Neuroscience: Mesoscope provides 100X imaging volume at high resolution

Neuroscience has shown that even relatively simple behaviors involve multiple brain regions. Advances in laser-power optimization and efficiency of two-photon excitation have brought effective tracking of neuron populations across these regions within grasp.

But as scientists at Howard Hughes Medical Institute's (HHMI's) Janelia Research Campus (Chevy Chase, MD) note, carrying out this kind of tracking is impossible with standard equipment. Microscopes providing sufficient resolution have fields of view (FOV) of <1 mm—narrower than most brain areas, and those with FOV large enough to capture an entire brain cannot resolve individual neurons.

Aiming to overcome these limitations, researchers in Karel Svoboda's HHMI lab have developed a two-photon random-access mesoscope (2p-RAM) that is optimized for high-resolution in vivo applications, with an imaging volume ~100-fold larger than other microscopes with comparable resolution.¹ Providing a 5 mm FOV, the instrument can image within a volume spanning multiple brain areas (5 × 1 mm cylinder) with near-diffraction-limited resolution over a broad range of excitation wavelengths (see figure). And the researchers have demonstrated the system's ability to capture neural activity in noncontiguous brain regions of transgenic mice expressing protein calcium sensors.

The researchers laid out key specifications for their mesoscope: They chose a 5 mm FOV to enable imaging of most cortical areas of mouse brains that are nearly coplanar. To achieve axial resolutions smaller than a typical soma, they specified a numerical aperture (NA) of 0.6, and to maximize fluorescence signal collection, they specified a 1.0 collection NA. To achieve diffraction-limited performance and high two-photon excitation efficiency using fluorescent protein sensors based on green fluorescent protein and various red fluorescent substances, they specified a 900-1070 nm spectral range.

Operation, implications, commercialization

Because neurons of interest often reside at different depths (either in the same or different regions), fast axial scanning is crucial for mesoscale functional imaging of intact brains. So the 2p-RAM was designed to change focus quickly by using a remote mirror that moves nearly conjugate to the specimen plane. Passing the beam through a remote-focus (RF) objective produces aberrations complementary to the imaging objective—doing so prior to lateral scanning enabled a reduction in size, complexity, and cost for the RF objective. Optimizing both coatings and the RF objective's element count reduces power loss, too.

The instrument's serial scanning capability enables both low-magnification imaging across large areas and high-resolution imaging across smaller, but widely separated, areas. Online analysis of low-magnification imagery facilitates identification of brain areas with particular activity patterns, which, once located, can be interrogated using high-resolution imaging to visualize individual neurons.

To maximize the delivery of peak power for any given average power, the researchers performed group delay dispersion compensation. This allows the use of greater peak power to maintain a consistent signal-to-noise ratio during rapid scanning, with shorter pixel dwell times. The researchers report that this fast scanning, in combination with briefer jump times between regions, will enable simultaneous tracking of tens of thousands neurons across the cortex at 10 Hz. This type of performance is necessary, the researchers note, to achieve single-trial decoding of neural activity patterns, which will allow correlation of behavior with underlying neural populations and enable study of a wider variety of behaviors, particularly more natural and unconstrained behaviors.

Thorlabs (Newton, NJ) has entered into a licensing agreement with HHMI to commercialize the 2p-RAM, and expects to have the instrument available for sale in Q1 2017.

REFERENCE

1. N. J. Sofroniew, D. Flickinger, J. King, and K. Svoboda, eLife, 5, 14472 (2016). doi:10.7554/eLife.14472.