Skip to content

[Robust cardiac MR image segmentation foundation model] This code contains the most powerful cardiac segmentation model trained from UK biobank dataset with superior performance on out-of-domain datasets. This model can be used out-of-box, and serve as a foundation model for further finetuning

cherise215/CardiacMRSegmentation

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 

Repository files navigation

CardiacMRSegmentation

Paper Improving the generalizability of convolutional neural network-based segmentation on CMR images
In arXiv 2019.

Code is available at https://gitlab.doc.ic.ac.uk/cc215/Cardiac_Multi_view_segmentation

Performance on intra-domain and out-of-domain public datasets

  • Our model trained on UKBB (~4000 training subjects), with extensive data augmentations (described in [1]) can achieve satisfactory performance on unseen out-of-domain public datasets, including ACDC, M&Ms. We report the segmentation performance on these datasets in terms of Dice score below.
UKBB test (600 subjects, 1200 frames: ED+ES) ACDC (100 subjects, 200 frames: ED+ES) (unseen domain) M&Ms (150 subjects, 300 frames: ED+ES) (unseen domain)
configurations LV MYO RV LV MYO RV LV MYO RV
batch_size = 1, roi size = 256, z_score 0.9383 0.8780 0.8979 0.8940 0.8034 0.8237 0.8862 0.7889 0.8168

Adversarially trained model performance on intra-domain and out-of-domain public datasets

  • We further enhance our model performance by applying our recent proposed adversarial data augmentation [2,3].
UKBB ACDC (unseen domain) M&Ms (unseen domain)
LV MYO RV LV MYO RV LV MYO RV
w/o Adv chain 0.9383 0.8780 0.8979 0.8940 0.8034 0.8237 0.8862 0.7889 0.8168
w/ Adv chain 0.9360 0.8732 0.8965 0.9060 0.8087 0.8404 0.8929 0.7987 0.8245

Please cite our work if you find this code useful.

[1] Chen, C. et al. (2020) ‘Improving the Generalizability of Convolutional Neural Network-Based Segmentation on CMR Images’, Frontiers in cardiovascular medicine, 7, p. 105. doi:10.3389/fcvm.2020.00105.

[2] Chen, C, et al. (2020). “Realistic Adversarial Data Augmentation for MR Image Segmentation.” In Medical Image Computing and Computer Assisted Intervention – MICCAI 2020, 667–77. Springer International Publishing.

[3] Chen, C, et al. "Enhancing MR Image Segmentation with Realistic Adversarial Data Augmentation." arXiv preprint arXiv:2108.03429 (2021).


@ARTICLE{Chen2020-gz,

title = "Improving the Generalizability of Convolutional Neural {Network-Based} Segmentation on {CMR} Images",

author = "Chen, Chen and Bai, Wenjia and Davies, Rhodri H and Bhuva, Anish N and Manisty, Charlotte H and Augusto, Joao B and Moon, James C and Aung, Nay and Lee, Aaron M and Sanghvi, Mihir M and Fung, Kenneth and Paiva, Jose Miguel and Petersen, Steffen E and Lukaschuk, Elena and Piechnik, Stefan K and Neubauer, Stefan and Rueckert, Daniel",

journal = "Front Cardiovasc Med",

pages = "105",

month = jun,

year = 2020,

}

@INPROCEEDINGS{Chen2020-ne,

title = "Realistic Adversarial Data Augmentation for {MR} Image Segmentation",

booktitle = "Medical Image Computing and Computer Assisted Intervention -- {MICCAI} 2020",

author = "Chen, Chen and Qin, Chen and Qiu, Huaqi and Ouyang, Cheng and Wang, Shuo and Chen, Liang and Tarroni, Giacomo and Bai, Wenjia and Rueckert, Daniel",

publisher = "Springer International Publishing",

pages = "667--677",

year = 2020

}

About

[Robust cardiac MR image segmentation foundation model] This code contains the most powerful cardiac segmentation model trained from UK biobank dataset with superior performance on out-of-domain datasets. This model can be used out-of-box, and serve as a foundation model for further finetuning

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published