Biology subjects

Goebel, R.

Publications and source records attributed to Goebel, R..

At least 19 recordsLinked to original sources

bioRxiv2026

A critical assessment of LGN segmentation quality based on volume, contour, and location

Reliable segmentation of the lateral geniculate nucleus (LGN), a small thalamic relay nucleus essential for visual processing, is necessary for characterizing its anatomy in healthy populations and structural alterations in neurological disorders. However, accurate delineation remains challenging on routinely acquired T1-weighted images because the LGN is small, deeply located, and has limited contrast relative to surrounding thalamic tissue. Although manual segmentation remains the reference standard, it is time consuming and susceptible to rater variability, limiting its use in large datasets. Automated approaches may therefore provide a more scalable alternative. In this study, we performed a systematic benchmarking of seven automated approaches: FreeSurfer, NextBrain, DeepThalamus, HIPS-THOMAS, MASSP 2.0, BigBrain alignment, and manual novice segmentation against expert-defined ground truth. We applied a multi-metric evaluation incorporating the Dice similarity coefficient, Hausdorff distance and balanced average Hausdorff distance, volumetric similarity, and Euclidean distance. Together, these measures evaluate overlap, contour, volume, and location against expert manual segmentations. Our results indicate that HIPS-THOMAS and DeepThalamus outperformed the other automated methods in all metrics. However, manual segmentation retained high precision despite exhibiting greater inter-subject variance than the top-performing automated segmentation tools. The different metrics do not necessarily report the same order of quality, highlighting the need of using multiple metrics for evaluation segmentation quality.

neuroscience↗

bioRxiv2026

Searchlight Optimization Using Representational Similarity Analysis for Subject-Level Voxel Selection in Emotional State Decoding

Identifying informative voxels is a critical, yet challenging step in functional magnetic resonance imaging (fMRI), particularly for multivariate analyses involving multiple related conditions. Existing approaches often rely on predefined regions of interest (ROIs) or activation-based criteria, which may be insufficient for capturing fine-grained representational differences. This challenge becomes particularly relevant in experimental settings and interventions such as neurofeedback training, where voxels are not only measured as neural responses but also used as targets for intervention based on their previously observed activity patterns. In this study, we propose a subject-level searchlight optimization framework that integrates voxel-wise general linear model (GLM)-based univariate analysis with representational similarity analysis (RSA)-based multivariate refinement to identify voxels that are both task-relevant and condition-sensitive. To enhance practical applicability, the framework further incorporates a data-driven hyperparameter tuning step based on Bayesian optimization, enabling efficient identification of high-performing configurations from small pilot datasets, with consistent performance when applied to larger samples. The proposed framework was evaluated using an emotion imagery fMRI dataset with four affective conditions. Results demonstrate that the multivariate refinement improves alignment between empirical and target representational structures compared with univariate selection alone. Compared with a classifier-based voxel selection approach, the RSA-based approach better preserves the representational geometry of emotional states while maintaining discriminative capacity. These findings highlight the effectiveness, efficiency, and robustness of the proposed RSA framework, providing a practical solution for identifying condition-sensitive voxels and supporting more precise multivariate investigation of affective brain states in multi-condition fMRI studies.

neuroscience↗

bioRxiv2026

All signals considered: Data quality partially explains inter-individual task differences in a large, open fNIRS dataset

