When microscopes and astronomy collide: correcting movement artifacts from <i>in vivo</i> microscopy using a decades old approach to image stars

In vivo microscopy of the central nervous system (CNS), especially of awake behaving animals, is a powerful tool that allows researchers to follow dynamic cellular processes with high spatial resolution in the brain and spinal cord tissues over periods of time ranging from seconds to months (Fenrich et al. 2013). Unfortunately, in vivo microscopy has a major problem – animals’ breath, their hearts beat, and they move around when awake. And when one part of the animal moves, there's a good chance the CNS moves too. All of this movement during image acquisition causes structures of interest to shift position within the image (x-y directions) and to come in and out of focus (z-direction), resulting in movement artifacts. The issue of movement artifacts goes beyond the aesthetics of an image. In the case of functional imaging (e.g. Ca2+ imaging), movement artifacts (especially in the z-direction) can cause huge changes in fluorescent intensities, especially of small structures such as synaptic boutons. Since fluorescent intensity is the primary readout for most genetically encoded optogenetic reporters, changes in intensity due to movement artifacts can be misinterpreted as changes in neuronal activity leading to confusion and erroneous conclusions. What can be done to mitigate movement artifacts? While stopping breathing or the heart beating is normally not practical, and restricting animal movements (either with anaesthesia or restraints) is not useful when imaging awake behaving animals, there are methods to compensate for movement artifacts (Vinegoni et al. 2014). For example, one approach is to move the microscope objective along with the tissues. This optical stabilization approach has been shown to be very effective at correcting for movements in the z-direction using a custom movement compensation controllers and a z-axis piezo objective positioners (Laffray et al. 2011). However, optical stabilization usually requires expensive and technically challenging modifications to microscopes. Alternatively, a more common post-processing approach is to identify common features between adjacent images then shift the images in the x-y directions so that the features are aligned (i.e. co-registration). This approach works well to co-register adjacent optical sections in a z-stack (i.e. volumetric imaging) or to co-register the same optical sections acquired over time. It's even possible to co-register multiple z-stacks acquired over time in the x-y-z directions; however, repeatedly collecting time-series z-stacks takes considerable time, thus reducing temporal fidelity. In this issue of The Journal of Physiology, Ryan et al. (2020) report a simple method for correcting fluorescent intensities due to movement artifacts in the z-direction. The elegance of this approach is that it requires no microscope modifications and functional imaging is done in a single optical plane, which simplifies implementation and increases scanning rates. Their approach, called zCorrect, is based on two observations: First, while it is well known that resolution of an object (or ROI) in the z-direction is lower than resolution in the x and y directions, Ryan et al. found that the fluorescence intensity profiles of ROIs in the z-direction are best described using a Moffat function (called the ROI-spread-function or RSF). Moffat functions have been used for decades by astronomers to account for optical aberrations in the atmosphere (Moffat, 1969), similar to the optical aberrations observed in CNS tissues. The second observation is that fluorescent blood vessels could be used as an anatomical marker to determine the z-position of each frame in a time-series acquisition with movement artifact. Combining these observations, Ryan et al. developed a method to estimate the RSFs of ROIs, determine the z-position of each frame in a time-series acquired of a single optical plane with movement artifact, and correct ROI intensities due to movement artifact. To test their approach, Ryan et al. implanted glass windows over the primary visual cortex (V1) of mice that express the genetically encoded calcium indicator GCaMP6f specifically in the synaptic boutons of VIP interneurons and recorded synaptic activity (GCaMP6f is more fluorescent during synaptic activity). Awake mice were head-fixed below the microscope objective and imaging was done while the mouse was at rest and running on a treadmill to increase movement artifacts. These mice were also injected with fluorescent dyes to stain blood vessels. Image stacks that included all of the RSFs of the synapses of interest were then acquired to generate a reference volume. Immediately after, a time-series image sequence was acquired from a single optical plane within the reference volume. In post-processing, the reference volume was used to estimate the RSFs and to create a blood-vessel ‘map’. In post-processing, the reference volumes were used to estimate the RSFs and to create a blood-vessel ‘maps’. Because there is z-artifact during the time-series, the blood vessel staining of each frame of the time-series is cross-referenced to the reference volume to determine the z-position of that frame. The intensity of each ROI is then scaled for that frame according to its RSF. Similar to previous studies, Ryan et al. observed considerable movement artifacts when a mouse was running compared to at rest. However, unlike previous work showing that VIP neuron activity increases in V1 during running, the raw time-series images indicated decreased synaptic activity. Using zCorrect, Ryan et al. found that the raw data was in fact misleading and that a majority of the synapses had actually increased in activity. These results highlight the consequences of movement artifacts on the interpretation of functional imaging data and the importance of mitigating these artifacts. As in vivo microscopy becomes increasingly popular for studying neural circuitry, the zCorrect approach could be a simple and easy way to correct for movement artifacts when imaging small structures in awake behaving animals. None declared.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.