fMRI bold signal relies on the differences in magnetic susceptibility between oxygenated and deoxygenated haemoglobin. But this bears all the limitation of inferring neuronal activity just based on blood oxygenated haemoglobin.

Would it then be feasible to move deeper into the neuronal level, to measure discrete changes happening during the excitation of neurons? Of course haemoglobin would not work for that.

I could envision an isotope with a non-zero spin... divalent cation and hence act as a bioisostere Ca2+... The idea is that this specie would bind Ca-binding proteins within the neuron, would be displaced upon endoplasmatic reticulum release of Ca2+, and magnetic resonance would reflect changes between the bound-free forms as signature of excited or resting state of the neuron.

Else would PET could get beyond fluorinated-glucose consumption? PET techniques rely on the decay of the nucleids. Few variables can affect the decay of radioisotopes, but the electronic surrounding and the chemical composition can slightly affect the reate of radioactive. I do not know if to a measurable extent.

Would it be possible to come up with a radiolabelled ligand whose electronic surroudings are affected by neuron excitabiilty --relying on the cell own machinery-- to get the decay rate affected to a measurable extent? I could think of a radiolabelled respiratory chain inhibitor/cofactors, or hopefully something more neuron-specific. If so, changes in the decay rate over time could indicate changes in neuronal activity relative to a basal.

Just looking for some food for thought I'd appreciate all kinds of feedbacks with regard on technical limitations, feasibility...