Significance: High inter-subject variability and limited reproducibility in functional near-infrared spectroscopy (fNIRS) research may partly reflect global systemic physiology and signal quality differences, possibly distorting task-evoked hemodynamic responses. Aim: We investigate how signal quality relates to inter-subject variability in motor-task fNIRS responses and introduce a large, open, multi-task, near whole-head fNIRS dataset with extensive peripheral physiology and short-channel recordings. Approach: Fifty-seven participants completed resting-state, motor action, motor imagery, emotion recognition, visual, and auditory tasks during fNIRS recording. Peripheral measures included pulse oximetry, heart rate, blood oxygen saturation, respiration, room temperature, galvanic skin response, electrocardiogram, and electromyography. Signal quality was assessed using the scalp coupling index (SCI), coefficient of variation (CV), signal-to-noise ratio (SNR) and a spectral measure here coined the coupling SNR (cSNR). Results: Quality metrics were weakly to moderately correlated, except SNR and CV, which showed the expected inverse relationship. All quality metrics were significantly related to channel length and associated with task-related activation estimates. Group-level analyses validated activation in expected task-related regions. Conclusions: The assessed metrics capture complementary features of fNIRS signal quality and may help explain individual activation differences. The dataset provides a comprehensive, open resource enabling future evaluation of physiological correction methods and confound mitigation.

neuroscience↗

bioRxiv2026

Intracortical vessel density mapping at mesoscale reveals regionally specific angioarchitecture in the human brain

The metabolic demands of the human brain are met by a complex vascular architecture, yet our characterization of this network remains incomplete. While we have mapped the macroscopic vessels on the brains surface and the microscopic capillaries within small tissue samples, the mesoscopic scale consisting of the penetrating vessels that plunge through cortex remains an anatomical terra incognita. Mapping the interface between the macroscopic and microscopic scales is essential to understanding the critical vascular supply that sustains brain health. Here, we leveraged the BigBrain dataset and developed custom detection and tracing algorithms to reveal a whole-cortex record of the mesoscopic vascular network. We find that vascular density is not uniform across the cortex, but is a heterogeneous landscape that shows clear relationships to traditional areal boundaries. Our results constitute a reference for human mesoscopic angioarchitecture and demonstrates the power of repurposing high-resolution histological at-lases. Ultimately, this work lays the groundwork for validating recently developed in vivo MRI techniques for imaging the human cerebrovascular system at mesoscale.

neuroscience↗

bioRxiv2025

Toward navigating emotional states using real-time representational similarity analysis fMRI neurofeedback - a feasibility study

Real-time functional magnetic resonance imaging neurofeedback (rt-fMRI-NF) is a promising non-invasive brain-computer-interface (BCI) technique for enhancing self-regulation of affective states in the brain. However, conventional univariate rt-fMRI-NF approaches are limited in their ability to distinguish neural patterns of distinct emotions that involve overlapping brain regions. In this study, we applied an rt-fMRI semantic neurofeedback (rt-fMRI-sNF) paradigm, incorporating real-time representational similarity analysis (rt-RSA) to enable navigation between emotional states. Four emotional patterns were first derived from functional localizer runs, each designed to evoke a specific emotion, and then applied as target patterns during neurofeedback. Using an RSA-informed circular semantic map (CSM), participants received real-time visual feedback indicating both the similarity and intensity of their current brain activity relative to target emotional patterns. Participants were instructed to use mental imagery to shift their brain activity toward the specific target pattern and increase its intensity. Twenty-four healthy participants completed the localizer runs, and two consecutive neurofeedback runs in the same session. Ten participants successfully engaged with both the similarity and intensity components of the CSM, showing effective modulations of their mental states. Analyses of the localizer runs revealed overlapping regional activations across emotions and demonstrated that RSA outperformed univariate analysis in distinguishing between them. For the neurofeedback runs, linear mixed-effects model (LMM) analyses across multiple performance metrics indicated consistent within-run improvements and higher initial performance in the second run, while significant between-run learning effects emerged only in exploratory models with quadratic time terms. A block-wise comparison also showed significantly higher performance at the end of each run compared to the beginning based on the intensity metric. These findings support the usability of RSA in differentiating multiple emotional states and demonstrate the feasibility of the rt-fMRI-sNF paradigm for emotion regulation.

neuroscience↗

bioRxiv2024

Navigating the Maze: Identifying Potential Pitfalls in Attention State Classification from fMRI Brain Patterns

