Head of the Magnetic Resonance Center

Prof. Dr. Klaus Scheffler


Secretary: Tina Schröder
Phone: +49 7071 601-701
Fax: +49 7071 601-702

Current and former Lab members

     PD. Dr. Gabriele Lohmann (Project Leader)
     Dr. Philipp Ehses (PostDoc)
     Christian Mirkes (PhD Student)
     Mario Baez (PhD Student)
     Ali Agheifar (PhD Student)
     Alexander Loktyusin (PhD Student)
     Marlon Arturo Perez Rodas (PhD Student)


Sequences and Signals

Our primary goal is to develop novel magnetic resonance techniques that are able to specifically probe the structural and biochemical composition of living tissue. This is closely linked with our interest to understand the details of magnetic resonance signal formation within a living environment, as nuclear magnetization is continuously influenced by different processes during its live time between excitation and relaxation. This is a simple, probably computationally demanding task, since we just have to forward the tiny fluctuating magnetic fields, which are sensed by the water during its random or oriented walk through tissue, to the Bloch or similar equations. A prominent example is the detection of neuronal activation with magnetic resonance, often called functional MRI or fMRI: increased neuronal activation increases the observed magnetic resonance signal, and sometimes vice versa. This BOLD effect currently is the working horse of numerous applications in cognitive and systems neurosciences. 
Since the last years the focus if this research group is to explore new ways, still based on MR, to detect and process neuronal activation. To reach this goal, several projects listed below are conducted in parallel to overcome the obsolete EPI technique.

Back to balanced SSFP

In 2006 we introduced a novel concept based on balanced SSFP to acquire functional maps, which was further developed by several other groups. Besides several papers on the basic spin physics of this approach, bSSFP so far has never been used as a brain mapping technique for neuroscientists. We have recently implemented (and modified) this technique at 9.4T and 7T, and were able to produce highly reproducible activation maps. The stability or temporal SNR was superior to EPI based on the completely balanced and thus flow and motion compensated nature of bSSFP. Preliminary comparisons to GE-EPI and SE-EPI suggest a mainly T2-related signal behavior (and diffusion effects) with reduced sensitivity to larger draining veins. So far, the temporal resolution is about 3 times below that of EPI, but advanced readouts as used for sodium imaging will increase the ADC duty cycle into the range of EPI.

Signal formation

The mechanism of signal generation within the neuronal network remains partially unclear. A major future goal is to model signal formation in tissue based on precise anatomical and physiological models, taking all parameters into account that might disturb the spins. On a microscopic scale these parameters are not directly accessible with MR, and we thus aim to either use high-resolution, ex-vivo microscopy information and/or to simultaneously combine MR with optical methods such as two-photon microscopy in vivo, as well as electrophysiological recordings. These additional parameters and detailed anatomical structures will help us via Monte Carlo simulations to understand the superb functional signal of bSSFP and other sequences at 9.4T.

Trying to eliminate unwanted signals: motion and shim

Beside the fundamental physiological limit in exposure of living tissue to changing electromagnetic fields, which basically sets a limit in SNR and acquisition speed, the measured signal itself is heavily disturbed by physiological and scanner fluctuations superimposed to the potentially interesting signal. To assess variations in B0 (caused mainly by breathing and motion) and gradient imperfections we recycled our in 1996 published method of using small NMR probes. Based on a running DFG project these probes will be fully integrated into a single chip (CMOS ASIC technology) and connected to an external acquisition system by fiber optics. In combination with optical tracking of motion (Kineticor system) we aim to acquire a comprehensive overview of B0 instabilities, and feed this information back to the scanner hardware and reconstruction. In addition, we will develop models to estimate the B0 distribution inside the head based on its position and surrounding B0 fields. This will involve the construction of an integrated matrix shim located and integrated within the 16-channel transmit array. Integration of local field probes, an array of loop shim coils and optical motion tracking into one coil system combined with appropriate feedback mechanisms will provide an all-in-one solution to capture and compensate physiological fluctuations. Combination of bSSFP with the aforementioned monitoring and correction for physiological noise and advanced matrix shim will probably set a new dimension in functional MRI.

Without gradients

Spatial resolution requires excessive switching of gradients and is thus time consuming. The observation of a free induction decay (which might be modulated be very rapid physiological changes within the acquired volume) from only a tiny volume is thus impossible except if small micro coils with intrinsic volume selection are used. Funded by DFG we will explore the possibility to measure proton FIDs with micro coils combined with electrophysiological recordings. The coils with integrated transmit-receive circuits are integrated within a 100 mm silicon needle which will be placed in the rat or monkey cortex. The ultimate goal is to observe fast, non-BOLD related modulations in the decaying FID as well as fast acquisition of neurotransmitter changes with proton spectroscopy. Using balanced SSFP without applying any gradients produces an infinite FID that continuously can sense any field fluctuations appearing within the small detection volume of the coil

Novel approaches for fMRI data analysis

Using different acquisition techniques then EPI and looking for deep brain responses requires novel methods to detect activation. The dynamic MR response function to activation not only depends on the brain structure and location, but also on the used MR method. In the past twenty years, task-based fMRI studies have primarily focused on signal amplitude changes, or on connectivity related to few selected nodes. We will follow an alternative view on fMRI data by shifting the focus away from signal amplitudes towards large-scale, task-induced synchronization networks. We developed a new data analysis algorithm called ``SyncMap'' that is designed to identify sets of brain locations that collectively synchronize in response to a task while being spatially connected in an underlying network. SyncMap does not require a presegmentation of the data for dimensionality reduction as it can handle large networks consisting of thousands of nodes. Because its conceptual basis is task-induced synchronization it does not depend on a hemodynamic response model. SyncMap identified several task-specific, large-scale patterns of synchronization. The strongest network hubs coincided with the sites of highest BOLD amplitude changes. However, additional hubs emerged where there were no significant amplitude changes, and several hubs even exhibited antagonistic behavior. This provides an entirely new window into the immense complexity of human brain function by taking large-scale synchronization patterns into account.
Last updated: Tuesday, 10.02.2015