El Duvelle NeuroIf you want to study #HippocampalReplay... Use ephys, not #CalciumImaging!!
(Calcium imaging doesn't detect single spikes well, but replay mostly involves single spikes)
#Neuroscience #SpatialCognition #Hippocampus
Oh and also.. Replay happens during #SWRs or at least "large, irregular activity" state. If you cannot access the #LFP as is the case with Calcium Imaging.. then you can't know if what you're recording is replay or #ThetaSequences!
Replay is also very fast.. A whole trajectory could be replayed in less than 100 ms.. Probably too fast for calcium imaging (although I'd be happy to see some quantification of the Ca timeline?)
Also if you want to study replay during #Planning.. maybe don't do it while the animals are eating? Maybe (just maybe) if the animals are eating they'll be thinking how did they get the food, and not about what to do next?
Ko-Fan Chen 陳克帆@elduvelle_neuro I was wondering last week, why those voltage sensors (like arclight sorry I am not behind with tech ) don't seem to offer the promise, or am I wrong? I do feel with #connnectome catching up we needs a way to access functional connectivity with temp resolution of ephy but better spatially? #neuroscience #LFP
@kofanchen I also don't know enough about the imaging techniques or advancements in those to say. But there are quite good techniques that combine optogenetic tracing and ephys, for example like in this study: Selective information routing by ventral hippocampal CA1 projection neurons - very small yield though, but I'm sure if you combined this with neuropixels recordings you could get the best of both worlds?
Albert CardonaDespite the existence of Voltron, a genetically-encoded voltage indicator, there seems to be a lack of talks and papers featuring this sensor. One wonders why.
Ben de Bivort@albertcardona @elduvelle_neuro @kofanchen getting imaging systems that can record fast enough to see spikes, without blasting out the visual system, is hard. Maybe that's why?
@debivort @elduvelle_neuro @kofanchen
Light-sheet two-photon isn't that rare nowadays?
@elduvelle_neuro neuropixel still way too big for #drosophila brain? But guess good enough for rodents?
@kofanchen Yes definitely fine for rodents! I'm not sure what is the smallest organism they've been used with.. I was thinking maybe they used NP in this butterfly study (Neural representation of goal direction in the monarch butterfly brain but no it's just tetrodes..
Neural representation of goal direction in the monarch butterfly brain - Nature Communications
Neural coding of goal direction remains unclear in insects. Here, the authors describe goal-direction neurons in the monarch butterfly brain that specifically encode the insect’s desired flight direction during spatial orientation.
Jesse MilesI absolutely would have agreed with you when I started grad school, but recent GCaMPs seem very fast and bright!
https://www.nature.com/articles/s41586-023-05828-9
Granted, the rest of what you need to do to get to the point where you can actually collect data (let alone analyze it) is pretty monumental. And obviously still worse temporal resolution than ephys. But ephys isn't a walk in the park either, and those videos of cells fluorescing were what hooked me on neuroscience in undergrad haha. So pretty!
@jessetm thanks for sharing this! They say the half-rise time is 2ms which seems pretty good indeed!
There are other parameters to take into account as well though, like the acquisition rate - this paper I'm reading has it at 20 frames per second (20Hz), which means the minimum resolution is 50ms, if I am not mistaken (ephys is generally sampled at 20 or 30 kHz). So again I am not sure why you'd want to use that for replay / reactivation analyses... There are so many other things one could look at with imaging, but this doesn't seem the best tool for this question.
(I checked the paper you cite and they sample at 200Hz, which is definitely better, but they had to interpolate and average to figure out that half-rise time:
"As the fluorescent signal was sampled at 200 Hz, fast rise times (less than 10 ms) could not be reliably computed for single trials."
)
@elduvelle_neuro yeah, I definitely agree that technology is pretty inaccessible to use well (for the purposes you're describing) if you are not one of the labs that has already gone all in on imaging techniques, but I do marvel at the innovations from that field!
Spencer LaVere Smith@jessetm @elduvelle_neuro Whole cell patch clamp in vivo is still the highest fidelity approach (and my first love). But it doesn’t scale. 1 at a time for pros, 2 at a time for a few people on the planet, and up to 4 at a time is basically just Jouhanneau and Poulet.
Electrode arrays are great for high speed and they catch single spikes, but sample mostly cells very close to the shank. That said, the shank can be long and it is as easy to sample 3 mm deep as it is to sample 0.5 mm deep.
Calcium imaging densely samples many more neurons than electrode arrays. Orders of magnitude more. But the cells have to be superficial (roughly < 1 mm for 2p, < 1.6 mm for 3p). Or you’ve got to remove some tissue. Imaging is slow, but multi spike events are larger so still detected and in ephy spikes are often binned into broader time windows, so the difference is often not that large. You can see spindles in calcium imaging.
Each of these three techniques have unique strengths relative to the others. I like them all.
Redish Lab@sls @jessetm @elduvelle_neuro
I think it's also important to point out that these three techniques (agree - all are important!) sample different aspects of neural signalling. It's not just fidelity or scale.
Patch clamp measures intracellular voltage fluctuations, but those are not necessarily transmitted down axons, and you can have axonal generation of signaling. Electrode arrays (wires, tetrodes, silicon arrays, neuropixels) measure spikes (and LFPs, which are also a different signal, providing different information about the system). Calcium imaging provides information about calcium levels.
Although these signals are correlated with each other, they are not identical and should *never* be substituted one for the other. You can have calcium without spikes, synaptic interactions (intracellular signals) that do calculations, spikes without calcium, etc etc etc.
Instead of asking "can we see replay with calcium signaling?" I think we should ask "what is the calcium signaling doing when hippocampal sequences are replaying?"
That's why we need a literature. That's why we need all of these techniques. The information is synergistic.
Indeed jGCaMP8 are quite something.