Multi-voxel pattern analysis (MVPA) is a powerful technique to decode brain states from functional magnetic resonance imaging (fMRI) activity patterns. In neurofeedback (NF) applications, it has been used to perform real-time classification of brain activity patterns, establishing a closed-loop system that provides immediate feedback to the participants, enabling them to learn to control a complex mental state. However, MVPA has many potential limitations when applied to fMRI datasets (especially in real-time analysis) arising from small effect sizes, small number of training samples, high dimensionality of the data and, more generally, design choices. All these factors might produce inaccurate classification results. In this work, we followed a previous NF paradigm for sustained attention training. Participants were presented with composite images superimposing faces and scenes. They were instructed to focus on one class (either face or scene) for an extended period. A logistic regression classifier was trained to determine whether participants were adequately focusing on the instructed category based on their fMRI data. We analysed the classification outputs of the no-feedback training runs using various classifier settings, including whole brain data and different masking approaches, combined with different methods for the computation of single-trial fMRI responses. Furthermore, a ventricle mask was used as a control condition for the classification task, and simulations were carried out to assess the influence of the class order on the classification performances. We found inflation of the decoding accuracy for several common design choices and confounders. In particular, motion artefacts and low frequency drifts coupled with the task timing might have artificially increased the accuracy scores. Furthermore, the simulations revealed that fixed order in the presentation of experimental conditions resulted in further inflation of the classification accuracies especially in GLM-based and average-based trial estimate methods. We discuss the drawbacks of applying MVPA using the analysed sustained attention paradigm and provide insights for future improvements.

neuroscience↗

bioRxiv2024

Visual and auditory object representations in ventral visual cortex after restoring sight in humans

Visual category-selective representations in human ventral occipital temporal cortex (VOTC) seem to emerge early in infancy. Surprisingly, the VOTC of congenitally blind humans features category-selectivity for auditory and haptic objects. Yet it has been unknown whether VOTC would show category-selective visual responses if sight were restored in congenitally blind humans. Assuming competition for synaptic space during development, cross-modal activation of VOTC as a consequence of congenital blindness might interfere with visual processing in sight-recovery individuals. To test this hypothesis, we investigated adults who had suffered a transient phase of congenital blindness due to bilateral dense cataracts before their sight was restored by cataract-removal surgery. In a functional magnetic resonance imaging (fMRI) study, participants watched movies of faces, scenes, body parts and other objects in the visual condition, while in the auditory condition they listened to the corresponding sounds. The most prominent group difference was the reduced face-selectivity in individuals with reversed congenital cataracts compared to age- and sex-matched normally-sighted individuals. In addition, a double dissociation was found: only sight-recovery individuals demonstrated significant decoding accuracy of visual categories based on auditory category representations in VOTC, while only normally-sighted individuals VOTC decoded auditory categories based on visual category representations. The present results uncovered the neural mechanisms of previously observed face processing impairments in individuals with reversed congenital blindness. We suggest that lower face-selectivity in the sight recovery group might arise from selective deficits in the cortical representation of the central visual field in lower-tier visual areas. Additionally, we speculate that in higher-order visual areas cross-modal activity might facilitate - rather than interfere - with visual functional recovery after congenital blindness.

neuroscience↗

bioRxiv2024

In the brain of the beholder: bi-stable motion reveals mesoscopic-scale feedback modulation in V1

