Artifact-free fat-water separation in Dixon MRI using deep learning

Journal of Big Data

Research

Yi Liu¹  Madeleine Cule¹  Elena P. Sorokin¹  Brandon Whitcher²  Jimmy D. Bell²  Marjola Thanaj²  E. Louise Thomas²  Nicolas Basty² 

[1] Calico Life Sciences LLC, South San Francisco, USA;Research Centre for Optimal Health, University of Westminster, London, UK;

关键词: Neural networks;    Magnetic resonance imaging;    Body composition;    Artifact reduction;   

DOI  :  10.1186/s40537-022-00677-1

 received in 2022-03-04, accepted in 2022-12-25,  发布年份 2022

来源: Springer

【 摘 要 】

Chemical-shift encoded MRI (CSE-MRI) is a widely used technique for the study of body composition and metabolic disorders, where derived fat and water signals enable the quantification of adipose tissue and muscle. The UK Biobank is acquiring whole-body Dixon MRI (a specific implementation of CSE-MRI) for over 100,000 participants. Current processing methods associated with large whole-body volumes are time intensive and prone to artifacts during fat-water separation performed by the scanner, making quantitative analysis challenging. The most common artifacts are fat-water swaps, where the labels are inverted at the voxel level. It is common for researchers to discard swapped data (generally around 10%), which is wasteful and may lead to unintended biases. Given the large number of whole-body Dixon MRI acquisitions in the UK Biobank, thousands of swaps are expected to be present in the fat and water volumes from image reconstruction performed on the scanner. If they go undetected, errors will propagate into processes such as organ segmentation, and dilute the results in population-based analyses. There is a clear need for a robust method to accurately separate fat and water volumes in big data collections like the UK Biobank. We formulate fat-water separation as a style transfer problem, where swap-free fat and water volumes are predicted from the acquired Dixon MRI data using a conditional generative adversarial network, and introduce a new loss function for the generator model. Our method is able to predict highly accurate fat and water volumes free from artifacts in the UK Biobank. We show that our model separates fat and water volumes using either single input (in-phase only) or dual input (in-phase and opposed-phase) data, with the latter producing superior results. Our proposed method enables faster and more accurate downstream analysis of body composition from Dixon MRI in population studies by eliminating the need for visual inspection or discarding data due to fat-water swaps.

