This blog post summarizes the FENS webinar entitled: Multiscale, multimethod human brain imaging. Organised by the Human Brain Mapping.
Organized by Lars Muckli and Lucy Petro.
- One neuron’s output is another neuron’s input. fMRI seems to be sensitive to the energy demand of the presynaptic output at the location of the receiving neurons dendrites.
- The fMRI layer-profile matches better with MUA (fast fluctuations) than FLPs (slow fluctuations).
- Is fMRI reliable and replicable? All speakers seem to agree that recent populist critique of fMRI can be a wrong message to the field. Layer fMRI results are being replicated across modalities and species to a great extent.
- Where does the name ‘Tiramisu’ come from? The name comes from the network architecture (sets of dense layers).
- Can one combine EEG-fMRI measurements with intersubject correlation analyses? With EEG it has been done already for neuromarketing taks and movie watching.
- How to deal with subcortical structures in EEG-fMRI? Conventional EEG is not really sensitive to deep brain regions. New methods will allow more informed investigations.
- Do you (Rainer) think it is possible to extend Mario Senden’s work to reconstruct natural images? Letter shapes are already pushing it. It is possible to do it in real time. However, it might be challenging to extend this to complex natural scenes.
- Can the Tiramisu be used for T1 weighted 3D-EPI? The network is specific to specific sequence protocols. Though, it can be easily trained to any other sequence protocol in as few as 5 training datasets.
- How fast and how locally specific is the neurovascular coupling? is line scanning already exceeding the specificity of the neurovascular coupling? There are very fast responses of the vasculature. Faster than originally thought. But there is still progress to be expected about these questions in human and animal research.
- To what extent could the knowledge presented here be generalized from the early visual cortex to other brain areas? Primary sensory areas are probably generalizable. There are already several layer-fMRI studies looking at frontal and prefrontal areas. Confirmations from electrophys are nice to have but are not always available, though. The architecture of the cortical column are rather consistent across areas. The concepts of “feedforward” and “feedback” might be generalizable across areas. The question is what these concepts mean in the association cortex.
- Do fMRI and electrophysiology always agree with each other? Are there differences? There are draining vein problems in GE-BOLD. fMRI is more sensitive to modulations (feedback) than electrophysiology.