Understanding the neural processes underlying conscious perception remains a central goal in neuroscience. Visual illusions, whether static or dynamic, provide an effective ecological paradigm for studying conscious perception, as they induce subjective experiences from constant visual inputs. While previous neuroimaging studies have dissociated perceptual interpretation of visual motion from sensory input within the motion-sensitive area (hMT+) in humans, less is known about the role of V1 and its relationship to hMT+ during a bistable perception. To address this, we conducted a layer-fMRI study at 7 T with human participants exposed to a bistable motion quartet stimulus. Despite a constant sensory input, the bistable motion quartet elicits switching horizontal and vertical apparent motion percepts likely due to lateral and feedback connections across low and high-level brain regions (feedback processing). As control, we used an "unambiguous" version of the motion quartet, hereafter referred to as "physical" motion stimulus, where horizontal and vertical motion is physically presented as visual stimulus in an alternated fashion (feedforward processing). With the advantage of a sub-millimeter resolution gained at ultra-high field (7 Tesla), we aimed to unveil the differential laminar modulation of V1 (low visual area) and hMT+ (high-visual area) during the physical and bistable condition. Our results indicate that: O_LIhMT+ functional activity correlates with conscious perception during both physical and ambiguous stimuli with similar strength. There is no evidence of differential laminar profiles in hMT+ between the two experimental conditions. C_LIO_LIBetween inducer squares, V1 shows a significantly reduced functional response to the ambiguous stimulus compared to the physical stimulus, as it primarily reflects feedback signals with diminished feedforward input. Distinct V1 laminar profiles differentiate the two experimental conditions. C_LIO_LIThe temporal dynamics of V1 and hMT+ become more similar during the ambiguous condition. C_LIO_LIV1 exhibits reduced specificity to horizontal and vertical motion perception during the ambiguous condition at the retinotopic locations corresponding to the perceived motion. C_LI Our findings demonstrate that during the ambiguous condition, there is a stronger temporal coupling between hMT+ and V1 due to feedback signals from hMT+ to V1. Such feedback to V1 might be contributing to the stabilization of the vivid perception of directed motion at the face of constant ambiguous stimulation.

neuroscience↗

bioRxiv2024

PrediRep: Modelling hierarchical predictive coding with an unsupervised deep learning network

Hierarchical predictive coding (hPC) proposes that the cortex continuously generates predictions of incoming sensory stimuli. Deep neural networks inspired by hPC are frequently used to probe the neurocomputational mechanisms suggested by the theory in silico and to generate hypotheses for experimental investigations. However, these networks often deviate from hPC by prioritizing computational efficiency over alignment with its principles. To remedy this, we introduce PrediRep, a deep learning model explicitly designed to emphasize alignment with the theory. PrediRep incorporates the principles of hPC found in the other networks, while avoiding their deviations from it. We evaluate the performance of PrediRep on a next-frame prediction task and its functional alignment with hPC, comparing it to other contemporary deep learning networks inspired by the theory. Our findings demonstrate that PrediRep achieves the closest functional alignment with hierarchical predictive coding without sacrificing computational performance.

neuroscience↗

bioRxiv2024

Optimal placement of high-channel visual prostheses in human retinotopic visual cortex

AO_SCPLOWBSTRACTC_SCPLOWO_ST_ABSObjectiveC_ST_ABSRecent strides in neurotechnology show potential to restore vision in individuals afflicted with blindness due to early visual pathway damage. As neuroprostheses mature and become available to a larger population, manual placement and evaluation of electrode designs becomes costly and impractical. An automatic method to optimize the implantation process of electrode arrays at large-scale is currently lacking. ApproachHere, we present a comprehensive method to automatically optimize electrode placement for visual prostheses, with the objective of matching pre-defined phosphene distributions. Our approach makes use of retinotopic predictions combined with individual anatomy data to minimize discrepancies between simulated and target phosphene patterns. While demonstrated with a 1000-channel 3D electrode array in V1, our pipeline is versatile, potentially accommodating any electrode design and allowing for design evaluation. Main resultsNotably, our results show that individually optimized placements in 362 brain hemi-spheres outperform average brain solutions, underscoring the significance of anatomical specificity. We further show how virtual implantation of multiple individual brains highlights the challenges of achieving full visual field coverage owing to single electrode constraints, which may be overcome by introducing multiple arrays of electrodes. Including additional surgical considerations, such as intracranial vasculature, in future iterations could refine the optimization process. SignificanceOur open-source software streamlines the refinement of surgical procedures and facilitates simulation studies, offering a realistic exploration of electrode configuration possibilities.

