Pre-trained Diffusion Models for Plug-and-Play Medical Image Enhancement

Jun Ma; Yuanzhi Zhu; Chenyu You; Bo Wang

doi:10.1007/978-3-031-43898-1_1

Pre-trained Diffusion Models for Plug-and-Play Medical Image Enhancement

Jun Ma
, Yuanzhi Zhu
, Chenyu You
, Bo Wang

Applied Mathematics and Statistics

University Health Network
University of Toronto
Vector Institute
Swiss Federal Institute of Technology Zurich
Centre of Excellence on Aging and Chronic Diseases of McGill Integrated University Health Network

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

11 Scopus citations

Abstract

Deep learning-based medical image enhancement methods (e.g., denoising and super-resolution) mainly rely on paired data and correspondingly the well-trained models can only handle one type of task. In this paper, we address the limitation with a diffusion model-based framework that mitigates the requirement of paired data and can simultaneously handle multiple enhancement tasks by one pre-trained diffusion model without fine-tuning. Experiments on low-dose CT and heart MR datasets demonstrate that the proposed method is versatile and robust for image denoising and super-resolution. We believe our work constitutes a practical and versatile solution to scalable and generalizable image enhancement.

Original language	English
Title of host publication	Medical Image Computing and Computer Assisted Intervention – MICCAI 2023 - 26th International Conference, Proceedings
Editors	Hayit Greenspan, Hayit Greenspan, Anant Madabhushi, Parvin Mousavi, Septimiu Salcudean, James Duncan, Tanveer Syeda-Mahmood, Russell Taylor
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	3-13
Number of pages	11
ISBN (Print)	9783031438974
DOIs	https://doi.org/10.1007/978-3-031-43898-1_1
State	Published - 2023
Event	26th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2023 - Vancouver, Canada Duration: Oct 8 2023 → Oct 12 2023

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	14222 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	26th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2023
Country/Territory	Canada
City	Vancouver
Period	10/8/23 → 10/12/23

Access to Document

10.1007/978-3-031-43898-1_1

Cite this

Ma, J., Zhu, Y., You, C., & Wang, B. (2023). Pre-trained Diffusion Models for Plug-and-Play Medical Image Enhancement. In H. Greenspan, H. Greenspan, A. Madabhushi, P. Mousavi, S. Salcudean, J. Duncan, T. Syeda-Mahmood, & R. Taylor (Eds.), Medical Image Computing and Computer Assisted Intervention – MICCAI 2023 - 26th International Conference, Proceedings (pp. 3-13). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14222 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-43898-1_1

Ma, Jun ; Zhu, Yuanzhi ; You, Chenyu et al. / Pre-trained Diffusion Models for Plug-and-Play Medical Image Enhancement. Medical Image Computing and Computer Assisted Intervention – MICCAI 2023 - 26th International Conference, Proceedings. editor / Hayit Greenspan ; Hayit Greenspan ; Anant Madabhushi ; Parvin Mousavi ; Septimiu Salcudean ; James Duncan ; Tanveer Syeda-Mahmood ; Russell Taylor. Springer Science and Business Media Deutschland GmbH, 2023. pp. 3-13 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{8cb942864470417da66199346dd2557a,

title = "Pre-trained Diffusion Models for Plug-and-Play Medical Image Enhancement",

abstract = "Deep learning-based medical image enhancement methods (e.g., denoising and super-resolution) mainly rely on paired data and correspondingly the well-trained models can only handle one type of task. In this paper, we address the limitation with a diffusion model-based framework that mitigates the requirement of paired data and can simultaneously handle multiple enhancement tasks by one pre-trained diffusion model without fine-tuning. Experiments on low-dose CT and heart MR datasets demonstrate that the proposed method is versatile and robust for image denoising and super-resolution. We believe our work constitutes a practical and versatile solution to scalable and generalizable image enhancement.",

author = "Jun Ma and Yuanzhi Zhu and Chenyu You and Bo Wang",

note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.; 26th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2023 ; Conference date: 08-10-2023 Through 12-10-2023",

year = "2023",

doi = "10.1007/978-3-031-43898-1\_1",

language = "English",

isbn = "9783031438974",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "3--13",

editor = "Hayit Greenspan and Hayit Greenspan and Anant Madabhushi and Parvin Mousavi and Septimiu Salcudean and James Duncan and Tanveer Syeda-Mahmood and Russell Taylor",

booktitle = "Medical Image Computing and Computer Assisted Intervention – MICCAI 2023 - 26th International Conference, Proceedings",

