Paper Advances in Spiral fMRI

Summary

This repository contains the analysis code and manuscript (including representation code for all figures) for the paper "Advances in Spiral fMRI - a High-resolution Study with Single-shot Acquisition", now published in NeuroImage 2022, volume 246.

This work is part of the Spiral Functional Imaging (SPIFI) project at the Institute for Biomedical Engineering conducted at the ETH Zurich and University of Zurich.

This paper gives the first account of spiral fMRI at ultra-high field (7 Tesla) and with sub-millimeter (0.8 mm) spatial resolution. This progress was enabled by expanded signal modeling of the MRI signal, including static and dynamic magnetic fields, their accurate concurrent measurement utilizing NMR field probes, and the corresponding iterative model inversion using conjugate-gradient SENSE.

The employed task is a simple visual quarter-field stimulation paradigm.

Installation

1. The preferred installation method is to clone this repository with the recursive option, in order to also checkout the required third-party tools as submodules. On Unix/Mac or Windows (e.g., Git Bash) command line, type the following:

   git clone --recursive https://github.com/mrikasper/paper-advances-in-spiral-fmri.git
   

   • If you download the .zip file instead, please make sure to also download the packages mentioned in the Requirements section below, and place each of them as individual folders in the Code/Toolboxes subfolder of this repository.

   • Afterwards, your Code/Toolboxes folder should contain at least the following folders - make sure they are not empty:

   export_fig
   ismrmrd
   spm12
   TAPAS/PhysIO
   TAPAS/UniQC
   

2. For proper interaction between SPM and PhysIO (see Requirements), the contents of the PhysIO/code subfolder should be copied/linked to a spm12/toolbox/PhysIO subfolder. In order to do so, run

   cd Code/Toolboxes/TAPAS/PhysIO
   tapas_physio_init()
   

   in Matlab.

Getting Started

To reproduce the fMRI analysis, run Code/main.m in Matlab. This should perform

1. The preprocessing, including slice timing correction, realignment and smoothing
2. The first level GLM analysis of the visual quarterfield task, including physiological noise modeling (RETROICOR).
3. The generation of all figure content for the accompanying manuscript. This can also be run separately in Code/Representation/main_create_figures.m.

Note: You will also have to download the single subject example data (ReconstructedImages/SPIFI_0007, direct file download link) from ETH Research Collection and update the paths in Code/spifi_get_analysis_options.m to reflect where you put the raw image and logfile data, and where you would like the results to be written out. Specifically, you have to set

• paths.data.root: study folder of raw data
• paths.results.root: root folder for results and figures

Once the data is available in a FAIR repository, we will also automatize the download to the specified data folder accordingly.

Timeline

• Project started in September 2016
• First successful imaging results November 2016
• Started Paper in October 2017
  • General structure repo: October 2017
  • Introduction: February 2018
• Paper Data
  • Started Acquisition in October 2017
  • Started Analysis (maps) in December 2017
  • Started pipeline recon on Euler (CPU Cluster) in February 2018
  • Started Analysis Code in February 2018
  • Improved B0 map processing for LAYMM and SPIFI: October - December 2018
  • Rerun Analysis of all subjects: March 2019
    • with LAYMM map improvements
  • 2nd Rerun Analysis of all subjects: June 2019
    • with LAYMM recon phase 2 map improvements
• First Full Story Bullets: September 06, 2019
• First Complete Figure Drafts: October 04, 2019
• First Full Draft: November 10, 2019
• First Preprint (BiorXiv): November 15, 2019
• First Submission (Neuroimage): January 18, 2020
• Starting Revision: February 22, 2020
  • ... Covid-19 Pandemic...
• Submitted Revision: June 01, 2021
• Starting Second Revision: July 25, 2021
• Submitted Second Revision: October 23, 2021
• Accepted: November 15, 2021
• Proofs Accepted: December 6, 2021
• Final Open Access Version in NeuroImage: December 8, 2021

First Author: Lars Kasper

Requirements

This code is written in Matlab, and tested with version R2019b on a Windows PC. It further relies on the following open-source Matlab toolboxes

• Statistical Parametric Mapping SPM12 r7771
• The TAPAS software collection, in particular the following toolboxes:
  • UniQC: Unified NeuroImaging Quality Control Toolbox for computing all quality control metrics (SFNR, SD, Mean), ROI analysis and reproducible figure generation (crops, slices, windowing, colormaps)
  • PhysIO Toolbox: for Physiological Noise Modeling in fMRI
• export_fig by Yair Altman, for exporting the paper figure elements to high-resolution PNGs
• ISMRMRD Toolbox for converting raw coil data and magnetic field dynamics to the vendor-independent ISMRM raw data format (ISMRMRD)