neuroscience↗

bioRxiv2023

Neural Correlates of High-Level Visual Saliency Models

Visual saliency highlights regions in a scene that are most relevant to an observer. The process by which a saliency map is formed has been a crucial subject of investigation in both machine vision and neuroscience. Deep learning-based approaches incorporate high-level information and have achieved accurate predictions of eye movement patterns, the overt behavioral analogue of a saliency map. As such, they may constitute a suitable surrogate of cortical saliency computations. In this study, we leveraged recent advances in computational saliency modeling and the Natural Scenes Dataset (NSD) to examine the relationship between model-based representations and the brain. Our aim was to uncover the neural correlates of high-level saliency and compare them with low-level saliency as well as emergent features from neural networks trained on different tasks. The results identified hV4 as a key region for saliency computations, informed by semantic processing in ventral visual areas. During natural scene viewing, hV4 appears to serve a transformative role linking low- and high-level features to attentional selection. Moreover, we observed spatial biases in ventral and parietal areas for saliency-based receptive fields, shedding light on the interplay between attention and oculomotor behavior.

neuroscience↗

bioRxiv2023

Unlocking the Secrets of the Primate Visual Cortex: A CNN-Based Approach Traces the Origins of Major Organizational Principles to Retinal Sampling

Primate visual cortex exhibits key organizational principles: Cortical magnification, eccentricity-dependent receptive field size and spatial frequency tuning as well as radial bias. We provide compelling evidence that these principles arise from the interplay of the non-uniform distribution of retinal ganglion cells (RGCs), and a quasi-uniform convergence rate from the retina to the cortex. We show that convolutional neural networks (CNNs) outfitted with a retinal sampling layer, which resamples images according to retinal ganglion cell density, develop these organizational principles. Surprisingly, our results indicate that radial bias is spatial-frequency dependent and only manifests for high spatial frequencies. For low spatial frequencies, the bias shifts towards orthogonal orientations. These findings introduce a novel hypothesis about the origin of radial bias. Quasi-uniform convergence limits the range of spatial frequencies (in retinal space) that can be resolved, while retinal sampling determines the spatial frequency content throughout the retina.

neuroscience↗

bioRxiv2022

Tactile Stimulation Designs Adapted to Clinical Settings Result in Reliable fMRI-based Somatosensory Digit Maps

A wide range of neurological diseases with impaired motor functioning of the upper extremities are accompanied by impairments of somatosensory functioning, which are often undescribed but can provide crucial information for diagnostics, treatment selection, and follow-up. Therefore, a reliable description of the functional representation of the digits in the somatosensory cortex would be a highly valuable, but currently lacking, tool in the clinical context. Task-based functional Magnetic Resonance Imaging of passive tactile stimulation provides an indirect, but valid description of the layout of the digit map in the primary somatosensory cortex. However, to fulfill the specific requirements for clinical application, the presently established approaches need to be adapted and subsequently assessed for feasibility and retest reliability, in order to provide informative parameters for the description of the evoked digit activations. Accordingly, the present high-field 3T fMRI study compares the performance of two established digit mapping designs - travelling wave (TW) and blocked design (BD) - for passive tactile stimulation of the five digits, adapted to reduce the time requirements to just below 15 minutes. To be able to assess the retest reliability unaffected by any clinical conditions, the study was performed on neurotypical participants. The results show that both stimulation designs evoke significant and distinct activation clusters in the primary somatosensory cortex of all participants for all five digits. The average spatial locations of the center of gravities across participants show the common succession of distinct digit representation along the central sulcus. The cortical extent elicited activation, which is generally larger for the thumb and the index finger, also shows comparable average values across the two approaches. Less overlap of activation between neighboring digits was obtained in BD, consistent with the distinct single digit neuronal representations. A high retest reliability was obtained for the location of the digit activation, displaying stable center of gravity locations across sessions for both stimulation designs. This is contrasted by only medium to low retest reliability for the extent and overlap of the digit activations, indicating discrepancies across sessions. These results demonstrate the capacity of shortened fMRI digit mapping approaches (both TW and BD) to obtain the full layout of single digit cortical activations on the level of the individual, which together with the high reliability of the location of the digit representation over time indicates both approaches are clinically applicable.

neuroscience↗

bioRxiv2022

Acquisition and processing methods of whole-brain layer-fMRI VASO and BOLD: The Kenshu dataset.

