Simulating the Impact of Vascular Draining Effects on Layer-Specific MVPA using GE-BOLD
Poster No:
1601
Submission Type:
Abstract Submission
Authors:
Jonas Karolis Degutis1, Romy Lorenz1, Denis Chaimow1
Institutions:
1Max Planck Institute for Biological Cybernetics, Tübingen, Germany
First Author:
Co-Author(s):
Introduction:
Cortical layers are fundamental to neocortical organization. Ultra-high-field fMRI enables investigations of layer-specific functioning in humans by measuring differences between laminar activations in various experiments [1]. Most studies use GE-BOLD imaging, despite its reduced spatial specificity due to signal leakage from deep to superficial layers [2]. As a result, superficial layers show a larger, mixed signal, compromising laminar findings in univariate analyses. Interestingly, multivariate approaches appear to mitigate these biases, with studies reporting higher decoding accuracy in deep and middle layers, suggesting that these methods better capture layer-specific neural origins [3,4]. However, the neurophysiological basis for this effect remains unclear. Here, we perform exhaustive simulation analyses to examine the impact of vascular draining effects on multivariate analyses of layer-specific GE-BOLD responses.
Methods:
We developed a mechanistic model to simulate multivariate patterns across layers, building on [5]. The model involved four stages: generating patterns, modeling neuronal responses, simulating spatial BOLD activation, and reconstructing MR voxel responses. Patterns were created with a given spatial frequency in each depth. They were convolved with depth-varying BOLD point-spread functions followed by a vascular model that drained the signal from deep to superficial layers [6]. The BOLD signal was then MR sampled and layer-dependent noise added (Fig. 1a).
The primary analysis simulated two distinguishable columnar patterns originating from one of the three layers: deep, middle, or superficial. A linear classifier was used to decode these patterns from each layer. We then varied the spatial scale of patterns and analyzed their effect on layer-specific decoding. Since the main simulations assumed unrealistic perfect laminar delineation, we tested the effect of layer mis-segmentation on MVPA results. We also inverted the vascular model with the aim of deconvolving the signal and investigated how incorrect deconvolution parameters impact decoding accuracies between layers.
The primary analysis simulated two distinguishable columnar patterns originating from one of the three layers: deep, middle, or superficial. A linear classifier was used to decode these patterns from each layer. We then varied the spatial scale of patterns and analyzed their effect on layer-specific decoding. Since the main simulations assumed unrealistic perfect laminar delineation, we tested the effect of layer mis-segmentation on MVPA results. We also inverted the vascular model with the aim of deconvolving the signal and investigated how incorrect deconvolution parameters impact decoding accuracies between layers.
Results:
1. We find that the layer of origin showed maximal decoding relative to other layers (Fig. 1b). However, vascular draining resulted in considerable spread of decodable information making it difficult to disentangle deep vs. middle and middle vs. superficial layers when results originate in deep or middle layers, respectively. Importantly, deconvolving the signal prior to decoding improved laminar specificity across all simulations (Fig. 1c).
2. The spatial scale of the columnar pattern impacts layer-specificity of decoding results. For finer columnar patterns, we see layer-specific decoding differences despite patterns being present in all layers (Fig. 2a). This effect is further enhanced by deconvolution (Fig. 2b).
3. Layer mis-segmentation results in relatively stable decoding differences irrespective of the percent of misaligned voxels in deep and superficial layers. However, for middle layers, we find the layer-specificity of the results being compromised for moderate voxel misalignment (Fig. 2c).
4. Vascular model parameter differences between convolution and deconvolution show relatively stable decoding differences irrespective of misalignment percentage (Fig. 2d).
2. The spatial scale of the columnar pattern impacts layer-specificity of decoding results. For finer columnar patterns, we see layer-specific decoding differences despite patterns being present in all layers (Fig. 2a). This effect is further enhanced by deconvolution (Fig. 2b).
3. Layer mis-segmentation results in relatively stable decoding differences irrespective of the percent of misaligned voxels in deep and superficial layers. However, for middle layers, we find the layer-specificity of the results being compromised for moderate voxel misalignment (Fig. 2c).
4. Vascular model parameter differences between convolution and deconvolution show relatively stable decoding differences irrespective of misalignment percentage (Fig. 2d).
Conclusions:
Contrary to expectations, vascular draining affects layer-specific decoding results to an extent. We illustrate how signal deconvolution can restore layer-specificity only when the signal originates in a single layer and show that decoding differences remain relatively immune to incorrect deconvolution parameters. However, we also find that signals originating from all layers and mis-segmentation can have detrimental effects on layer-specific decoding. Our findings underscore the necessary caution needed when interpreting laminar MVPA results both with and without deconvolution.
Modeling and Analysis Methods:
Exploratory Modeling and Artifact Removal 2
Multivariate Approaches 1
Keywords:
Cortical Layers
HIGH FIELD MR
Modeling
Multivariate
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
Provide references using APA citation style.
2. Koopmans, P. J., Barth, M., & Norris, D. G. (2010). Layer‐specific BOLD activation in human V1. Human brain mapping, 31(9), 1297-1304.
Iamshchinina, P., Kaiser, D., Yakupov, R., Haenelt, D., Sciarra, A., Mattern, H., ... & Cichy, R. M. (2021). Perceived and mentally rotated contents are differentially represented in cortical depth of V1. Communications biology, 4(1), 1069.
3. Bergmann, J., Petro, L. S., Abbatecola, C., Li, M. S., Morgan, A. T., & Muckli, L. (2024). Cortical depth profiles in primary visual cortex for illusory and imaginary experiences. Nature Communications, 15(1), 1002.
4. Markuerkiaga, I., Marques, J. P., Gallagher, T. E., & Norris, D. G. (2021). Estimation of laminar BOLD activation profiles using deconvolution with a physiological point spread function. Journal of neuroscience methods, 353, 109095.
5. Chaimow, D., Uğurbil, K., & Shmuel, A. (2018). Optimization of functional MRI for detection, decoding and high-resolution imaging of the response patterns of cortical columns. NeuroImage, 164, 67-99.