Are there any microscopy modalities or techniques for in vivo imaging at higher resolutions than the diffraction limit?
I was looking at this list: Super resolution microscopy, but they don't talk about in vivo imaging.
There have been experiments with Live-cell dSTORM with SNAP-tag fusion proteins in Markus Sauer's lab in Würzburg. If you are using the physiology definition of in-vivo, I'm pretty sure some of my co-workers have measured on live organisms. However, this is work in progress and not fully published; if you are interested, you should ask Markus Sauer directly.
You can use the dSTORM-like techniques (STORM, dSTORM, GSDIM, RPM, etc.) in live cells because the chemical environment in the cell is favourable and the redox system of the cell works for you. The key problem is to get the fluorophores into the cell, but SNAP-tags can help you to do it.
As one of the co-authors, I can tell you that live cell was some pain, but considerably less than getting 3D localization microscopy to work.
PALM is a different story because you generally have longer acquisition times. STED is difficult: If the cell wasn't dead before, the STED beam will ensure it is, either through the intensity or by unbalancing the chemical environment sufficiently. You might have your image before the cell dies, or maybe not.
Since the question was asked, the field has seen tremendous growth and even a Nobel prize has been awarded for Super-resolution microscopy. Since other answers have already touched upon how STED, PALM, STORM could help, I would talk about Expansion Microscopy (ExM) and how it has been recently used to image intact zebrafish brain tissue and gastrulating zebrafish embryos [Freifeld et al]. Since the optical resolution is limited by the wavelength of light which gets diffracted if the object being imaged is smaller than the wavelength of light (~300 nm). In order to image small object, what if we could enlarge the object itself to cross the diffraction barrier? This has been done by a team of researchers at MIT who imaged mouse brain tissue in 2015 [Chen et al].
Chen et al. have solved this problem by digesting the mouse brain away after transforming it into a polymer gel but before inducing matrix expansion. But how can one study a brain in which all the neurons are gone? Here, a second idea came into play: For some questions, it would be enough to study a “shadow” of the brain in the polymer matrix. This was realized by first marking molecules of interest in neurons with antibodies. The antibodies then were labeled with fluorescent markers that bind both to the antibodies and the matrix. The brain was then digested away and the matrix expanded. The fluorescent neuronal shadows also expanded and could be studied in “superresolution.” Chen et al. could thus increase specimen size by about a factor of 5 while preserving general morphology, and could visualize cultured neurons and brain slices at 70-nm resolution. This allowed them to observe proteins localized to synapses.
The best part is that conventional confocal microscopes can be used to image these enlarged samples, so the imaging speed is also good.
Any superresolution techniques relying on stochastic sampling (e.g. PALM, STORM) is extremely difficult to accomplish in vivo, as the tissue would be moving (e.g. due to respiration) and this would hinder a lot the reconstruction of the image.
* Note that the images in that article come from this paper, which is not done in vivo.
Thanks to @vkehayas for pointing out that in vivo STED has been done for instance here Nanoscopy in a living mouse brain
Yes. A decade ago I've been working on a software for microscope that does what you want.
It's optical, however it produces Z map instead if image, and has resolution is order of magnitudes below the diffraction limit: http://www.amphoralabs.com/production/10404/10708