Hi-Fi BOLD fMRI of the human auditory cortex.
Poster No:
2324
Submission Type:
Abstract Submission
Authors:
Carolina Tammurello1, Helene Vitali1, Mauro Leidi2, Ludovica Romanin3, Elodie Savary4, Jessica Bastiaansen5, Adèle Mackowiak6, Oscar Esteban7, Micah Murray4, Monica Gori1, Benedetta Franceschiello8, Eleonora Fornari4
Institutions:
1Italian Institute of Technology, Genoa, Italy, 2The Sense Innovation and Research Centre, Sion, Switzerland, 3Advanced Clinical Imaging Technology, Siemens Healthineers International AG, Lausanne, Switzerland, 4Lausanne University Hospital and University of Lausanne (CHUV), Lausanne, Switzerland, 5Bern University Hospital; Swiss Institute for Translational and Entrepreneurial Medicine, Bern, Switzerland, 6Lausanne University Hospital and University of Lausanne (CHUV), Lausanne, Vaud, 7Lausanne University Hospital and University of Lausanne (CHUV), Lausanne, VD, 8The Sense Innovation and Research Centre; School of Engineering, HES-SO Valais-Wallis, Lausanne & Sion, Switzerland
First Author:
Co-Author(s):
Ludovica Romanin
Advanced Clinical Imaging Technology, Siemens Healthineers International AG
Lausanne, Switzerland
Advanced Clinical Imaging Technology, Siemens Healthineers International AG
Lausanne, Switzerland
Jessica Bastiaansen
Bern University Hospital; Swiss Institute for Translational and Entrepreneurial Medicine
Bern, Switzerland
Bern University Hospital; Swiss Institute for Translational and Entrepreneurial Medicine
Bern, Switzerland
Micah Murray, Ph.D.
Lausanne University Hospital and University of Lausanne (CHUV)
Lausanne, Switzerland
Lausanne University Hospital and University of Lausanne (CHUV)
Lausanne, Switzerland
Benedetta Franceschiello
The Sense Innovation and Research Centre; School of Engineering, HES-SO Valais-Wallis
Lausanne & Sion, Switzerland
The Sense Innovation and Research Centre; School of Engineering, HES-SO Valais-Wallis
Lausanne & Sion, Switzerland
Eleonora Fornari
Lausanne University Hospital and University of Lausanne (CHUV)
Lausanne, Switzerland
Lausanne University Hospital and University of Lausanne (CHUV)
Lausanne, Switzerland
Introduction:
Recent fMRI acceleration techniques have allowed for sub-second temporal resolution1,2, but are still impacted by decreased Signal to Noise Ratio (SNR), inter-slice crosstalk and eddy current artefacts3. A novel acquisition and reconstruction approach, dubbed Hi-Fi fMRI, combines a spiral phyllotaxis sampling trajectory3, a compressed-sensing algorithm4, and application of motion correction directly to k-space5 to obtain minimal distortions and high SNR6,7 as well as a 250msec temporal resolution (chosen during image reconstruction). To date, this has been successfully applied in visual cortices. Here, we employ complex auditory stimuli8 to activate the deeper and smaller auditory cortices, so as to test the robustness of the technique.
Methods:
Three healthy adult volunteers underwent a passive auditory stimulation task (Fig 1a). Auditory stimuli were delivered in no-stereo mode to both ears through MR compatible noise-cancelling headphones. Participants fixated centrally a gray patch while listening to a total of 49, 40s-long trials. Sequences of meaningful environmental sounds (e.g. bells, dog barking) superposed onto an everyday auditory scene (e.g. market, street)8 were played during the ON phase of each trial (0-15s)6. Pink noise was played during the REF phase (15-40s), so as to prevent the rhythmicity of the scanner noise from eliciting peaks of BOLD activity.
An uninterrupted gradient recalled echo (GRE) research application sequence was acquired with a 3T clinical scanner (MAGNETOM PrismaFit, Siemens Healthcare, Erlangen, Germany), with TE/TR=25/28.24ms, FoV=192x192x192mm3, 0.8 mm3 isotropic resolution, FA=12°, TA = 33min. A 3D radial spiral phyllotaxis sampling trajectory was employed, guaranteeing uniform readout distribution in k-space3.
Images were reconstructed in 5D along the x-y-z-repetition (trial) -activity (ON/REF) dimensions with a k-t-sparse SENSE algorithm (image under-sampling 20%)4, resulting in 98 3D volumes (Fig 1B2). Total variation regularization was applied along the repetition dimension. Readouts acquired in 2 time-intervals were selected: 5-15s, capturing BOLD signal at peak during the ON phase, and 30-40s, corresponding to BOLD baseline during the REF phase (pink noise). Readouts belonging to the same activity type were unified via sum of squares and the absolute difference between ON and REF was computed. ON and REF images for each trial underwent a statistical analysis and clusters with a statistical difference at p<0.05 (FWE-corrected, T>6.38) were selected; this threshold corresponded, in the ON-REF image calculation, to 83au. Within these clusters, a separate set of 4D reconstructions was performed along x-y-z-t (time) dimensions, with bins of 2.5sec, where Fourier transform regularization was applied along the time dimension (Fig 1B1). Motion correction translation and rotational coefficients were estimated using SPM129 and applied to the original k-space.
An uninterrupted gradient recalled echo (GRE) research application sequence was acquired with a 3T clinical scanner (MAGNETOM PrismaFit, Siemens Healthcare, Erlangen, Germany), with TE/TR=25/28.24ms, FoV=192x192x192mm3, 0.8 mm3 isotropic resolution, FA=12°, TA = 33min. A 3D radial spiral phyllotaxis sampling trajectory was employed, guaranteeing uniform readout distribution in k-space3.
