Publications

{Introduction: Traditional models of learning and memory consolidation postulate two interacting memory systems, with rapid encoding supported by the hippocampus and only gradually developing, stable storage in neocortical circuits (McClelland et al., 1995). There is an ongoing debate about which neocortical regions subserve long-term memory consolidation, including the idea, that the same regions originally involved in processing the input later provide the neural substrate for the memory trace. In a recently published fMRI study we have shown rapidly emerging neocortical memory-related activity in the posterior parietal cortex (PPC) that over learning repetitions becomes independent of hippocampal signaling and fulfills all criteria for a long-term memory representation (Brodt et al., 2016). Besides changes in functional activity, the site where a memory representation is stored for the long-term should also undergo structural changes. These changes can be assessed by diffusion MRI already several hours after learning (Sagi et al., 2012). In the current study, we investigated functional and structural changes in the neocortex over the course of learning. Methods: Two groups of human subjects (n\textequals41) learned object-place associations over 8 learning-recall repetitions in two sessions spaced 13 hours apart. Neural activity during learning and recall was tracked with fMRI. To assess structural changes, dMRI was acquired at three time points: immediately before the first learning session, 90 minutes after the first learning session and again before the second learning session. Results: Confirming our previous results, functional activity in the PPC, specifically in the precuneus, increases rapidly over learning repetitions, is stable over an offline period and mirrors the progression of recall performance rates. The same holds true for functional activity during recall. Concerning structural changes, when controlling for circadian effects, PPC areas as well as other areas along the dorsal and ventral processing streams show a marked decrease in mean diffusivity after the first learning session, which is stable for at least 12 hours and correlates with performance improvement between the task sessions. Conclusions: The simultaneous investigation of functional and structural changes confirms the rapid build-up of long-term memory representations in areas involved in the original processing of later remembered stimuli and emphasize the special role that posterior parietal areas play for memory function.}

{Functional neuroimaging studies have led to understanding the brain as a collection of spatially segregated functional networks. It is thought that each of these networks is in turn composed of a set of distinct sub-regions that together support each network\textquotesingles function. Considering the sub-regions to be an essential part of the brain\textquotesingles functional architecture, several strategies have been put forward that aim at identifying the functional sub-units of the brain by means of functional parcellations. Current parcellation strategies typically employ a bottom-up strategy, creating a parcellation by clustering smaller units. We propose a novel top-down parcellation strategy, using time courses of instantaneous connectivity to subdivide an initial region of interest into sub-regions. We use split-half reproducibility to choose the optimal number of sub-regions. We apply our Instantaneous Connectivity Parcellation (ICP) strategy on high-quality resting-state FMRI data, and demonstrate the ability to generate parcellations for thalamus, entorhinal cortex, motor cortex, and subcortex including brainstem and striatum. We evaluate the subdivisions against available cytoarchitecture maps to show that our parcellation strategy recovers biologically valid subdivisions that adhere to known cytoarchitectural features.}

{Functional 31P spectroscopy has been investigated in several studies with greatly varying results, which may be due to the low sensitivity of the 31P nucleus. We have taken advantage of the high SNR at 9.4 T to acquire spectra from the human visual cortex under stimulation. Experiments were performed with different localization volumes, defined by saturation pulses. In spite of the excellent quality of the obtained data, no stimulation-related changes in metabolite concentrations or resonance frequencies could be detected.}

{Anxiety disorders are common and debilitating conditions with higher prevalence in women. However, factors that predispose women to anxiety phenotypes are not clarified. Here we investigated potential contribution of the single nucleotide polymorphism rs2236418 in GAD2 gene to changes in regional inhibition/excitation balance, anxiety\textemdashlike traits and related neural activity in both sexes. 105 healthy individuals were examined with high-field (7T) multimodal magnetic resonance imaging (MRI), including resting state fMRI in combination with assessment of GABA and Glutamate (Glu) levels via MR spectroscopy (MRS). Regional GABA/Glu levels in ACC subregions were assessed as mediators of gene\textemdashpersonality interaction for the trait harm avoidance and moderation by sex was tested. In AA homozygotes, with putatively lower GAD2 promoter activity, we observed increased intrinsic neuronal activity and higher inhibition/excitation balance in pregenual ACC (pgACC), as compared to G carriers. The pgACC drove a significant interaction of genotype, region and sex, where inhibition/excitation balance was significantly reduced only in female AA carriers. This finding was specific for rs2236418 as other investigated SNPs of the GABA synthesis related enzymes (GAD1, GAD2 and GLS) were not significant. Furthermore, only in women there was a negative association of pgACC GABA/Glu ratios with harm avoidance. A moderated\textemdashmediation model revealed that pgACC GABA/Glu also mediated the association between the genotype variant and level of harm avoidance, dependent on sex. Our data thus provide new insights into the neurochemical mechanisms that control emotional endophenotypes in humans and constitute predisposing factors for the development of anxiety disorders in women.}

{It is our great pleasure to welcome you to the 25th IEEE Conference on Virtual Reality and 3D User Interfaces (IEEE VR), the premiere international conference focused on research in these domains! This year the name of the conference changed since the highly successful 12-year IEEE Symposium on 3D User Interfaces is brought back into the main VR conference. It is a tremendous honor to host the conference in Reutlingen on its first return to Germany since 2005. Reutlingen is situated nicely with respect to our research institutes, the city has the charm of a southern German town, and the recently built (2013) Stadthalle and concert venue has a professional team that is experienced in conferences highlighting innovation in technology. This venue has enabled us to create a large program both in scientific content and in terms of demonstration and exhibit space.}

