Understanding the normal and diseased human brain crucially depends on reliable knowledge of its anatomical microstructure and functional micro-organization. Even subtle changes at the microscopic level can cause debilitating diseases. To resolve the microstructure and its changes in health and disease, unprecedented spatial resolution, minimal artifact levels and high tissue specificity of the imaging are essential. To address these methodological challenges, the MR-Methods group pursues an interdisciplinary approach, developing novel MRI and fMRI acquisition methods, image reconstruction and processing techniques.

Advanced MR Reconstruction

Gillad Liberman, Deni Kurban, Dimo Ivanov, Benedikt Poser

Advanced MRI scanning at high fields comes along with multiple challenges: geometric distortions, noise amplification with parallel imaging methods (GRAPPA, SMS), spatio-temporal field instabilities, echo-time constraints, signal blurring, and many others. Many of these challenges can be addressed with advanced readout trajectory modifications and corresponding modifications in the image reconstruction. The MR-Methods group is working on multiple of those advanced reconstitution methods:

• Alteration of GRAPPA reference line acquisition order from segmented EPI to FLASH and FLEET approaches has been proven to provide superior insensitivity of field perturbations. As such, the MR-Methods group is investigating the optimal GRAPPA reconstruction parameter space in ASL, VASO and BOLD fMRI.
• Spiral trajectories can be beneficial for BOLD and non-BOLD fMRI due to their TE-flexibility and efficient acceleration capabilities. However, their reconstruction is computational expensive and requires the knowledge of the exact gradient wave forms. The the MR-Methods group is working on more efficient reconstruction methods by employing Minimal Neural Networks and Skope field cameras.

Example trajectory of an SMS-spiral readout with intermediate blips

Layer-(f)MRI

Renzo Huber, Sri Kashyap, Dimo Ivanov, Benedikt Poser

In recent years, there have been several highly publicized studies showing the value of high-field (≥7 Tesla), high-resolution (functional) MRI for human neuroscience. With recent advancements, high-field (f)MRI allows researchers to approach the spatial scale of cortical layers and cortical columns. This revolutionizes the ability to tackle cortical information processing at the mesoscale (< 1mm), closing the gap with invasive animal electrophysiology. With layer-fMRI, neuroscientists can ask questions about directional information flow, about afferent vs. efferent connectivity, and about feed-forward vs. feed-back input between brain areas. However, the method of layer-fMRI has just started to prove its applicability in neuroscientist research studies and will benefit from further development to improve its interpretability, user-friendliness and specificity-sensitivity advancements.

Example of layer dependent activity pattern in the motor cortex.

The MR-Methods group is working on obtaining a better understanding of the signal generation mechanism in layer-fMRI and implementing advanced imaging and analysis tools for efficient mapping of layer activity:

• Pushing the limits of fMRI resolution up to 100 micro meter voxels.
• Investigating the spatio-temporal layer signal across luminance contrasts in V1/V2
• Developing evaluation tools to map layer-fMRI signals across topographical representations within brain areas.
• Reviewing the applicability of non-BOLD fMRI methods: ASL, VASO, Calibrated BOLD.
• Investigating the input/output microcircuits in the primary motor system.
• Investigating the applicability of layer-fMRI in cognitive brain areas: DLPFC

Non-BOLD fMRI: CBF and CBV

Dimo Ivanov, Renzo Huber, Sri Kashyap, Benedikt Poser

While BOLD fMRI revolutionized the ability to tackle cortical information processing across brain systems, one of the major limitations of conventional gradient-echo BOLD fMRI is that it probes neuronal activity changes only indirectly via evoked cerebral blood flow (CBF), cerebral blood volume (CBV) and cerebral metabolic rate of oxygen (CMRO2) changes. Thus, conventional GE-BOLD is limited by its quantifiability and localization specificity. Measuring CBF, CBV or CMRO2 directly on a laminar and columnar level could overcome these limitations. However, the MR-Methods group showed in a recent review paper that their acquisition procedures come along with technical challenges and sensitivity constraints.

The MR-Methods group is working on making non-BOLD fMRI methods more applicable to address neuroscientists’ research questions. In recent papers, we could improve the non-BOLD fMRI sequences by combining contrast-generating adiabatic preparation pulses with efficient signal readout modules:

• Sub-millimeter fMRI with large coverage laminar ASL.
• Investigating the optimal ASL approaches for high-field non-BOLD fMRI
• Increasing the CBV-VASO fMRI coverage by a factor of 3-5 with SMS-readouts
• Optimizing parallel imaging parameters for high-field CBF mapping
• Increasing sub-millimeter sensitivity of CBV-VASO fMRI with 3D-EPI readouts

Ultra High Field Imaging

Benedikt Poser, Sri Kashyap, Renzo Huber, Dimo Ivanov

The SNR and CNR benefits of ultra-high field have helped push the envelope of achievable spatial resolution in MRI. However, the push for high resolution comes at a cost of a large encoding burden leading to very lengthy scans. The MR-Methods group is working on UHF-specific developments in parallel imaging with controlled aliasing and the move away from 2D slice-by-slice imaging to much more SNR-efficient simultaneous multi-slice (SMS) and 3D acquisitions to address this issue. Furthermore, the MR-Methods group is working on the implementation of new UHF RF pulse classes to tackle the B1+ and SAR issues of UHF and the increased SAR and power requirement of SMS RF pulses.

Current projects involve:

• Developing new workflows to overcome the challenges of B0 and RF excitation inhomogeneity at ultra-high field MRI with volumetric B0 and flip-angle homogenisation at 9.4 T.
• Optimizing pulse sequences and parallel imaging for high spatio-temporal resolution MRI at ultra-high field
• Pushing the capabilities of ultra-high resolution blood volume fMRI and BOLD fMRI in humans at 9.4Understanding the normal and diseased human brain crucially depends on reliable knowledge of its anatomical microstructure and functional micro-organization. Even subtle changes at the microscopic level can cause debilitating diseases. To resolve the microstructure and its changes in health and disease, unprecedented spatial resolution, minimal artifact levels and high tissue specificity of the imaging are essential. To address these methodological challenges, the MR-Methods group pursues an interdisciplinary approach, developing novel MRI and fMRI acquisition methods, image reconstruction and processing techniques.
• Developing high-resolution gradient-recalled echo imaging methods at 9.4T using 16-channel parallel transmit simultaneous multislice spokes excitations with slice-by-slice flip angle homogenization.

AROMA

Alejandro Monreal, Benedikt Poser