The connectivity between neurons is most definitely an important piece of the information needed to understand the structure of the brain, but it doesn’t paint the entire picture. The finer structures within the soma, dendrites, axons and synapses contain information about the strength of the connections and the processing that takes place within these components.
My question is, apart from the simple connectivity, is there enough resolution in these images to form a judgement as to these finer details so that things like the differentiation between inhibitory or excitatory neurons, prediction of neuron spiking patterns, AP propagation speed and so on, can be determined?
It seems to me these are important bits of information that will need to be captured in order to truly determine the state of the brain.
Welcome! The data set that we are currently working with is stained such that only the cell walls show up. The good part about this staining method is that it makes automatic segmentation much easier. That translates to the chunks that you guys are working with being larger and more correct. The bad news is that all those great things that you are talking about don’t show up in the results.
There are/We have other data sets which are stained differently and they include all sorts of organelle goodies, vesicles, etc. The problem with them is that our automatic segmentation algorithms don’t work on them nearly as well. At this point, we aren’t ready to run an Eyewire like project on those data sets, but as we improve the algorithms, we should be able to work with them as well.
Good to know.
Have a look at some of prof Seung’s comments on that topic in this thread…
Yes this dataset is not perfect, but that’s always the case. We should still be able to discover some interesting science in it.