{Brain information processing likely relies on cooperative interactions between neural populations at multiple scales. Growing evidence suggests that network oscillations, as observed in Local Field Potentials (LFP), are instrumental to the spatiotemporal coordination of these interactions. Therefore, investigating the coupling between spatiotemporal patterns of LFP and spiking activity is instrumental to understand distributed neural information processing. Common approaches to investigate this coupling are restricted to pairwise spike-LFP interactions, which are suboptimal formodern datasets with hundreds of simultaneous recording sites. Capturing efficiently the overall spike-LFP coupling structure in this high dimensional setting is of paramount importance to exploit the full potential of modern electrophysiology recording techniques. We develop a Generalized Phase Locking Analysis (GPLA), a multivariate extension of phase locking analysis, by gathering pairwise complex phase locking information in a rectangular matrix and summarize its structure with the largest singular value and the corresponding singular vectors. Singular vectors represent the dominant LFP and spiking patterns and the singular value, called generalized Phase Locking Value (gPLV), characterizes the strength of the coupling between LFP and spike patterns. We further investigate statistical properties of the gPLV and develop a statistical testing framework. Compared to univariate pairwise approaches, simulations with networks of Leaky Integrate and Fire (LIF) neurons [1, 2] show that GPLA: (i) can reliably retrieve the coupling between spikes and LFP with lesser amount of data and (ii) exploits optimally the activity of multiple units to increase the statistical power while preserving individual coupling properties. Application to recordings from Utah arrays in macaque prefrontal cortex reveals a previously undetected large-scale coupling through an LFP traveling wave in the beta band (15\textendash30 Hz) synchronized with an array-wide synchronous spiking event. We hypothesize that it reflects a spatially distributed population with enhanced horizontal connectivity that activated by the incoming traveling wave. These results illustrate the interest of GPLA to assess global relationships between spatiotemporal patterns of spikes and network oscillations.}

{In environments where orientation is ambiguous, the visual system uses prior knowledge about lighting coming from above to recognize objects, determine which way is up, and reorient the body. Here we investigated the extent with which assumed light from above preferences are affected by body orientation and the orientation of the retina relative to gravity. We tested the ability to extract shape-from-shading with seven human male observers positioned in multiple orientations relative to gravity using a modified KUKA anthropomorphic robot arm. Observers made convex-concave judgments of a central monocularly viewed stimulus with orientations of a shading gradient consistent with being lit from one of 24 simulated illumination directions. By positioning observers in different roll-tilt orientations relative to gravity and when supine, we were able to monitor change in the light-from-above prior (the orientation at which a shaded disk appears maximally convex). The results confirm previous findings that the light-from-above prior changes with body orientation relative to gravity. Interestingly, the results varied also with retinal orientation as well as an additional component that was approximately twice the frequency of retinal orientation. We use a modelling approach to show that the data are well predicted by summing retinal orientation with cross-multiplied utricle and saccule signals of the vestibular system, yielding gravity-dependent biases in the ability to extract shape-from-shading. We conclude that priors such as light coming from above appear to be constantly updated by neural processes that monitor self-orientation to achieve optimal object recognition over moderate deviations from upright posture at the cost of poor recognition when extremely tilted relative to gravity.}

{The postulate of independence of cause and mechanism (ICM) has recently led to several new causal discovery algorithms. The interpretation of independence and the way it is utilized, however, varies across these methods. Our aim in this paper is to propose a group theoretic framework for ICM to unify and generalize these approaches. In our setting, the cause-mechanism relationship is assessed by perturbing it with random group transformations. We show that the group theoretic view encompasses previous ICM approaches and provides a very general tool to study the structure of data generating mechanisms with direct applications to machine learning.}

Magnetic Resonance Imaging (MRI) is a widely used imaging technology in medicine. Its advantages include good soft tissue contrast and the use of non-ionizing radiation in contrast to for example computed tomography (CT). One drawback are the long acquisition times that are needed. They depend on the diagnostic use case but are usually within the range of minutes. These long scan times make the images prone to patient motion during image acquisition which can lead to blurring or ghosting artifacts. Those artifacts might render the diagnostic value of the images useless which requires the image to be reacquired or the patient to be sedated before the scan to prevent motion artifacts. This is where motion correction comes into play. One can distinguish between retrospective and prospective motion correction (PMC) methods. Retrospective motion correction tries to improve image quality after the image acquisition by post-processing and possibly using additional motion tracking information, if available. Prospective motion correction relies on a motion tracking modality that is used to provide motion information to update imaging parameters during image acquisition. Both motion correction methods can also be used in combination with each other. This thesis, however, will focus on the implementation and validation of a system for prospective head motion correction. The system consisted of four nuclear magnetic resonance (NMR) field probes using. Those feld probes were attached to the head and used to measure the spatiotemporal evolution of magnetic felds. By switching spatially varying magnetic fields, this information can be used to track the field probes\textquotesingle positions and calculate the corresponding head motion in order to perform prospective motion correction.

{Magnetic resonance spectroscopic imaging (MRSI) is a promising technique for mapping the spatial distribution of multiple metabolites in the human brain. These metabolite maps can be used as a diagnostic tool to gain insight into several biochemical processes and diseases in the brain. In comparison to lower field strengths, MRSI at ultra-high field strengths benefits from a higher signal to noise ratio (SNR) as well as higher chemical shift dispersion, and hence spectral resolution. This study combines the benefits of an ultra-high field magnet with the advantages of an ultra-short TE and TR single-slice FID-MRSI sequence (such as negligible J-evolution and loss of SNR due to T2 relaxation effects) and presents the first metabolite maps acquired at 9.4 T in the healthy human brain at both high (voxel size of 97.6 $\micro$L) and ultra-high (voxel size of 24.4 $\micro$L) spatial resolutions in a scan time of 11 and 46 min respectively. In comparison to lower field strengths, more anatomically-detailed maps with higher SNR from a larger number of metabolites are shown. A total of 12 metabolites including glutamate (Glu), glutamine (Gln), N-acetyl-aspartyl-glutamate (NAAG), Gamma-aminobutyric acid (GABA) and glutathione (GSH) are reliably mapped. Comprehensive description of the methodology behind these maps is provided.}