}

Ma, J, Zhu, Y, You, C & Wang, B 2023, Pre-trained Diffusion Models for Plug-and-Play Medical Image Enhancement. in H Greenspan, H Greenspan, A Madabhushi, P Mousavi, S Salcudean, J Duncan, T Syeda-Mahmood & R Taylor (eds), Medical Image Computing and Computer Assisted Intervention – MICCAI 2023 - 26th International Conference, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 14222 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 3-13, 26th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2023, Vancouver, Canada, 10/8/23. https://doi.org/10.1007/978-3-031-43898-1_1

Pre-trained Diffusion Models for Plug-and-Play Medical Image Enhancement. / Ma, Jun; Zhu, Yuanzhi; You, Chenyu et al.
Medical Image Computing and Computer Assisted Intervention – MICCAI 2023 - 26th International Conference, Proceedings. ed. / Hayit Greenspan; Hayit Greenspan; Anant Madabhushi; Parvin Mousavi; Septimiu Salcudean; James Duncan; Tanveer Syeda-Mahmood; Russell Taylor. Springer Science and Business Media Deutschland GmbH, 2023. p. 3-13 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14222 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Pre-trained Diffusion Models for Plug-and-Play Medical Image Enhancement

AU - Ma, Jun

AU - Zhu, Yuanzhi

AU - You, Chenyu

AU - Wang, Bo

N1 - Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.

PY - 2023

Y1 - 2023

N2 - Deep learning-based medical image enhancement methods (e.g., denoising and super-resolution) mainly rely on paired data and correspondingly the well-trained models can only handle one type of task. In this paper, we address the limitation with a diffusion model-based framework that mitigates the requirement of paired data and can simultaneously handle multiple enhancement tasks by one pre-trained diffusion model without fine-tuning. Experiments on low-dose CT and heart MR datasets demonstrate that the proposed method is versatile and robust for image denoising and super-resolution. We believe our work constitutes a practical and versatile solution to scalable and generalizable image enhancement.

AB - Deep learning-based medical image enhancement methods (e.g., denoising and super-resolution) mainly rely on paired data and correspondingly the well-trained models can only handle one type of task. In this paper, we address the limitation with a diffusion model-based framework that mitigates the requirement of paired data and can simultaneously handle multiple enhancement tasks by one pre-trained diffusion model without fine-tuning. Experiments on low-dose CT and heart MR datasets demonstrate that the proposed method is versatile and robust for image denoising and super-resolution. We believe our work constitutes a practical and versatile solution to scalable and generalizable image enhancement.

UR - https://www.scopus.com/pages/publications/85174715406

U2 - 10.1007/978-3-031-43898-1_1

DO - 10.1007/978-3-031-43898-1_1

M3 - Conference contribution

AN - SCOPUS:85174715406

SN - 9783031438974

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

SP - 3

EP - 13

BT - Medical Image Computing and Computer Assisted Intervention – MICCAI 2023 - 26th International Conference, Proceedings

A2 - Greenspan, Hayit

A2 - Madabhushi, Anant

A2 - Mousavi, Parvin

A2 - Salcudean, Septimiu

A2 - Duncan, James

A2 - Syeda-Mahmood, Tanveer

A2 - Taylor, Russell

PB - Springer Science and Business Media Deutschland GmbH

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

Y2 - 8 October 2023 through 12 October 2023

ER -

Ma J, Zhu Y, You C, Wang B. Pre-trained Diffusion Models for Plug-and-Play Medical Image Enhancement. In Greenspan H, Greenspan H, Madabhushi A, Mousavi P, Salcudean S, Duncan J, Syeda-Mahmood T, Taylor R, editors, Medical Image Computing and Computer Assisted Intervention – MICCAI 2023 - 26th International Conference, Proceedings. Springer Science and Business Media Deutschland GmbH. 2023. p. 3-13. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-43898-1_1