Cortical depth-dependent functional magnetic resonance image (fMRI), also known as layer-fMRI, has the potential to capture directional neural information flow of brain computations within and across large-scale cortical brain networks. E.g., layer-fMRI can differentiate feedforward and feedback cortical input in hierarchically organized brain networks. Recent advancements in 3D-EPI sampling approaches and MR contrast generation strategies have allowed proof-of-principle studies showing that layer-fMRI can provide sufficient data quality for capturing laminar changes in functional connectivity. These studies have however not shown how reliable the signal is and how repeatable the respective results are. It is especially unclear whether whole-brain layer-fMRI functional connectivity protocols are widely applicable across common neuroscience-driven analysis approaches. Moreover, there are no established preprocessing fMRI methods that are optimized to work for whole-brain layer-fMRI datasets. In this work, we aimed to serve the field of layer-fMRI and build tools for future routine whole-brain layer-fMRI in application-based neuroscience research. We have developed publicly available sequences, acquisition protocols, and processing pipelines for whole-brain layer-fMRI. These protocols are validated across 60 hours of scanning in nine participants. Specifically, we identified and exploited methodological advancements for maximizing tSNR efficiency and test-retest reliability. We are sharing an extensive multi-modal whole-brain layer-fMRI dataset (20 scan hours of movie-watching in a single participant) for the purpose of benchmarking future method developments: The Kenshu dataset. With this dataset, we are also exemplifying the usefulness of whole brain layer-fMRI for commonly applied analysis approaches in modern cognitive neuroscience fMRI studies. This includes connectivity analyses, representational similarity matrix estimations, general linear model analyses, principal component analysis clustering, etc. We believe that this work paves the road for future routine measurements of directional functional connectivity across the entire brain. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=132 SRC="FIGDIR/small/504502v1_ufig1.gif" ALT="Figure 1"> View larger version (89K): org.highwire.dtl.DTLVardef@1c6579forg.highwire.dtl.DTLVardef@a2bbecorg.highwire.dtl.DTLVardef@1dfa41aorg.highwire.dtl.DTLVardef@18ee67f_HPS_FORMAT_FIGEXP M_FIG C_FIG

neuroscience↗

bioRxiv2022

Imaging the columnar functional organization of human area MT+ to axis-of-motion stimuli using VASO at 7 Tesla

Cortical columns of direction-selective neurons in the motion sensitive area (MT) have been successfully established as a microscopic feature of the neocortex in animals. The same property has been investigated at mesoscale (<1 mm) in the homologous brain area (hMT+, V5) in living humans by using ultra-high field functional magnetic resonance imaging (fMRI). Despite the reproducibility of the selective response to axis-of-motion stimuli, clear quantitative evidence for the columnar organization of hMT+ is still lacking. Using cerebral blood volume (CBV)-sensitive fMRI at 7 Tesla with submillimeter resolution and high spatial specificity to microvasculature, we investigate the columnar functional organization of hMT+ in 5 participants perceiving axis-of-motion stimuli for both blood oxygenation level dependent (BOLD) and vascular space occupancy (VASO) contrast mechanisms provided by the used Slab-Selective Slice Inversion (SS-SI)-VASO sequence. With the development of a new searchlight algorithm for column detection, we provide the first quantitative columnarity map that characterizes the entire 3D hMT+ volume. Using voxel-wise measures of sensitivity and specificity, we demonstrate the advantage of using CBV-sensitive fMRI to detect mesoscopic cortical features by revealing higher specificity of axis-of-motion cortical columns for VASO as compared to BOLD contrast. These voxel-wise metrics also provide further insights on how to mitigate the highly debated draining veins effect. We conclude that using CBV-VASO fMRI together with voxel-wise measurements of sensitivity, specificity and columnarity offers a promising avenue to quantify the mesoscopic organization of hMT+ with respect to axis-of-motion stimuli. Furthermore, our approach and methodological developments are generalizable and applicable to other human brain areas where similar mesoscopic research questions are addressed.