{Introduction: Unarguably, the thalamus is a core structure of the human brain (Sherman, 2016), and all individual thalamic nuclei are integral components of very different functional systems in the brain (Jones, 2007). However, only a few studies have addressed the structural and functional diversity of the thalamic substructures in detail. Using diffusion tensor (DTI) or resting state functional MRI (r-fMRI) most 1.5, 3, and 7 Tesla MRI studies were able to determine 7 to 31 thalamic parcels with different spatial resolution (Calamante et al., 2012; Johansen-Berg et al., 2005; Kim et al., 2013; Kumar et al., 2017). Although most nuclei like the pulvinar complex, the lateral geniculate nucleus (LGN), and others are composed of numerous subnuclei (Baldwin et al., 2011, Andrews et al., 1997), we thought to examine whether a higher imaging resolution could assess a more detailed functional diversity of thalamic substructures. We, therefore, decided to use 1 mm isotopic resting state MRI data acquired at 9.4 Tesla to assess a high dimensional parcellation of the human thalamus. Methods: r-fMRI Acquisition: The r-fMRI was acquired at 9.4 T Siemens (Erlangen, Germany) in six right-handed male volunteers. We used the psf method for the distortion correction. The FOV covered thalamus (s. Fig. 1a) consisted of 45 slices, TR of 2.5 sec., 1 mm isotropic resolution and 220 scans. The volunteers kept their eyes closed during the resting state scans. Besides, two structural scans, i.e., MP2RAGE (600 microns iso) and a 3D GRE (400 microns iso) were acquired. r-fMRI Analysis: The preprocessing was performed with a modified pipeline using SPM12. It included slice timing, motion correction, coregistration, normalization and smoothing (3x3x3mm kernel). ICA Analysis: The normalized subject data were temporally concatenated. We performed the probabilistic-ICA (Beckmann et al., 2005) on the left and right thalamus. The optimal number of component estimation was done using default melodic model order selection. In the last step, the z-stat of all components were used to compute the winner map for the right and left thalamus. Results: The ICA automatic model order selection detected 82 components within the right thalamus and 83 components within the left thalamus (s. Fig 1b). In comparison with the histological atlas of Morel (Morel et al., 1997), which is restricted to a set of 29 bilateral nuclei (s. Fig. 1c), we observed an of 2.8 fold increase of parcels. Furthermore, our parcels varied in size, location, and distribution between both hemispheres (Fig. 2). This variable distribution with different temporal pattern within the left and right thalamus probably reflects functional differences between both hemispheres. The composite analysis in respect to the histological atlas revealed a varying number of parcels for each Morel nucleus; for example, we observed 6-7 left and right sub-parcels within the layered lateral geniculate nuclei (Andrews et al., 1997). Conclusions: Our study revealed that the thalamus exhibits a high-dimensional functional segregation even at rest. The detected parcels differed in size, location, and lateralization. In comparison with the histological defined thalamic nuclei, we observed a variable parcel assignment to all major nuclei groups in both hemispheres. However, further work is required to establish a valid and high-dimensional functional atlas of the thalamus, which could enhance our understanding of the concerted thalamo-cortical interaction at rest and under task conditions.}

{In this study, the acquisition of a high resolution (64x64) water reference MRSI data is accelerated by a factor of R\textequals14 using compressed sensing. The results show that this highly accelerated water reference can reliably be used for eddy current and phase correction purposes, as well as internal referencing and quantification. This enables the acquisition of the high resolution water reference MRSI data in 80 seconds at 9.4T.}

{Increasing evidence suggests a hypoglutamatergic state in major depressive disorder (MDD), however spatial- and metabolite specific abnormalities have not been fully characterized. Using short TE/TM STEAM MRS, we evaluated Glu, Gln, Gln/Glu and GABA metabolism in two histoarchitectonically distinct subdivisions of the anterior cingulate cortex (ACC). The pregenual ACC, involved in emotion processing, showed altered glutamine-glutamine cycling but not altered GABAergic metabolism in MDD, whereas no differences between patients and controls were found in the anteromedial ACC. Increased Gln/Glu in MDD in pgACC but not aMCC confirms a regionally specific role of altered glutamatergic metabolism and neuronal-glial interaction.}

{MRCDI is a novel technique for non-invasive measurement of weak currents in the human head, which is important in several neuroscience applications. Here, we present reliable in-vivo MRCDI measurements in the human brain based on SSFP-FID, yielding an unprecedented accuracy. We demonstrate the destructive influences of stray magnetic fields caused by the current passing through feeding cables, and propose a correction method. Also, we show inter-individual differences in MRCDI measurements for two different current profiles, and compare the measurements with simulations based on individualized head models. The simulations of the current-induced magnetic fields show good agreement with in-vivo brain measurements.}

{Among the multitude of elements making up visual experience, color stands out in that it can specify both subjective experience and objective properties of the outside world. Whereas most neuroimaging research on human color vision has focused on external stimulation, the present study addressed this duality by investigating how externally elicited color vision is linked to subjective color experience induced by object imagery. We recorded fMRI activity while showing our participants abstract color stimuli that were either red, green, or yellow in half of the runs (\textquotedblleftreal-color runs\textquotedblright) and asked them to produce mental images of colored objects corresponding to the same three categories in the remaining half (\textquotedblleftimagery runs\textquotedblright). To make sure that participants were engaged in visual imagery, they performed a 1-back same/different color judgment task on the imagined objects. We trained color classifiers using MVPA to distinguish between fMRI responses to the three color stimuli and cross-validated them on data from real-color or imagery runs. Although real-color percepts could be predicted from all retinotopically mapped visual areas, only color decoders trained on hV4 responses could additionally predict the color category of an object that was being imagined. This suggests that sensory-driven and self-induced colors share a common neural code in hV4. Using a hierarchical drift diffusion model, we furthermore demonstrated that the decoding accuracy in hV4 was predictive of performance in the color judgment task on a trial-by-trial basis. The commonality between neural representations of perceived and imagined object color, in combination with the behavioral modeling evidence, hence identifies area hV4 as a \textquotedblleftperceptual bridge\textquotedblright linking externally triggered color vision with color in self-generated object imagery.}

{Color is special among basic visual features in that it can form a defining part of objects that are engrained in our memory. Whereas most neuroimaging research on human color vision has focused on responses related to external stimulation, the present study investigated how sensory-driven color vision is linked to subjective color perception induced by object imagery. We recorded fMRI activity in male and female volunteers during viewing of abstract color stimuli that were red, green, or yellow in half of the runs. In the other half we asked them to produce mental images of colored, meaningful objects (such as tomato, grapes, banana) corresponding to the same three color categories. Although physically presented color could be decoded from all retinotopically mapped visual areas, only hV4 allowed predicting colors of imagined objects when classifiers were trained on responses to physical colors. Importantly, only neural signal in hV4 was predictive of behavioral performance in the color judgment task on a trial-by-trial basis. The commonality between neural representations of sensory-driven and imagined object color and the behavioral link to neural representations in hV4 identifies area hV4 as a perceptual hub linking externally triggered color vision with color in self-generated object imagery.}