【 授权许可】

CC BY   
© The Author(s) 2023

【 图 表 】

41116_2022_35_Article_IEq279.gif

41116_2022_35_Article_IEq272.gif

41116_2022_35_Article_IEq266.gif

41116_2022_35_Article_IEq258.gif

41116_2022_35_Article_IEq256.gif

41116_2022_35_Article_IEq255.gif

41116_2022_35_Article_IEq254.gif

41116_2022_35_Article_IEq253.gif

41116_2022_35_Article_IEq252.gif

41116_2022_35_Article_IEq251.gif

41116_2022_35_Article_IEq250.gif

41116_2022_35_Article_IEq248.gif

Fig. 1

41116_2022_35_Article_IEq244.gif

41116_2022_35_Article_IEq243.gif

41116_2022_35_Article_IEq242.gif

41116_2022_35_Article_IEq241.gif

41116_2022_35_Article_IEq240.gif

41116_2022_35_Article_IEq239.gif

41116_2022_35_Article_IEq238.gif

Fig. 4

Fig. 3

Fig. 2

41116_2022_35_Article_IEq234.gif

Fig. 1

Fig. 2

41116_2022_35_Article_IEq231.gif

41116_2022_35_Article_IEq230.gif

41116_2022_35_Article_IEq229.gif

41116_2022_35_Article_IEq228.gif

41116_2022_35_Article_IEq227.gif

41116_2022_35_Article_IEq226.gif

41116_2022_35_Article_IEq225.gif

41116_2022_35_Article_IEq224.gif

41116_2022_35_Article_IEq223.gif

41116_2022_35_Article_IEq222.gif

41116_2022_35_Article_IEq221.gif

41116_2022_35_Article_IEq220.gif

41116_2022_35_Article_IEq219.gif

41116_2022_35_Article_IEq218.gif

41116_2022_35_Article_IEq217.gif

41116_2022_35_Article_IEq216.gif

41116_2022_35_Article_IEq215.gif

Fig. 1

41116_2022_35_Article_IEq213.gif

41116_2022_35_Article_IEq212.gif

41116_2022_35_Article_IEq211.gif

41116_2022_35_Article_IEq210.gif

41116_2022_35_Article_IEq209.gif

40249_2023_1056_Article_IEq5.gif

41116_2022_35_Article_IEq208.gif

41116_2022_35_Article_IEq207.gif

41116_2022_35_Article_IEq206.gif

41116_2022_35_Article_IEq205.gif

41116_2022_35_Article_IEq204.gif

41116_2022_35_Article_IEq203.gif

41116_2022_35_Article_IEq201.gif

Fig. 2

41116_2022_35_Article_IEq199.gif

Fig. 1

41116_2022_35_Article_IEq197.gif

41116_2022_35_Article_IEq196.gif

Fig. 2

41116_2022_35_Article_IEq194.gif

41116_2022_35_Article_IEq193.gif

41116_2022_35_Article_IEq192.gif

41116_2022_35_Article_IEq191.gif

Fig. 1

41116_2022_35_Article_IEq189.gif

41116_2022_35_Article_IEq188.gif

41116_2022_35_Article_IEq173.gif

41116_2022_35_Article_IEq172.gif

Fig. 7

Fig. 5

Fig. 2

41116_2022_35_Article_IEq170.gif

41116_2022_35_Article_IEq169.gif

41116_2022_35_Article_IEq168.gif

Fig. 2

Fig. 4

41116_2022_35_Article_IEq165.gif

Fig. 2

41116_2022_35_Article_IEq163.gif

Fig. 1

Fig. 1

41116_2022_35_Article_IEq159.gif

41116_2022_35_Article_IEq158.gif

41116_2022_35_Article_IEq157.gif

41116_2022_35_Article_IEq155.gif

41116_2022_35_Article_IEq154.gif

41116_2022_35_Article_IEq153.gif

41116_2022_35_Article_IEq152.gif

Fig. 1

41116_2022_35_Article_IEq150.gif

41116_2022_35_Article_IEq149.gif

41116_2022_35_Article_IEq148.gif

41116_2022_35_Article_IEq147.gif

41116_2022_35_Article_IEq146.gif

41116_2022_35_Article_IEq145.gif

41116_2022_35_Article_IEq144.gif

41116_2022_35_Article_IEq143.gif

41116_2022_35_Article_IEq142.gif

41116_2022_35_Article_IEq141.gif

41116_2022_35_Article_IEq140.gif

Fig. 2

41116_2022_35_Article_IEq138.gif

41116_2022_35_Article_IEq137.gif

41116_2022_35_Article_IEq136.gif

41116_2022_35_Article_IEq135.gif

Fig. 5

41116_2022_35_Article_IEq133.gif

41116_2022_35_Article_IEq132.gif

41116_2022_35_Article_IEq131.gif

Fig. 2

41116_2022_35_Article_IEq129.gif

Fig. 5

Fig. 24

Fig. 23

41116_2022_35_Article_IEq127.gif

Fig. 4

Fig. 21

Fig. 5

Fig. 20

Fig. 3

41116_2022_35_Article_IEq125.gif

41116_2022_35_Article_IEq124.gif

41116_2022_35_Article_IEq123.gif

41116_2022_35_Article_IEq122.gif

Fig. 2

42004_2022_796_Article_IEq4.gif

41116_2022_35_Article_IEq119.gif

41116_2022_35_Article_IEq118.gif

41116_2022_35_Article_IEq117.gif

41116_2022_35_Article_IEq116.gif

41116_2022_35_Article_IEq115.gif

Fig. 16

41116_2022_35_Article_IEq106.gif

41116_2022_35_Article_IEq105.gif

41116_2022_35_Article_IEq103.gif

41116_2022_35_Article_IEq99.gif

41116_2022_35_Article_IEq76.gif

12888_2022_4506_Article_IEq2.gif

Fig. 1 

41116_2022_35_Article_IEq60.gif