- The talk at imaging and spectroscopy on mNeptune looks like it will be incredibly useful. It's a new fluorescent protein that just sneaks into the water window. You can excite it with 633 nm, and detect it at 650, allowing you to do intravital imaging while avoiding hemoglobin absoprtion. I suspect many people will find that very useful.
- Two different talks on superresolution imaging of neurons are starting to show the real promise of superresolution imaging. The attempt to use it to trace all the neurons in a brain sample looks very ambitious, and I appreciate ambitious. Of course, the problem is it's sloooowwwww. If you have to fix the cell, why can't you use EM? That's a genuine question, by the way. Why can't you just do EM on brain slices and trace neurons that way, with way more resolution than you can get with any fluorescence technique? There must be a reason this doesn't work. Are they hard to stain with EM probes? Do they degrade rapidly under EM? Inquiring minds want to know.
Monday, February 22, 2010
On a more positive note
I'm a bit of a curmudgeon (actually, I'm a lot of a curmudgeon) so it's easy for me to complain, and as a postscript, I thought I'd say some nice things:
Subscribe to:
Post Comments (Atom)
3 comments:
Dmitri Chklovskii at Janelia Farms is aiming to do that sort of EM-based reconstruction. I'm afraid this description is a couple of years old, though. Alignment of overlapping images is a big issue, but, he says, a soluble one (at least at the Drosophila level).
Jeff Lichtman's lab is also working on EM based approaches to trace neurons in ultra-thin (~40nm) brain slices. This is also very ambitious and incredibly challenging. Fluorescence provides some advantages, like color (you don't need to trace as much if you can distinguish some based on color). Also another big advantage is molecular specificity for synaptic markers, which is hard to get with EM. One of the major problems with EM so far has also been staining artifacts, nonspecific staining that forms huge chunks that obscure a lot of detail. This is not as much of a problem with fluorescence. I think both approaches are worth pursuing in the end.
EM can't hit the resolution on the nail, and you need to rely on sections. Reconstructing 3d datasets from sections, anyone can tell you, is a nightmare and highly innacurate even with the best stack alignment algorithms. Getting data in situ, even at lower resolution, is the only real way to do it.
Post a Comment