{A key question in vision research concerns how the brain compensates for self-induced eye and head movements to form the world-centered, spatiotopic representations we perceive. Although human V3A and V6 integrate eye movements with vision, it is unclear which areas integrate head motion signals with visual retinotopic representations, as fMRI typically prevents head movement executions. Here we examined whether human early visual cortex V3A and V6 integrate these signals. A previously introduced paradigm allowed participant head movement during trials, but stabilized the head during data acquisition utilizing the delay between blood-oxygen-level-dependent (BOLD) and neural signals. Visual stimuli simulated either a stable environment or one with arbitrary head-coupled visual motion. Importantly, both conditions were matched in retinal and head motion. Contrasts revealed differential responses in human V6. Given the lack of vestibular responses in primate V6, these results suggest multi-modal integration of visual with neck efference copy signals or proprioception in V6.}

{A key question in vision research concerns visual stability: how is visual information in retinal coordinates integrated with non-visual cues of self-induced motion to form the spatiotopic representations of the world that we perceive? Eye movements have been found to modulate retinotopic representations at multiple stages along the visual stream, yet a special role has been attributed to human areas V3A and V6, as both cancel self-induced retinal planar motion during eye movements to a near complete extent (Fischer, B\"ulthoff, Logothetis, Bartels, 2012). Beyond that, only little is known about which human visual processing stages integrate head motion signals with retinotopic representations as human fMRI is typically incompatible with execution of voluntary head movements. We recently circumvented these limitations and introduced a novel paradigm that allows participants to move their heads during fMRI scanning (Schindler and Bartels, 2018). The functional characteristics of the BOLD signal allowed us to temporally decouple stimulus presentation from the acquisition of stimulus evoked responses. Our custom-built air pressure based head-stabilization system permitted head-rotation during trials, but stabilized head position during data acquisition. Video-based head-tracking and head-mounted goggles allowed for real-time generation of visual stimuli taking head-motion into account. Observers viewed approaching visual flow through head-mounted MR-compatible goggles. A congruent condition simulated constant forward motion while the observer rotated the head relative to the body, as when looking around while being driven along a straight road. In the incongruent condition, observers performed identical head rotations, but the visual consequences were inversed such that visual and extra-retinal cues did not combine in any meaningful way. Crucially, both conditions were matched regarding head and retinal motion. Based on this paradigm, we previously examined the integration of head motion and visual signals in regions with established vestibular processing. Here we asked whether early visual cortex as well as areas V3A and V6 may integrate retinotopic visual representations with voluntary head motion. Contrasting congruent versus incongruent conditions revealed differential responses in human V6 but not in early visual regions or V3A, consistent with multi-modal integration of visual cues with head motion in human area V6. Our results extend previous evidence for multimodal integration in V6 to head-motion cues and are in line with the hypothesis of V6 as a crucial hub for compensation of self-induced motion.}

{Purpose Information on the electrical tissue conductivity might be useful for the diagnosis and characterization of pathologies such as tumors [1]. MRCDI and MREIT are two emerging non-invasive techniques for imaging of weak currents and ohmic conductivities. In this study, we demonstrated human in vivo brain MRCDI to pave the way for its clinical use [2], [3]. Methods In short, weak alternating currents up to 1\textendash2 mA are injected into human head in synchrony with tailored phase-sensitive MRI. The currents create a magnetic field , which shifts the precession frequency of the magnetization and modulates the acquired MR images. The acquired images are used to measure and reconstruct the current flow and conductivity distributions. We employed a steady-state free precession free-induction-decay (SSFP-FID) sequence in five subjects, and injected currents of 1 mA by an MR-conditional current source via electrodes attached to the scalp (two current profiles: Right-left (RL), electrodes placed near the temporoparietal junctions; anterior-posterior (AP), one attached to the forehead and one above the inion). Additionally, an ultra-short-echo-time sequence was performed to track the feeding cables for correcting the stray magnetic fields induced by cable currents. Corrected measurements were used to calculate current flow distributions and compared with Finite-Element simulations of the current flow based on individualized head models [4]. Results The current-induced magnetic field with was reliably measured and the reconstructed current flows showed good agreement with the simulations (average coefficient of determination R2\textequals 71\textpercent). The injected current flow differed substantially among individuals according to the electrode placements and anatomical differences. The calculated currents are stronger in CSF-filled highly conductive regions, e.g. the longitudinal fissure. Conclusions The strong correlation between the simulations and measurements validates the accuracy of the method and demonstrates the potential of the method for determining accurate brain tissue conductivities. These initial current flow recordings pave the way for human brain MREIT that might complement standard MR methods for tumor characterization.}

{Magnetic resonance current density imaging (MRCDI) and MR electrical impedance tomography (MREIT) are two emerging modalities, which combine weak time-varying currents injected via surface electrodes with magnetic resonance imaging (MRI) to acquire information about the current flow and ohmic conductivity distribution at high spatial resolution. The injected current flow creates a magnetic field in the head, and the component of the induced magnetic field $\Delta$Bz,c parallel to the main scanner field causes small shifts in the precession frequency of the magnetization. The measured MRI signal is modulated by these shifts, allowing to determine $\Delta$Bz,c for the reconstruction of the current flow and ohmic conductivity. Here, we demonstrate reliable $\Delta$Bz,c measurements in-vivo in the human brain based on multi-echo spin echo (MESE) and steady-state free precession free induction decay (SSFP-FID) sequences. In a series of experiments, we optimize their robustness for in-vivo measurements while maintaining a good sensitivity to the current-induced fields. We validate both methods by assessing the linearity of the measured $\Delta$Bz,c with respect to the current strength. For the more efficient SSFP-FID measurements, we demonstrate a strong influence of magnetic stray fields on the $\Delta$Bz,c images, caused by non-ideal paths of the electrode cables, and validate a correction method. Finally, we perform measurements with two different current injection profiles in five subjects. We demonstrate reliable recordings of $\Delta$Bz,c fields as weak as 1 nT, caused by currents of 1 mA strength. Comparison of the $\Delta$Bz,c measurements with simulated $\Delta$Bz,c images based on FEM calculations and individualized head models reveals significant linear correlations in all subjects, but only for the stray field-corrected data. As final step, we reconstruct current density distributions from the measured and simulated $\Delta$Bz,c data. Reconstructions from non-corrected $\Delta$Bz,c measurements systematically overestimate the current densities. Comparing the current densities reconstructed from corrected $\Delta$Bz,c measurements and from simulated $\Delta$Bz,c images reveals an average coefficient of determination R2 of 71\textpercent. In addition, it shows that the simulations underestimated the current strength on average by 24\textpercent. Our results open up the possibility of using MRI to systematically validate and optimize numerical field simulations that play an important role in several neuroscience applications, such as transcranial brain stimulation, and electro- and magnetoencephalography.}

