Fig. 2: The optical design of the FANTASIA system.

a A single-pair AOD based two-photon imaging system with a simple architecture. BE (beam expander); M (mirror); DpCG (dispersion compensation grating); SRR (sliding retroreflector); DCU (dispersion compensation unit); HWP (halfwave plate); AODs (acousto-optical deflectors); DvCL (divergence compensation lens); SL (scan lens); TL (tube lens); TM (turning mirror); DF (dichroic filter); Obj (objective); Sub (subject); CLs (collection lenses); PMT (photomultiplier tube). b–d Schematics of scanning modes supported by the system. A basic scanning mode supported by an orthogonal pair of AODs is 2D random-access pointing (b). The system further supports 2D fast raster scanning (c) by chirping the AOD control signals on both axes at the same speed, while the beam divergence generated by the AODs is compensated with an additional convex lens (DvCL). The focal plane depth can be additionally shifted by adjusting the chirping speed of the AOD control signals, resulting in a 3D multi-layer raster scanning mode (d). e The effect of AOD chirping speed on the frame rate, the DvCL needed, and the FOV size in the 2D fast full-frame raster scanning mode in (c). The AOD chirping speed is expressed in line scan time, the time to linearly chirp across the AOD bandwidth (30 MHz). A shorter line scan time would result in a higher full-frame raster scanning rate, a need for a DvCL with a shorter focal length to fully compensate for beam divergence, and a wider dark periphery in the FOV due to the necessary transition time between the scanning lines. f Two-photon image acquired with the 2D fast full-frame raster scanning mode demonstrated in (c). The image is of a neuronal population labeled with GCaMP6s in marmoset auditory cortex at a ~200 μm depth. Scale bar: 100 μm. g The effect of AOD chirping speed on focal depth shift in the 3D multi-layer full-frame raster scanning mode (d) with an f = 1000 mm DvCL. The AOD chirping speed is expressed in line scan time as in (e). This effect is also DvCL- and objective- dependent. h Two-photon images acquired with the 3D multi-layer raster scanning mode demonstrated in (d). The five scanned layers are spaced by a 5 μm inter-layer distance. The images are of a testing slide (Invitrogen F36924) tilted at a 3.4° angle (~25 μm depth change over the FOV). Scale bar: 100 μm. i The spatial dispersion effect demonstrated by wide-field images of two randomly accessed spots pointed on a fluorescence reference slide (Ted Pella 2273) without (uncompensated) and with (compensated) the DCU. These two points are at the opposite corners of the FOV (x-, y+ and x+ , y-). Scale bar: eq. 1 mrad of AOD scanning angle. j The temporal dispersion effect on laser pulse width. The autocorrelation functions were measured at the laser output (original) and after the AODs without (uncompensated) and with (compensated) the DCU. Pulse width estimation is based on sech² deconvolution of 0.65 times autocorrelation. k The two-photon point spread functions (PSFs) of the system. The PSFs were measured with the Thorlabs 10x, 0.5 NA objective on 0.5-µm microbeads (1^st and 2^nd row) and with the Olympus 25x, 1.05 NA objective on 0.2-µm microbeads (3^rd row). Measurements were taken under random access pointing mode (1^st row) and raster scanning mode (2^nd and 3^rd row). Scale bar: 1 μm. FWHM (lateral, axial): 1.17 ± 0.22 µm, 6.38 ± 0.19 µm (1^st row); 1.45 ± 0.23 µm, 7.78 ± 0.65 µm (2^nd row); 0.56 ± 0.06 µm, 1.42 ± 0.04 µm (3^rd row)