Genetically encoded voltage indicators combined with light sheet microscopy show promise for parallel recording of neuronal membrane potentials. However, the high acquisition speeds needed (500-1000 Hz) typically limit imaging to a single focal plane due to the challenges of fast focusing through 3D samples. While moving the imaging objective is slow, remote focusing offers a faster alternative. Existing remote focusing methods, utilizing either tunable focusing elements in a Fourier plane or focusing on a remote image plane, suffer from drawbacks. The latter method, while offering fast, aberration-free imaging, loses more than 50% of light due to polarization optics (a quarter-wave plate and polarizing beamsplitter). This light loss significantly impacts signal-to-noise ratio (SNR), especially in high-speed voltage imaging with sub-millisecond integration times. Therefore, existing remote focusing approaches have not been widely adopted for high-speed, volumetric voltage imaging, instead primarily used for fast scanning in two-photon microscopy. This paper addresses this limitation by presenting a new remote focusing design to improve light efficiency and enable high-speed volumetric imaging.

Light sheet microscopy in combination with genetically encoded voltage indicators has shown potential for recording the membrane potential of numerous neurons simultaneously. However, the high acquisition rates necessary for voltage sensors typically restrict imaging to a single focal plane. Existing remote focusing techniques, which aim to address this limitation by moving the focusing element away from the objective and sample, are either too slow (due to the inertia of moving the objective) or too light-inefficient, losing a substantial portion of light through polarization optics. This inefficiency severely impacts signal-to-noise ratio (SNR), particularly crucial for high-speed voltage imaging.

The proposed method, FLIRF (flipped image remote focusing), enhances light efficiency by replacing the conventional beamsplitter and quarter-wave plate with a microscopic retroreflector. This retroreflector flips and folds the image, spatially separating the incoming and outgoing light. The system uses a 16×0.8 NA water-dipping primary objective and a 2×0.75 NA air remote objective, with a custom-built retroreflector from two 0.5 mm aluminum-coated right-angle prisms. A non-standard 124 mm focal length for the remote system tube lens provided an angular magnification of 1.29, close to the ideal value. The retroreflector's axial movement, achieved with a linear coil motor, allowed for volumetric imaging at up to 500 Hz over a 150 µm range. The point spread function (PSF) was measured at various z-positions to characterize the system's resolution and intensity. Zebrafish larvae expressing the voltage indicator Volttron2 were used to demonstrate FLIRF's capabilities in volumetric voltage imaging. Data acquisition involved 20-second recordings of spontaneous neuronal activity at 500 frames/sec, with data alignment, motion correction, 3D segmentation, and signal extraction performed post-acquisition. Motion correction accounted for the non-linear trajectory of the retroreflector and the camera's rolling shutter. 3D segmentation was done using Cellpose, and signal extraction used an adapted method to extract fluorescence from the 3D ROIs.

FLIRF achieved volumetric imaging at up to 500 Hz across a 150 µm z-range and a 358 µm x-range (limited by the retroreflector size). The system demonstrated a maximal lateral resolution of 0.53 µm, maintaining 80% of maximal intensity over a 100 µm z-range at the center of the field of view. In voltage imaging experiments, FLIRF successfully recorded spontaneous activity from over 100 neurons in the zebrafish spinal cord simultaneously, with clear observation of both rhythmic membrane oscillations and individual spike activity. Minimal reduction in action potential SNR was observed over the 20-second recording duration, despite a measured fluorescence bleaching decay constant of τ = 72 s. The light efficiency of FLIRF was nominally doubled compared to beamsplitter-based systems, a significant improvement for high-speed voltage imaging.

FLIRF successfully addresses the limitations of existing remote focusing methods by providing a light-efficient and high-speed approach for volumetric voltage imaging. The two-fold increase in light efficiency compared to beamsplitter-based systems is crucial for improving the SNR of high-speed recordings. While sacrificing half the field of view, the approach offers a valuable trade-off for achieving the necessary light sensitivity. The ability to record from over 100 neurons simultaneously demonstrates the method's potential for high-throughput screening in neuroscience. The observed effects of the non-linear movement of the retroreflector and the camera's rolling shutter highlight the need for careful consideration and correction in data analysis.

FLIRF represents a significant advancement in volumetric voltage imaging, offering a light-efficient and high-speed solution. Its successful application to recording neuronal activity in zebrafish demonstrates its practical value. Future work could focus on further optimizing the system's speed and imaging range, exploring alternative retroreflector designs, and expanding its applicability to other biological systems.

The current implementation of FLIRF sacrifices half of the field of view for improved light efficiency. The retroreflector's non-linear movement and the camera's rolling shutter necessitate digital correction during data processing. The volume that can be imaged at a given rate depends on both camera speed and the voice coil motor's capabilities, potentially limiting applicability to densely labeled samples.