{Studies of human and rodent navigation often reveal a remarkable cross-species similarity between the cognitive and neural mechanisms of navigation. Such cross-species resemblance often overshadows some critical differences between how humans and nonhuman animals navigate. In this review, I propose that a navigation system requires both a storage system (i.e., representing spatial information) and a positioning system (i.e., sensing spatial information) to operate. I then argue that the way humans represent spatial information is different from that inferred from the cellular activity observed during rodent navigation. Such difference spans the whole hierarchy of spatial representation, from representing the structure of an environment to the representation of sub-regions of an environment, routes and paths, and the distance and direction relative to a goal location. These cross-species inconsistencies suggest that what we learned from rodent navigation does not always transfer to human navigation. Finally, I argue for closing the loop for the dominant, unidirectional animal-to-human approach in navigation research, so that insights from behavioral studies of human navigation may also flow back to shed light on the cellular mechanisms of navigation for both humans and other mammals (i.e., a human-to-animal approach).}

{Despite its relevance for navigation surprisingly little is known about how goal direction bearings to distant locations are computed. Behavioural and neuroscientific models proposing the path integration of previously navigated routes are supported indirectly by neural data, but behavioral evidence is lacking. We show that humans integrate navigated routes post-hoc and incrementally while conducting goal direction estimates. Participants learned a multi-corridor layout by walking through a virtual environment. Throughout learning, participants repeatedly performed pairwise pointing from the start location, end location, and each turn location between segments. Pointing latency increased with the number of corridors to the target and decreased with pointing experience rather than environmental familiarity. Bimodal pointing distributions indicate that participants made systematic errors, for example, mixing up turns or forgetting segments. Modeling these error sources suggests that pointing did not rely on one unified, but rather multiple representations of the experimental environment. We conclude that participants performed incremental on-the-fly calculations of goal direction estimates within compartmentalised representations, which was quicker for nearby goals and became faster with repeated pointing. Within navigated environments humans do not compute difference vectors from coordinates of a globally consistent integrated \textquotedblleftmap in the head\textquotedblright.}

{Purpose Treatments using high-intensity focused ultrasound (HIFU) in the abdominal region remain challenging as a result of respiratory organ motion. A novel method is described here to achieve 3D motion-compensated ultrasound (US) MR-guided HIFU therapy using simultaneous ultrasound and MRI. Methods A truly hybrid US-MR-guided HIFU method was used to plan and control the treatment. Two-dimensional ultrasound was used in real time to enable tracking of the motion in the coronal plane, whereas an MR pencil-beam navigator was used to detect anterior\textendashposterior motion. Prospective motion compensation of proton resonance frequency shift (PRFS) thermometry and HIFU electronic beam steering were achieved. Results The 3D prospective motion-corrected PRFS temperature maps showed reduced intrascan ghosting artifacts, a high signal-to-noise ratio, and low geometric distortion. The k-space data yielded a consistent temperature-dependent PRFS effect, matching the gold standard thermometry within approximately 1\mbox{$^\circ$}C. The maximum in-plane temperature elevation ex vivo was improved by a factor of 2. Baseline thermometry acquired in volunteers indicated reduction of residual motion, together with an accuracy/precision of near-harmonic referenceless PRFS thermometry on the order of 0.5/1.0\mbox{$^\circ$}C. Conclusions Hybrid US-MR-guided HIFU ablation with 3D motion compensation was demonstrated ex vivo together with a stable referenceless PRFS thermometry baseline in healthy volunteer liver acquisitions.}

{Manual assembly at production is a mentally demanding task. With rapid prototyping and smaller production lot sizes, this results in frequent changes of assembly instructions that have to be memorized by workers. Assistive systems compensate this increase in mental workload by providing "just-in-time" assembly instructions through in-situ projections. The implementation of such systems and their benefits to reducing mental workload have previously been justified with self-perceived ratings. However, there is no evidence by objective measures if mental workload is reduced by in-situ assistance. In our work, we showcase electroencephalography (EEG) as a complementary evaluation tool to assess cognitive workload placed by two different assistive systems in an assembly task, namely paper instructions and in-situ projections. We identified the individual EEG bandwidth that varied with changes in working memory load. We show, that changes in the EEG bandwidth are found between paper instructions and in-situ projections, indicating that they reduce working memory compared to paper instructions. Our work contributes by demonstrating how design claims of cognitive demand can be validated. Moreover, it directly evaluates the use of assistive systems for delivering context-aware information. We analyze the characteristics of EEG as real-time assessment for cognitive workload to provide insights regarding the mental demand placed by assistive systems.}

