Abstract
Purpose
Use a conference challenge format to compare machine learning-based gamma-aminobutyric acid (GABA)-edited magnetic resonance spectroscopy (MRS) reconstruction models using one-quarter of the transients typically acquired during a complete scan.
Methods
There were three tracks: Track 1: simulated data, Track 2: identical acquisition parameters with in vivo data, and Track 3: different acquisition parameters with in vivo data. The mean squared error, signal-to-noise ratio, linewidth, and a proposed shape score metric were used to quantify model performance. Challenge organizers provided open access to a baseline model, simulated noise-free data, guides for adding synthetic noise, and in vivo data.
Results
Three submissions were compared. A covariance matrix convolutional neural network model was most successful for Track 1. A vision transformer model operating on a spectrogram data representation was most successful for Tracks 2 and 3. Deep learning (DL) reconstructions with 80 transients achieved equivalent or better SNR, linewidth and fit error compared to conventional 320 transient reconstructions. However, some DL models optimized linewidth and SNR without actually improving overall spectral quality, indicating a need for more robust metrics.
Conclusion
DL-based reconstruction pipelines have the promise to reduce the number of transients required for GABA-edited MRS.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Data and code availability
All code pertaining to the simulated data generation or deep learning models in this article can be found at the following github repository: https://github.com/rmsouza01/Edited-MRS-challenge. In vivo data is available in Big GABA public repository [23], https://www.nitrc.org/projects/biggaba/. This project was supported by NIH grant R01 EB016089.
References
Mullins PG, McGonigle DJ, O’Gorman RL, Puts NAJ, Vidyasagar R, Evans CJ, Edden RAE (2014) Current practice in the use of MEGA-PRESS spectroscopy for the detection of GABA. Neuroimage 86:43–52
Xiao Y, Lanz B, Lim S-I, Tkac I, Xin L (2023) Improved reproducibility of γ-aminobutyric acid measurement from short-echo-time proton MR spectroscopy by linewidth-matched basis sets in LCModel. NMR Biomed 37(2):e5056
Puts NAJ, Edden RAE (2023) In vivo magnetic resonance spectroscopy of GABA: a methodological review. Prog Nucl Magn Reson Spectrosc 60:29–41
Mescher M, Merkle H, Kirsch J, Garwood M, Gruetter R (1998) Simultaneous in vivo spectral editing and water suppression. NMR Biomed 6:260–272
Harris AD, Saleh MG, Edden RAE (2017) Edited 1 H magnetic resonance spectroscopy in vivo: methods and metabolites: edited 1 H MRS. Magn Reson Med 77(4):1377–1389
Peek AL, Rebbeck TJ, Leaver AM, Foster SL, Refshauge KM, Puts NAJ, Oeltzschner G (2023) A comprehensive guide to MEGA-PRESS for GABA measurement. Anal Biochem 669(15):115113
Harris AD, Glaubitz B, Near J, Evans CJ, Puts NAJ, Schmidt-Wilcke T, Tegenthoff M, Barker PB, Edden RAE (2014) Impact of frequency drift on gamma-aminobutyric acid-edited MR spectroscopy. Magn Reson Med 72(4):941–948
Chen D, Hu W, Liu H, Zhou Y, Qiu T, Huang Y, Wang Z, Lin M, Lin L, Wu Z, Wang J, Chen H, Chen X, Yan G, Guo D, Lin J, Qu X (2023) Magnetic resonance spectroscopy deep learning denoising using few in vivo data. IEEE Trans Comput Imaging 9:448–458
Kyathanahally SP, Döring A, Kreis R (2018) Deep learning approaches for detection and removal of ghosting artifacts in MR spectroscopy: detection and removal of ghosting artifacts in MRS using deep learning. Magn Reson Med 80(3):851–863
Lee HH, Kim H (2020) Deep learning-based target metabolite isolation and big data-driven measurement uncertainty estimation in proton magnetic resonance spectroscopy of the brain. Magn Reson Med 84(4):1689–1706
Wang J, Ji B, Lei Y, Liu T, Mao H, Yang X (2023). Denoising Magnetic Resonance Spectroscopy (MRS) Data Using Stacked Autoencoder for Improving Signal-to-Noise Ratio and Speed of MRS. arXiv: 2303.16503
Shamaei A, Starcukova J, Pavlova I, Starcuk Z Jr (2023) Model-informed unsupervised deep learning approaches to frequency and phase correction of MRS signals. Magn Reson Med 89(3):1221–1236
Ma DJ, Le HAM, Ye Y, Laine AF, Lieberman JA, Rothman DL, Small SA, Guo J (2022) MR spectroscopy frequency and phase correction using convolutional neural networks. Magn Reson Med 87(4):1700–1710
Tapper S, Mikkelsen M, Dewey BE, Zöllner HJ, Hui SCN, Oeltzschner G, Edden RAE (2021) Frequency and phase correction of J-difference edited MR spectra using deep learning. Magn Reson Med 85(4):1755–1765
Van De Sande DMJ, Merkofer JP, Amirrajab S, Veta M, Van Sloun RJG, Versluis MJ, Jansen JFA, Van Den Brink JS, Breeuwer M (2023) A review of machine learning applications for the proton MR spectroscopy workflow. Magn Reson Med 90(4):1253–1270
Muckley MJ, Riemenschneider B, Radmanesh A et al (2021) Results of the 2020 fastMRI challenge for machine learning MR image reconstruction. IEEE Trans Med Imaging 40(9):2306–2317
