal brains, and a clinical dataset with 316 pathological brains. Our experimental results demonstrate that DisC-Diff outperforms other state-of-the-art methods both quantitatively and visually.

AB - Multi-contrast magnetic resonance imaging (MRI) is the most common management tool used to characterize neurological disorders based on brain tissue contrasts. However, acquiring high-resolution MRI scans is time-consuming and infeasible under specific conditions. Hence, multi-contrast super-resolution methods have been developed to improve the quality of low-resolution contrasts by leveraging complementary information from multi-contrast MRI. Current deep learning-based super-resolution methods have limitations in estimating restoration uncertainty and avoiding mode collapse. Although the diffusion model has emerged as a promising approach for image enhancement, capturing complex interactions between multiple conditions introduced by multi-contrast MRI super-resolution remains a challenge for clinical applications. In this paper, we propose a disentangled conditional diffusion model, DisC-Diff, for multi-contrast brain MRI super-resolution. It utilizes the sampling-based generation and simple objective function of diffusion models to estimate uncertainty in restorations effectively and ensure a stable optimization process. Moreover, DisC-Diff leverages a disentangled multi-stream network to fully exploit complementary information from multi-contrast MRI, improving model interpretation under multiple conditions of multi-contrast inputs. We validated the effectiveness of DisC-Diff on two datasets: the IXI dataset, which contains 578 normal brains, and a clinical dataset with 316 pathological brains. Our experimental results demonstrate that DisC-Diff outperforms other state-of-the-art methods both quantitatively and visually.

KW - Conditional diffusion model

KW - Magnetic resonance imaging

KW - Multi-contrast super-resolution

U2 - 10.1007/978-3-031-43999-5_37

DO - 10.1007/978-3-031-43999-5_37

M3 - Conference contribution

AN - SCOPUS:85174677244

SN - 9783031439988

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 387

EP - 397

BT - Medical Image Computing and Computer Assisted Intervention – MICCAI 2023

A2 - Greenspan, Hayit

A2 - Madabhushi, Anant

A2 - Mousavi, Parvin

A2 - Salcudean, Septimiu

A2 - Duncan, James

A2 - Syeda-Mahmood, Tanveer

A2 - Taylor, Russell

PB - Springer

CY - Cham

Y2 - 8 October 2023 through 12 October 2023

ER -