{Introduction: Brain morphometry studies typically utilize anatomical data with 1 mm3 isotropic resolution. However, the widespread availability of high-field MRI scanners and receive coil arrays have led to increased interest in submillimeter resolution data. While higher resolution may have an advantage (Bazin et al., 2014; L\"usebrink et al., 2013; Zaretskaya et al., 2017), lifting this constraint on voxel size opens the door to a variety of acquisition protocols with a wide range of image contrast and SNR. Here we aim to systematically investigate the effects of image SNR on surface reconstruction quality based on sub-millimeter MPRAGE data acquired at 3T. Methods: To systematically vary image SNR without changing voxel size, we acquired multiple repetitions of high-resolution (0.6 mm isotropic) MPRAGE data from 9 participants, and for each subject generated multiple image volumes with progressively higher SNR by averaging together increasing number of repetitions (i.e. averaging 1 to 6 (n\textequals5) or 1 to 8 (n\textequals4) repetitions, depending on the total number of repetitions acquired within the session). Images were acquired on a 3T Siemens MAGNETOM Trio Tim system using a 32 channel coil and a 0.6 mm isotropic multiecho MPRAGE (van der Kouwe et al., 2008) protocol (TR/TE1/TE2/TI/FA/BW/ESP/matrix\textequals2510 ms/2.88 ms/ 5.6 ms/1200 ms/7\mbox{$^\circ$}/420 Hz/px/8.4 ms/400$\times$400). To minimize blurring during the inversion recovery, we used acceleration in the partition direction (R\textequals2) without partial Fourier in any direction and a slab-selective axial acquisition (slices per slab\textequals224) to minimize the number of partition encoding steps. Each average volume was used to generate cortical surfaces using FreeSurfer\textquotesingles native "hires" sub-millimeter reconstruction stream (Zaretskaya et al., 2017). To assess surface quality, for each set of surfaces (the gray-white and gray-CSF interface surfaces of each hemisphere) we computed the following parameters: (1) image SNR (the mean divided by the standard deviation of voxel intensities within the FreeSurfer white matter mask), (2) surface smoothness (defined as the median value of per-vertex local mean curvature), (3) number of topological defects in the initial surface, identified by FreeSurfer (Segonne et al., 2007), (4) median defect size. We also report gray-white matter contrast, derived from the reconstruction with 6 repetitions, and defined as the median of vertex-wise contrast values expressed as (white matter intensity \textendash gray matter intensity)/(white matter intensity + gray matter intensity)$\cdot$100. Results: Automatic reconstruction completed successfully in all but 2 cases. Both failed cases were comprised of a single repetition and could not complete automatically due to low SNR. The gray-white matter contrast of our data, based on averaging 6 repetitions for each subject, was 26.8 $\pm$ 0.4 (mean across subjects $\pm$ S.E.M). We observed a consistent increase in image SNR - accompanied by an increase in surface smoothness and by a decrease in the number of topological defects - with increasing number of repetitions included in the average. There was no change in the median defect size. While surface quality clearly improved with increasing SNR, gray-white surface smoothness and the number of defects did not reach a clear plateau even after 7 repetitions, suggesting that more repetitions may be needed to determine the SNR beyond which there is no noticeable improvement in surface quality. However, for this protocol and scanner, and across this small group of subjects, we find that qualitatively the performance of the surface reconstruction is adequate after averaging 5\textendash6 repetitions. Conclusions: Overall we show that averaging multiple repetitions can compensate for the corresponding loss of image SNR due to increase in image resolution. High-quality reconstructions can hence be generated from voxels far smaller than conventional acquisitions, even at standard field strength, provided sufficient data quality.}

{Reliable assessment and analysis of spontaneous mechanical activities in musculature (SMAM) visible in repetitive DWI is a relatively new technique for non-invasive characterization of skeletal musculature. To correct for data corrupted by intentional contractions, a surface electromyography-based contraction state analysis was investigated to reject undesired DWI data. It is demonstrated that the presented method enables a more reliable quantification of SMAMs and improved spontaneous activity maps.}

{Purpose To perform exchange-rate measurements on the in vivo human brain downfield spectrum (5\textendash10 ppm) at 9.4 T and to compare the variation in concentrations of the downfield resonances and of known upfield metabolites to determine potential peak labels. Methods Non-water-suppressed metabolite cycling was used in combination with an inversion transfer technique in two brain locations in healthy volunteers to measure the exchange rates and T1 values of exchanging peaks. Spectra were fitted with a heuristic model of a series of 13 or 14 Voigt lines, and a Bloch\textendashMcConnell model was used to fit the exchange rate curves. Concentrations from non-water-inverted spectra upfield and downfield were compared. Results Mean T1 values ranged from 0.40 to 0.77 s, and exchange rates from 0.74 to 13.8 s\textminus1. There were no significant correlations between downfield and upfield concentrations, except for N-acetylaspartate, with a correlation coefficient of 0.63 and P \textless 0.01. Conclusions Using ultrahigh field allowed improved separation of peaks in the 8.2 to 8.5 ppm amide proton region, and the exchange rates of multiple downfield resonances including the 5.8-ppm peak, previously tentatively assigned to urea, were measured in vivo in human brain. Downfield peaks consisted of overlapping components, and largely missing correlations between upfield and downfield resonances\textemdashalthough not conclusive\textemdashindicate limited contributions from metabolites present upfield to the downfield spectrum. Magn Reson Med, 2017. \copyright 2017 International Society for Magnetic Resonance in Medicine.}

{Purpose The transverse relaxation times T2 of 17 metabolites in vivo at 3T is reported and region specific differences are addressed. Methods An echo-time series protocol was applied to one, two, or three volumes of interest with different fraction of white and gray matter including a total number of 106 healthy volunteers and acquiring a total number of 128 spectra. The data were fitted with the 2D fitting tool ProFit2, which included individual line shape modeling for all metabolites and allowed the T2 calculation of 28 moieties of 17 metabolites. Results The T2 of 10 metabolites and their moieties have been reported for the first time. Region specific T2 differences in white and gray matter enriched tissue occur in 16 of 17 metabolites examined including single resonance lines and coupled spin systems. Conclusion The relaxation time T2 is regions specific and has to be considered when applying tissue composition correction for internal water referencing.}

{We investigated the possibility to observe the anatomical details of the superior colliculus (SC), a layered structure located on the tectum of the midbrain, by in vivo MRI at 9.4T. Through image analysis in native space, several brain structures of the mid brain could be identified. The signal variation of all imaging modalities (T1, R2\textasteriskcentered and QSM) along and across the superior colliculus consistently highlighted the deep white layer VII, adjacent to the periaqueductal grey; the myelinated fibres in the superficial optic layer (layer III) and an iron-rich layer attributed to the intermediate grey layer (IV).}

{Off-resonant spin-lock imaging enables a lot of possibilities for T1$\varrho$ and chemical exchange (CE) sensitive applications. For this purpose, a matching amplitude of the tipping and the locking pulse is required, which can be difficult due to the high power requirements of adiabatic pulses. In this work, we present a newly shaped adiabatic half-passage pulse, usable at low power to match the amplitude of the pulses. Off- and on-resonant saturated images acquired at 9.4T are shown. The new pulse shape is able to generate robust images with comparatively low power at ultra-high-field strengths.}