Images were reconstructed in 5D along the x-y-z-repetition (trial) -activity (ON/REF) dimensions with a k-t-sparse SENSE algorithm (image under-sampling 20%)4, resulting in 98 3D volumes (Fig 1B2). Total variation regularization was applied along the repetition dimension. Readouts acquired in 2 time-intervals were selected: 5-15s, capturing BOLD signal at peak during the ON phase, and 30-40s, corresponding to BOLD baseline during the REF phase (pink noise). Readouts belonging to the same activity type were unified via sum of squares and the absolute difference between ON and REF was computed. ON and REF images for each trial underwent a statistical analysis and clusters with a statistical difference at p<0.05 (FWE-corrected, T>6.38) were selected; this threshold corresponded, in the ON-REF image calculation, to 83au. Within these clusters, a separate set of 4D reconstructions was performed along x-y-z-t (time) dimensions, with bins of 2.5sec, where Fourier transform regularization was applied along the time dimension (Fig 1B1). Motion correction translation and rotational coefficients were estimated using SPM129 and applied to the original k-space.
·Representation of experimental paradigm and analysis. Figure 1a shows the experimental and auditory stimulation protocol. Figure 1B1 shows 4D reconstructions and figure 1B2 shows 5D reconstructions.
Results:
All subjects showed a bilateral activation along the superior temporal gyrus adjacently to the Heschl's gyrus and restricted to the gray matter (for coordinates in MNI space of each cluster see Fig 2b). Maps represent the signal change intensity in all statistically significant clusters. The signal dynamics associated with the activations were then extracted and shown to follow the classical hemodynamic response function (HRF). A normalized average of these HRFs is shown in figure 2c.
·Figure 2a shows all the active clusters of each subject; figure 2b shows their coordinates and figure 2c displays the average (mean±std) across subjects of the HRFs extracted from the clusters.
Conclusions:
The present paradigm proved that Hi-Fi fMRI can record activation with submillimeter spatial resolution and located into the gray matter of the superior temporal gyrus involved in processing natural sounds10. Moreover, the signal HI-FI is sensitive to follows the classical HRF. While the temporal resolution extracted was 2.5sec, the high sampling-rate and its radial trajectory allow for a flexible reconstruction of the images with time bins of different sizes. These results support the validity of the HI-FI approach as well as its broad applicability.
The research was supported by the MYSpace project (PI Monica Gori), which has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program (grant agreement No 948349).
The research was supported by the MYSpace project (PI Monica Gori), which has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program (grant agreement No 948349).
Modeling and Analysis Methods:
Methods Development
Novel Imaging Acquisition Methods:
BOLD fMRI 1
Perception, Attention and Motor Behavior:
Perception: Auditory/ Vestibular 2
Keywords:
Cortex
FUNCTIONAL MRI
Other - Auditory, Spiral phyllotaxis, SNR, Motion Correction
1|2Indicates the priority used for review
Provide references using author date format
2 Shafieizargar, B., Dekker, A. J. Den, Sijbers, J., & Klein, S. (2023). Systematic review of reconstruction techniques for accelerated quantitative MRI. June, 1172–1208. https://doi.org/10.1002/mrm.29721
3 Piccini, D., Littmann, A., Nielles-vallespin, S., & Zenge, M. O. (2011). Spiral Phyllotaxis : The Natural Way to Construct a 3D Radial Trajectory in MRI. 1056, 1049–1056. https://doi.org/10.1002/mrm.22898
4 Feng, L., Axel, L., Chandarana, H., Block, K. T., Sodickson, D. K., & Otazo, R. (2016). XD‐GRASP: golden‐angle radial MRI with reconstruction of extra motion‐state dimensions using compressed sensing. Magnetic resonance in medicine, 75(2), 775-788.
5 Roy, C. W., Heerfordt, J., Piccini, D., Rossi, G., Pavon, A. G., Schwitter, J., & Stuber, M. (2021). Motion compensated whole-heart coronary cardiovascular magnetic resonance angiography using focused navigation (fNAV). Journal of Cardiovascular Magnetic Resonance, 23, 1-17.
6 Franceschiello, B., Rumac, S., Hilbert, T., Nau, M., Dziadosz, M., Degano, G., Roy, C. W., Gaglianese, A., Petri, G., Stuber, M., Kober, T., Van, R. B., Murray, M. M., & Fornari, E. (2023). Technical report: whole-brain BOLD functional imaging at 3T MRI. im(6), 2–3.
7 Franceschiello, B., Rumac, S., Hilbert, T., Nau, M., Dziadosz, M., Degano, G., Roy, C. W., Gaglianese, A., Petri, G., Stuber, M., Kober, T., Van, R. B., Murray, M. M., & Fornari, E. (2023). Hi-Fi fMRI: High-resolution, fast-sampled and sub-second whole-brain functional MRI at 3T in humans. 8, 1–15.
8 Meuli, R. A., Maeder, P., Pittett, A., Adriani, M., Fornari, E., Thirano, J., & Clarke, S. (2000). OPTIMISATION OF STIMULI AND ACQUISITION TECHNIQUE FOR fMR1 OF THE AUDITORY SYSTEM. 5, 2000.
9 Andersson, J. et al. Statistical Parametric Mapping: The Analysis of Functional Brain Images. (ACADEMIC PRESS, 2006).
10 Moerel, M., De Martino, F., & Formisano, E. (2012). Processing of natural sounds in human auditory cortex: tonotopy, spectral tuning, and relation to voice sensitivity. Journal of Neuroscience, 32(41), 14205-14216.