Beauferris Y, Teuwen J, Karkalousos D et al (2022) Multi-coil MRI reconstruction challenge—assessing brain MRI reconstruction models and their generalizability to varying coil configurations. Front Neurosci 16:919186
Harris AD, Amiri H, Bento M et al (2023) Harmonization of multi-scanner in vivo magnetic resonance spectroscopy: ENIGMA consortium task group considerations. Front Neurol 13:1045678
Edden RAE, Puts NAJ, Harris AD, Barker PB, Evans CJ (2014) Gannet: a batch processing tool for the quantitative analysis of gamma-aminobutyric acid-edited MR spectroscopy spectra: Gannet: GABA analysis toolkit. J Magn Reson Imaging 40(6):1445–1452
Simpson R, Devenyi GA, Jezzard P, Hennessy TJ, Near J (2017) Advanced processing and simulation of MRS data using the FID appliance (FID-A)—an open source, MATLAB-based toolkit. Magn Reson Med 77(1):23–33
Near J, Andersson J, Maron E, Mekle R, Gruetter R, Cowen P, Jezzard P (2013) Unedited in vivo detection and quantification of gamma-aminobutyric acid in the occipital cortex using short-TE MRS at 3T. NMR Biomed 26(11):1353–1362
Zhang Y, An L, Shen J (2017) Fast computation of full density matrix of multispin systems for spatially localized in vivo magnetic resonance spectroscopy. Med Phys 44(8):4169–4178
Mikkelsen M, Barker PB, Bhattacharyya PK et al (2017) Big GABA: edited MR spectroscopy at 24 research sites. Neuroimage 159(32–45):32–45
Lin A, Andronesi O, Bogner W et al (2021) Minimum reporting standards for in vivo magnetic resonance spectroscopy (MRSinMRS): experts’ consensus recommendations. NMR Biomed 34(5):e4484
Choi I, Andronesi O, Barker P, Bogner W, Edden RAE, Kaiser LG, Lee P, Marjańska M, Terpstra M, de Graaf RA (2021) Spectral editing in 1H magnetic resonance spectroscopy: experts’ consensus recommendations. NMR Biomed 34(5):e4411
Near J, Harris AD, Juchem C, Kreis R, Marjańska M, Öz G, Slotboom J, Wilson M, Gasparovic C (2021) Preprocessing, analysis and quantification in single-voxel magnetic resonance spectroscopy: experts’ consensus recommendations. NMR Biomed 34(5):e4257
Ronneberger O, Fischer P, Brox T (2015) U-Net: Convolutional Networks for Biomedical Image Segmentation. In: Navab N, Hornegger J, Wells W, Frangi A (eds) Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015. MICCAI 2015. Lecture Notes in Computer Science(), vol 9351. Springer, Cham. pp. 234–241. https://doi.org/10.1007/978-3-319-24574-4_28
Merkofer JP, van de Sande DMJ, Amirrajab S, Drenthen GS, Veta M, Jansen JFA, Breeuwer M, van Sloun RJG (2023) A Deep Learning Approach Utilizing Covariance Matrix Analysis for the ISBI Edited MRS Reconstruction Challenge. arXiv: 2306.02984
Funding
The challenge organizers were supported by the NSERC Discovery Grant (#RGPIN-2021-02867), NSERC Discovery Grant (#RGPIN-2017-03875), NSERC Brain CREATE Award, and the Alberta Graduate Excellence Scholarship. Team Deep Spectral Divers was supported by the DeepMind Scholarship Program, the National Council for Scientific and Technological Development (CNPq Process #313598/2020-7), by the BI0S—Brazilian Institute of Data Science, grant #2020/09838-0, São Paulo Research Foundation (FAPESP)-BRAINN—Brazilian Institute of Neuroscience and Neurotechnology, grant #2013/07559-3, São Paulo Research Foundation (FAPESP), and by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—Brasil (CAPES)—Finance Code 001. Team Spectralligence was in part funded by Spectralligence (EUREKA IA Call, ITEA4 project 20209). Team Dolphins was supported by the Technology Missions Fund under the EPSRC Grant EP/X03870X/1, the British Heart Foundation (RG/20/4/34 803), and The Alan Turing Institute, London, UK.
Author information
Authors and Affiliations
Contributions
RB, HB, RS, and ADH were responsible for designing and organizing the challenge, providing baseline data and code, tutorials on how to use these, results presentation, and creating and maintaining a challenge website and code repository. These authors were also responsible for writing the majority of this manuscript. GD, MO, LU, SD, PDPC, and LR were responsible for the Deep Spectral Divers submission. JPM, DMJVDS, SA, GSD, MV, JFAJ, MB, and RJGVS were responsible for the Spectralligence submission. AQ, CR, and SN were responsible for the Dolphins submission. All authors provided feedback throughout the writing process and reviewed and approved the final submitted version.
Corresponding author
Ethics declarations
Conflict of interest
All co-authors have nothing to declare.
Ethics statement
The work completed in this manuscript adheres to the ethical guidelines outlined by MIT Press.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Roberto Souza and Ashley D. Harris are co-senior authors.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Berto, R.P., Bugler, H., Dias, G. et al. Results of the 2023 ISBI challenge to reduce GABA-edited MRS acquisition time. Magn Reson Mater Phy (2024). https://doi.org/10.1007/s10334-024-01156-9
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s10334-024-01156-9