{Aims and objectives: Alzheimer\textquoterights Disease (AD) is the most common cause of dementia worldwide. So far univocal diagnosis of AD is only achieved by postmortem histology. Beta-Amyloid plaques are known to be classical hallmarks of the post mortem Alzheimer\textasciiacutes Disease brain. In-vivo, Beta-Amyloid deposits are currently detec[...] Methods and materials: Two patients with autosomal dominant AD (female 51y, male 35y) and two age and sex-matched healthy subjects (HS) were scanned at 9.4T using a multi-echo (N\textequals5) 3D-GRE sequence (0.375x0.375x0.8mm3 voxel size, TR\textequals35ms; TE\textequals6 to 30ms in steps of 6ms, TA\textequals9min, FOV\textequals192x174x70.4mm3, matrix size\textequals512x464x88) [...] Results: A variation of 70ms-1ca. in the effective transverse relaxation rate was observed between grey and white matter in AD patient compared to healthy subject (Figure 1). A pattern between grey and white matter, corresponding to paramagnetic effects, was also detected in the susceptibility map (0-0.04 pp[...] Conclusion: Both R2\textasteriskcentered and QSM methods at ultra-high field hold promise for detecting Beta-Amyloid load within the cortical rim providing a potential means for early diagnosis of AD in-vivo. Optimization of the QSM algorithm would likely increase the power of Beta-Amyloid detection ex-vivo and in-vivo.}

{We explored anatomical details of the superior colliculus (SC) by in vivo magnetic resonance imaging (MRI) at 9.4T. The high signal-to-noise ratio allowed the acquisition of high resolution, multi-modal images with voxel sizes ranging between 176 $\times$ 132 $\times$ 600 $\mu$m and (800)3$\mu$m. Quantitative mapping of the longitudinal relaxation rate R1, the effective transverse relaxation rate R2\textasteriskcentered, and the magnetic susceptibility QSM was performed in 14 healthy volunteers. The images were analyzed in native space as well as after normalization to a common brain space (MNI). The coefficient-of-variation (CoV) across subjects was evaluated in prominent regions of the midbrain, reaching the best reproducibility (CoV of 5) in the R2\textasteriskcentered maps of the SC in MNI space, while the CoV in the QSM maps remained high regardless of brain-space. To investigate whether more complex neurobiological architectural features could be detected, depth profiles through the SC layers towards the red nucleus (RN) were evaluated at different levels of the SC along the rostro-caudal axis. This analysis revealed alterations of the quantitative MRI parameters concordant with previous post mortem histology studies of the cyto- and myeloarchitecture of the SC. In general, the R1 maps were hyperintense in areas characterized by the presence of abundant myelinated fibers, and likely enabled detection of the deep white layer VII of the SC adjacent to the periaqueductal gray. While R1 maps failed to reveal finer details, possibly due to the relatively coarse spatial sampling used for this modality, these could be recovered in R2\textasteriskcentered maps and in QSM. In the central part of the SC along its rostro-caudal axis, increased R2\textasteriskcentered values and decreased susceptibility values were observed 2 mm below the SC surface, likely reflecting the myelinated fibers in the superficial optic layer (layer III). Towards the deeper layers, a second increase in R2\textasteriskcentered was paralleled by a paramagnetic shift in QSM suggesting the presence of an iron-rich layer about 3 mm below the surface of the SC, attributed to the intermediate gray layer (IV) composed of multipolar neurons. These results dovetail observations in histological specimens and animal studies and demonstrate that high-resolution multi-modal MRI at 9.4T can reveal several microstructural features of the SC in vivo.}

{Our phenomenological experience of the stable world is maintained by continuous integration of visual self-motion with extra-retinal signals. However, due to conventional constraints of fMRI acquisition in humans, neural responses to visuo-vestibular integration have only been studied using artificial stimuli, in the absence of voluntary head-motion. We here circumvented these limitations and let participants to move their heads during scanning. The slow dynamics of the BOLD signal allowed us to acquire neural signal related to head motion after the observer\textquotesingles head was stabilized by inflatable aircushions. Visual stimuli were presented on head-fixed display goggles and updated in real time as a function of head-motion that was tracked using an external camera. Two conditions simulated forward translation of the participant. During physical head rotation, the congruent condition simulated a stable world, whereas the incongruent condition added arbitrary lateral motion. Importantly, both conditions were precisely matched in visual properties and head-rotation. By comparing congruent with incongruent conditions we found evidence consistent with the multi-modal integration of visual cues with head motion into a coherent \textquotedblleftstable world\textquotedblright percept in the parietal operculum and in an anterior part of parieto-insular cortex (aPIC). In the visual motion network, human regions MST, a dorsal part of VIP, the cingulate sulcus visual area (CSv) and a region in precuneus (Pc) showed differential responses to the same contrast. The results demonstrate for the first time neural multimodal interactions between precisely matched congruent versus incongruent visual and non-visual cues during physical head-movement in the human brain. The methodological approach opens the path to a new class of fMRI studies with unprecedented temporal and spatial control over visuo-vestibular stimulation.}

{Deep generative models such as Generative Ad- versarial Networks (GANs) and Variational Auto- Encoders (VAEs) are important tools to capture and investigate the properties of complex empiri- cal data. However, the complexity of their inner elements makes their functioning challenging to interpret and modify. In this respect, these archi- tectures behave as black box models. In order to better understand the function of such network, we analyze the modularity of these system by quantifying the disentanglement of their intrinsic parameters. This concept relates to a notion of invariance to transformations of internal variables of the generative model, recently introduced in the field of causality. Our experiments on generation of human faces with VAEs supports that modu- larity between weights distributed over layers of generator architecture is achieved to some degree, and can be used to understand better the function- ing of these architectures. Finally, we show that modularity can be enhanced during optimization.}