neuroscience↗

bioRxiv2022

Retrieving fMRI data in real-time: difficulties and pitfalls

One of the significant challenges in real-time fMRI environments is to ensure that the functional images are exported in real-time. The prerequired ability to reconstruct these images immediately after the acquisition has already been resolved in 2004. Nowadays, more sophisticated sequences allow for higher resolution and faster repetition times and thereby challenging the ability to export this data in real-time. In this article, we tackle the potentially arising problem of sending the reconstructed data from the MRI to an external PC to perform the real-time fMRI analysis. We show that depending on the implementation of the data transfer, long delays can occur that can differ drastically in time and how often they occur. In addition, we propose a solution for SIEMENS MRI devices which was tested and applied already on multiple MRI devices including 3T and 7T machines on different vendor software versions. This new technique can be used as a blueprint that can be directly applied to other manufacturers. We also provide the source code of the described solution and show that the delay in the data transfer can be significantly reduced to a tolerable level using our proposed procedure. Finally, we integrate measurement options for the data transfer times to improve quality measures in real-time fMRI environments (e.g., clinical) that can implement the proposed solution. Efforts should be taken by the real-time community and MRI manufacturers to employ a standardized real-time export e.g., similar to the lab streaming layer which is used as a standard export method in EEG environments.

neuroscience↗

bioRxiv2021

Mesoscopic Quantification of Cortical Architecture in the Living Human Brain

AO_SCPLOWBSTRACTC_SCPLOWMesoscopic (0.1-0.5 mm) interrogation of the living human brain is critical for advancing neuroscience and bridging the resolution gap with animal models. Despite the variety of MRI contrasts measured in recent years at the mesoscopic scale, in vivo quantitative imaging of T2* has not been performed. Here we provide a dataset containing empirical T2* measurements acquired at 0.35 x 0.35 x 0.35 mm3 voxel resolution using 7 Tesla MRI. To demonstrate unique features and high quality of this dataset, we generate flat map visualizations that reveal fine-scale cortical substructures such as layers and vessels, and we report quantitative depth-dependent T2* (as well as R2*) values in primary visual cortex and auditory cortex that are highly consistent across subjects. This dataset is freely available at https://doi.org/10.17605/OSF.IO/N5BJ7, and may prove useful for anatomical investigations of the human brain, as well as for improving our understanding of the basis of the T2* -weighted (f)MRI signal.

neuroscience↗

bioRxiv2021

ConvNets Develop Organizational Principles of the Visual Cortex when using Ganglion Cell-Based Sampling

AO_SCPLOWBSTRACTC_SCPLOWThe distribution of retinal ganglion cells in primate visual systems portrays a densely distributed central region, with an incrementally decreasing cell density as the angle of visual eccentricity increases. This results in a non-uniform sampling of the retinal image that resembles a wheelbarrow distortion. We propose that this sampling gives rise to several organizational properties of the primate visual system, including cortical magnification, linear relationship between eccentricity and receptive field sizes, eccentricity-dependent drop-off in spatial-frequency preference, and radial bias. We test this hypothesis by training a convolutional neural network to classify the orientation of sine gratings and Gabor stimuli, resampled according to retinal ganglion cell distributions. Our simulations show that introducing this sampling step gives rise to the aforementioned organizational principles in convolutional layers while only minimally affecting their classification performance. This lends credence to the notion that the retinal ganglion cell distribution is an important factor for the emergence of these organizational principles in visual systems.

neuroscience↗