Hybrid light-sculpting-scanning for versatile high-speed calcium imaging

We have developed strategies for optimizing of the present tradeoffs in light-sculpting microscopy by evaluating various scanning modalities theoretically and experimentally and identifying approaches to maximize the obtainable volume speeds, and depth penetration in scattering brain tissue using different laser systems.

To understand how dynamics of neuronal networks are related to brain function tools that allow capturing the functional dynamics of large cellular populations at high speed are required. One approach is based on temporal focusing (TeFo). We have recently applied wide-field TeFo as a technique for high-speed brain-wide calcium imaging in C. elegans. (Links: Paper or Overview) The effective decoupling of axial and lateral confinement of the excitation beam in TeFo allows to excite laterally wide yet axial confined volumes which in turn allow the use of parallel, camera-based detection schemes. While using this approach a large area can be in principle excited simultaneously the actual increase in effective frame (or volume) rate also depends on the available laser power, pulse repetition rate and peak pulse energy.  Moreover, such a wide-field approach to volumetric imaging may also poses technical limitations due to the required high peak power, which can usually only be provided by amplified laser systems. Finally the parallel detection scheme via cameras poses limitations on the obtainable depth due to scattering which leads to cross talk between neighboring pixels.

In this work we have addressed some of these shortcomings. First, we investigate the present tradeoffs in light-sculpting microscopy both theoretically and experimentally, evaluating various scanning modalities and devising strategies to maximize the obtainable volume speeds, and depth penetration in brain tissue based on the experimental requirements. Doing so we found that the imaging field-of-view can be increased to >200×200 µm2, while retaining a physiologically relevant imaging speed of 10 ms/plane by using line- or spiral-scanning methods in combination with temporal focusing (see Fig.1).

Fig. 2 Trading off area versus imaging speed and fluorescence for different excitation modalities. A. Theoretical estimates and experimental measurements of the fluorescence signal as a function of the excited area for different combinations of laser systems and excitation modalities. Numbers indicate parameter configurations at which experimental data were obtained with the respective images shown in D. Dashed lines indicate regions for which the scanning speed of the galvo mirrors would not be sufficient. The values on the y-axis are typical signal intensities for practical calcium imaging with Na> 1⋅107 representing an empirical lower bound (dashed grey line). B. Time required for scanning an area in order to achieve Na~ 3⋅108 hence sufficient signal for imaging as discussed in the main text. Coloring identical to A. C. Isolines for sufficient fluorescence signal (Na~ 3⋅108) during different image exposure times in dependence on scanned and non-scanned directions for a line-scan configuration (w ~3 µm) using an amplifier system. D. Images of convallaria rhizome, taken with different configurations at 10 % of available power and t<sub srcset=
exp = 10 ms exposure time per plane. The scale bar applies to all subfigures.” width=”1134″ height=”590″ /> Figure 1: Trading off area versus imaging speed and fluorescence for different excitation modalities. A. Theoretical estimates and experimental measurements of the fluorescence signal as a function of the excited area for different combinations of laser systems and excitation modalities. Numbers indicate parameter configurations at which experimental data were obtained with the respective images shown in D. Dashed lines indicate regions for which the scanning speed of the galvo mirrors would not be sufficient. The values on the y-axis are typical signal intensities for practical calcium imaging with Na> 1⋅10^7 representing an empirical lower bound (dashed grey line). B. Time required for scanning an area in order to achieve Na~ 3⋅10^8 hence sufficient signal for imaging as discussed in the main text. Coloring identical to A. C. Isolines for sufficient fluorescence signal (Na~ 3⋅10^8) during different image exposure times in dependence on scanned and non-scanned directions for a line-scan configuration (w ~3 µm) using an amplifier system. D. Images of convallaria rhizome, taken with different configurations at 10 % of available power and texp = 10 ms exposure time per plane. The scale bar applies to all subfigures.

In particular, we also found that one can achieve an imaging FOV of ~150 µm diameter at this imaging speed using galvanometer mirrors and a standard pulsed Ti:Sa laser instead of an amplifier laser system. The experimental setup, the various scanning modalities is shown in Fig. 2 and typical experimental results can be seen in Fig. 3.

BioMedOptExp_website_fig1
Figure 2: (A) Schematic of the setup including its various modalities. The laser beam, from either a Ti:Sa source or an amplified system, is expanded via lenses and directed towards a pair of galvo mirrors, after which, depending on the scanning modality either no lens (wide-field), a cylindrical scan lens (fcyl – line-scan 1D-LS) or a spherical scan lens (fsp – spiral-scan 2D-SS) are employed. (B) Left panel illustrates the schematics of various excitation modalities, including wide-field (top), line-scan 1D-LS (middle) and spiral-scan 2D-SS (bottom), while the right panel shows the experimental excitation pattern taken with a fluorescent plastic slide.

Furthermore, in order to suppress the detrimental effects of scattering we used the line-scan TeFo modality in combination with a synchronized rolling shutter read-out mode of our sCMOS camera, in which the detection area of the camera is restricted to a small line that moves synchronously with the illuminated line in the sample. The reduction of the active sensor area in one dimension to the width of a line allows to detect photons originating from the line-shaped illumination area on the sample only, while rejecting scattered photons. Again, we theoretically evaluated the expected effects of scattering reduction at different depths using Monte Carlo simulations, and then validated our theoretical predictions experimentally in acute mouse brain slices expressing GCaMP6m.

BioMedOptExp_website_fig2
Figure 3: Ca2+-imaging of acute mouse brain slices at various imaging depths, and using different TeFo excitation modalities. Individual somata as well as neuropil can clearly be seen at different depths as indicated in the top right corner. (A) Ti:Sa 2D-SS TeFo (B) Amplifier 1D-LS TeFo. All images were taken with an exposure time of 10 ms and no frame averaging has been performed. (C) Extracted Ca2+-signal from an acute brain slice that showed brief and sporadic spontaneous activity, imaged with 200 µm FOV 2D-SS TeFo. The blue trace shows the baseline of an inactive neuron, while the black trace corresponds to another neuron showing brief bursts of activity. Scale bar is 50 µm across all images.(A)

Relevant publication:
P. Rupprecht, R. Prevedel, F. Groessl, W. Haubensak, & A. Vaziri
Optimizing and extending light-sculpting microscopy for fast functional imaging in neuroscience
Biomedical Optics Express 6:2, 353-368 (2015). (Download)