{Introduction \& Aim: Mentally demanding tasks are often, but not always, those that place a high load on working memory. In fact, the N-back task is often relied on as the "ground truth" for evaluating brain-computer interfaces that seek to infer user experienced workload [Brouwer2011]. Manual production lines is a real-world example that requires workers to memorize complex assembly instructions, an aspect that is becoming more demanding in recent times as production lot sizes decrease. To assist manual assembly, in-situ displays have been developed that provide just-in-time instructions for assembly [Funk2016]. The utility of this in alleviating mental workload has been validated with measures such as questionnaires or semi-structured interviews, but never with neuroimaging. In this work, we employ electroencephalography (EEG) to evaluate the extent to which in-situ displays alleviate visuospatial working memory during manual assembly. Methods: An Emotiv Epoc was used to record EEG data throughout the whole experiment. In a within-subject study design (N\textequals12), participants were instructed to use either paper or projected in-situ instructions to assemble a Lego Duplo construction [Funk2016]. Paper instructions were printed on A4 sized sheets of paper. In-situ instructions were projected directly into the workplace (see Figure 1). Before the start of the assembly task, participants performed a one-minute eyes-opened/eyes-closed task followed by two N-back tasks (N\textequals0 and N\textequals2) to ensure a correct setup of the Emotiv Epoc through alpha desynchronization and to determine the individual bandwidth for alpha power. Afterward, participants started to assemble two different Lego Duplo starting with either paper or projected in-situ instructions based on the balanced Latin square. Ground truth for perceived workload is measured via NASA-TLX questionnaires filled out by participants after every N-back and assembly trial. Results: Alpha power was derived and noise suppressed by using a spatio-spectral decomposition filter [Nikulin2011]. Our findings show significantly higher alpha desynchronization when conducting an N-back task with N\textequals2 than N\textequals0, F(1, 11) \textequals 34.82, p \textless 0.001. By comparing projected in-situ and paper instructions, projected in-situ instructions showed significantly less alpha desynchronization than paper instructions, F(1, 11) \textequals 14.92, p \textless 0.003. This converges with collected NASA-TLX questionnaires evaluating perceived workload levels. We also found that item selection errors have significantly increased when using paper instructions instead of projected in-situ instructions, F(1, 11) \textequals 10.75, p \textless 0.007. Discussion and Conclusion: Our study shows encouraging results of using a low-cost EEG to evaluate assembly instruction system. Possible reasons for paper instructions causing higher alpha desynchronization, and thus working memory load, might be that every step (e.g. picking and placing a brick) has to be remembered by participants while projected in-situ instructions obviate the need for remembering single steps. This is backed up by a reduced number of item selection errors and supported by NASA-TLX questionnaires. This provides insights for developers to evaluate the feasibility of their assistive system design on a cognitive basis.}

{Touch is interpersonal, as it is shared between beings who have some mutual relation to one another, whether an intimate long-term relationship or a superficial one, and can be used to convey thoughts and feelings between conspecifics. We asked, participants to evaluate their own and someone else\textquoterights preference of \textquotedblleftwhich body part they prefer to touch\textquotedblright, during observed human social touch. Participants were shown pictures of an avatar, face-on, which had an upper body to either side of it, comprised of the neck, arm and palm. Avatars expressed preferences to one of the two bodies (left or right) and one of the three body parts (neck, arm and palm), indicated through combining positive/negative facial expressions, positive/negative gestures, and head orientation. Participants were asked to infer their own (1st-person mentalising) and the avatar\textquoterights (3rd-person mentalising) preference. Furthermore, driven by the notion that humans are highly susceptible to the happenings in their social environment, we introduced a third condition where the avatar showed no preference towards either body/body part (neutral condition), and asked participants to again evaluate their own preference. This allowed us to examine how social touch interactions may influence our subjective preferences. Overall, participants were faster in the neutral condition, than in the 1st-person mentalising and 3rd-person mentalising conditions. Although, participants\textquoteright preference was centred on the palm during the neutral condition, this was extended to include the arm and neck during 1st-person mentalising condition, suggesting that the avatar\textquoterights interaction with a specific body part influenced participants\textquoteright preference. In addition, participants were highly accurate in inferring the avatar\textquoterights \textquoteleftpreference\textquoteright, and showed that they were inclined to respond faster to positive facial expressions and gestures than negative, implying that they may share the touch recipient\textquoterights experience.}

{Information integration across the senses is fundamental for effective interactions with our environment. The extent to which signals from different senses can interact in the absence of awareness is controversial. Combining the spatial ventriloquist illusion and dynamic continuous flash suppression (dCFS), we investigated in a series of two experiments whether visual signals that observers do not consciously perceive can influence spatial perception of sounds. Importantly, dCFS obliterated visual awareness only on a fraction of trials allowing us to compare spatial ventriloquism for physically identical flashes that were judged as visible or invisible. Our results show a stronger ventriloquist effect for visible than invisible flashes. Critically, a robust ventriloquist effect emerged also for invisible flashes even when participants were at chance when locating the flash. Collectively, our findings demonstrate that signals that we are not aware of in one sensory modality can alter spatial perception of signals in another sensory modality.}

{Previous research on own body size estimation has only looked at estimates made by comparing own body size to a test body in front view (e.g., M\"olbert et al. 2017). However, people constantly see and compare themselves to bodies in different viewpoints. Depending on the viewpoint, shape cues potentially used to judge body size, such as the waist-to-hip ratio or the overall body outline, vary. Here, we asked whether viewpoint influences estimates of own body size in female participants. For each participant, a personalized female avatar was generated using weight, height, inseam, and arm span, and then variations of the personalized avatar having different weights ($\pm$5, $\pm$10, $\pm$15, $\pm$20, and $\pm$25) were created using a statistical body model. These eleven test bodies were presented in life-size in immersive virtual reality in six viewpoints: 0\mbox{$^\circ$}, $\pm$45\mbox{$^\circ$}, $\pm$90\mbox{$^\circ$}, 180\mbox{$^\circ$}. In a one-alternative forced choice paradigm, participants were asked to judge whether the test body was thinner or fatter than themselves. Results showed no significant influence of viewpoint on either the accuracy of body size estimation (PSE) or the sensitivity to weight changes (JND). Across all viewpoints, participants on average slightly overestimated their body weight (3.1) and could detect a weight difference of 5.2 in 50 of the trials. To further investigate whether females are also able to estimate own body size when the shape of the test bodies is clearly different to theirs, a set of personalized male avatars was generated for each participant and presented in front view using the same task. There was no difference in results between female and male test bodies. These results suggest that people are rather good at extracting body size independent of the viewpoint, and also from bodies with a very different shape.}

{In this work, we combined scanner\textquotesingles spherical harmonic shim coils with a local multi-coil shim array to work in parallel for dynamic shimming of the human brain at 9.4 T. Performance of the combined method is compared with global shimming with scanner\textquotesingles built-in shim setup, global shimming with multi-coil and dynamic shimming with multi-coil.}