FDDM: unsupervised medical image translation with a frequency-decoupled diffusion model

Diffusion models have demonstrated significant potential in producing high-quality images in medical image translation to aid disease diagnosis, localization, and treatment. Nevertheless, current diffusion models often fall short when it comes to faithfully translating medical images. They struggle...

Full description

Saved in:
Bibliographic Details
Main Authors: Yunxiang Li, Hua-Chieh Shao, Xiaoxue Qian, You Zhang
Format: Article
Language:English
Published: IOP Publishing 2025-01-01
Series:Machine Learning: Science and Technology
Subjects:
Online Access:https://doi.org/10.1088/2632-2153/adc656
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850216799397740544
author Yunxiang Li
Hua-Chieh Shao
Xiaoxue Qian
You Zhang
author_facet Yunxiang Li
Hua-Chieh Shao
Xiaoxue Qian
You Zhang
author_sort Yunxiang Li
collection DOAJ
description Diffusion models have demonstrated significant potential in producing high-quality images in medical image translation to aid disease diagnosis, localization, and treatment. Nevertheless, current diffusion models often fall short when it comes to faithfully translating medical images. They struggle to accurately preserve anatomical structures, especially when working with unpaired datasets. In this study, we introduce the frequency decoupled diffusion model (FDDM) for magnetic resonance (MR)-to-computed tomography (CT) conversion. The differences between MR and CT images lie in both anatomical structures (e.g. the outlines of organs or bones) and the data distribution (e.g. intensity values and contrast within). Therefore, FDDM first converts anatomical information using an initial conversion module. Then, the converted anatomical information guides a subsequent diffusion model to generate high-quality CT images. Our diffusion model uses a dual-path reverse diffusion process for low-frequency and high-frequency information, achieving a better balance between image quality and anatomical accuracy. We extensively evaluated FDDM using two public datasets for brain MR-to-CT and pelvis MR-to-CT translations. The results show that FDDM outperforms generative adversarial network (GAN)-based, variational autoencoder (VAE)-based, and diffusion-based models. The evaluation metrics included Fréchet inception distance (FID), mean absolute error, mean squared error, structural similarity index measure, and Dice similarity coefficient (DICE). FDDM achieved the best scores on all metrics for both datasets, particularly excelling in FID, with scores of 25.9 for brain data and 29.2 for pelvis data, significantly outperforming the other methods. These results demonstrate that FDDM can generate high-quality target domain images while maintaining the accuracy of translated anatomical structures, thereby facilitating more precise/accurate downstream tasks including anatomy segmentation and radiotherapy planning.
format Article
id doaj-art-8fd7fde79b4f43f88d3aaa5e71ce2a0e
institution OA Journals
issn 2632-2153
language English
publishDate 2025-01-01
publisher IOP Publishing
record_format Article
series Machine Learning: Science and Technology
spelling doaj-art-8fd7fde79b4f43f88d3aaa5e71ce2a0e2025-08-20T02:08:12ZengIOP PublishingMachine Learning: Science and Technology2632-21532025-01-016202500710.1088/2632-2153/adc656FDDM: unsupervised medical image translation with a frequency-decoupled diffusion modelYunxiang Li0https://orcid.org/0000-0003-0622-4710Hua-Chieh Shao1https://orcid.org/0009-0003-7664-7257Xiaoxue Qian2https://orcid.org/0000-0002-6185-1074You Zhang3https://orcid.org/0000-0002-8033-2755Department of Radiation Oncology, UT Southwestern Medical Center , Dallas, TX 75390, United States of AmericaDepartment of Radiation Oncology, UT Southwestern Medical Center , Dallas, TX 75390, United States of AmericaDepartment of Radiation Oncology, UT Southwestern Medical Center , Dallas, TX 75390, United States of AmericaDepartment of Radiation Oncology, UT Southwestern Medical Center , Dallas, TX 75390, United States of AmericaDiffusion models have demonstrated significant potential in producing high-quality images in medical image translation to aid disease diagnosis, localization, and treatment. Nevertheless, current diffusion models often fall short when it comes to faithfully translating medical images. They struggle to accurately preserve anatomical structures, especially when working with unpaired datasets. In this study, we introduce the frequency decoupled diffusion model (FDDM) for magnetic resonance (MR)-to-computed tomography (CT) conversion. The differences between MR and CT images lie in both anatomical structures (e.g. the outlines of organs or bones) and the data distribution (e.g. intensity values and contrast within). Therefore, FDDM first converts anatomical information using an initial conversion module. Then, the converted anatomical information guides a subsequent diffusion model to generate high-quality CT images. Our diffusion model uses a dual-path reverse diffusion process for low-frequency and high-frequency information, achieving a better balance between image quality and anatomical accuracy. We extensively evaluated FDDM using two public datasets for brain MR-to-CT and pelvis MR-to-CT translations. The results show that FDDM outperforms generative adversarial network (GAN)-based, variational autoencoder (VAE)-based, and diffusion-based models. The evaluation metrics included Fréchet inception distance (FID), mean absolute error, mean squared error, structural similarity index measure, and Dice similarity coefficient (DICE). FDDM achieved the best scores on all metrics for both datasets, particularly excelling in FID, with scores of 25.9 for brain data and 29.2 for pelvis data, significantly outperforming the other methods. These results demonstrate that FDDM can generate high-quality target domain images while maintaining the accuracy of translated anatomical structures, thereby facilitating more precise/accurate downstream tasks including anatomy segmentation and radiotherapy planning.https://doi.org/10.1088/2632-2153/adc656medical image translationdiffusion modelgenerative model
spellingShingle Yunxiang Li
Hua-Chieh Shao
Xiaoxue Qian
You Zhang
FDDM: unsupervised medical image translation with a frequency-decoupled diffusion model
Machine Learning: Science and Technology
medical image translation
diffusion model
generative model
title FDDM: unsupervised medical image translation with a frequency-decoupled diffusion model
title_full FDDM: unsupervised medical image translation with a frequency-decoupled diffusion model
title_fullStr FDDM: unsupervised medical image translation with a frequency-decoupled diffusion model
title_full_unstemmed FDDM: unsupervised medical image translation with a frequency-decoupled diffusion model
title_short FDDM: unsupervised medical image translation with a frequency-decoupled diffusion model
title_sort fddm unsupervised medical image translation with a frequency decoupled diffusion model
topic medical image translation
diffusion model
generative model
url https://doi.org/10.1088/2632-2153/adc656
work_keys_str_mv AT yunxiangli fddmunsupervisedmedicalimagetranslationwithafrequencydecoupleddiffusionmodel
AT huachiehshao fddmunsupervisedmedicalimagetranslationwithafrequencydecoupleddiffusionmodel
AT xiaoxueqian fddmunsupervisedmedicalimagetranslationwithafrequencydecoupleddiffusionmodel
AT youzhang fddmunsupervisedmedicalimagetranslationwithafrequencydecoupleddiffusionmodel