Skip to content

MR Brain Image Super-Resolution with Domain-Distance Adapted Down-Sampling

Notifications You must be signed in to change notification settings

Cuikailong/DDASR

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

31 Commits
 
 
 
 
 
 
 
 

Repository files navigation

DDASR: Domain-Distance Adapted Super-Resolution Reconstruction of MR Brain Images

Abstract

High detail and fast magnetic resonance imaging (MRI) sequences are highly demanded in clinical settings, as inadequate imaging information can lead to diagnostic challenges. MR image super-resolution (SR) is a promising approach to address this issue, but its performance is limited by the practical difficulty of acquiring paired low- and high-resolution (LR and HR) images. Most existing methods generate these pairs by down-sampling HR images, which often fails to capture complex degradations and domain-specific variations. In this study, we propose a domain-distance adapted SR framework (DDASR), which consists of two stages: the domain-distance adapted down-sampling network (DSN) and the GAN-based super-resolution network (SRN). The DSN incorporates characteristics from unpaired LR images during the down-sampling process, enabling the generation of domain-adapted paired LR images. Furthermore, we introduce a novel GAN with an enhanced attention U-Net and multi-scale perceptual loss. The proposed approach yields visually convincing textures and successfully restores 1.5T LR MRI data from the ADNI dataset, outperforming state-of-the-art SR approaches in both perceptual and quantitative evaluations.

The code will be released after the paper is accepted.

Framework

pipeline

Implementation details

Proposed model is trained on 2 NVIDIA GeForce RTX 3090 GPUs with 48 GB memory using the PyTorch.

Dataset

ADNI1 can be accessed from here.

Results presentation

Comparison with the state-of-the-arts

We perform super-resolution on the ADNI1 image and compare it with other models to achieve SOTA. result

Our DSN Synthesizes More Realistic LR Image

We compared the downsampling network and proved that our downsampling network can generate more realistic LR images.

reults_2

Other results

We also conducted ablation experiments on the model to verify the effectiveness of each module and component, see the article for details.

Quick Use

Requirements

  • python 3.6
  • pytorch 1.4.0
  • PIL, numpy, scipy

Hyper parameters

Detailed hyperparameter settings will be made public after the article is accepted.

Test

DSN Test Command:

python DSN/test.py --structure_dir input/dataset/ --texture_dir input/texture_dataset/ --output out

SRN Test Command:

python SRN/test.py --ops SRN/options/DDASR.yml --model_path=pretrained_models/DDASR.pth --input=inputs

Acknowledgments

This tool is developed in Yao Lab. We thank all the contributors and collaborators for their support.

About

MR Brain Image Super-Resolution with Domain-Distance Adapted Down-Sampling

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%