Whenever you try out a new method or piece of equipment in the lab, you’re bound to have many questions. Perhaps one of the most important questions is a rather simple one: are these data that I’m collecting any good? Am I recording a decent signal, or is this noise that I can’t yet recognize in real time?
Sometimes it is hard to answer this question until you’ve progressed some part of the way through your analysis pipeline. Still, it’s never a good feeling to realize you’ve spent time processing data that could have been much nicer if you had tweaked some settings during the collection phase. Here, we’ve laid out some important considerations to give you a jump-start on the road to learning how to identify good photometry data in real time. If you are brand new to fiber photometry, we recommend checking out our resources to get familiar with the basic concept behind the method.
Getting Connected and Checking for Signal
When you first connect your animal to your fiber photometry system, you’ll want to check to see if you have any fluorescent signal. To check for signal, you should set your driver box to constant mode and turn on the LED corresponding to the color of fluorophore you are interested in recording from (470nm if imaging a green fluorescent indicator, 560nm if imaging a red fluorescent indicator).
A live depiction of your fluorescent trace is shown as a scrolling trace on a graph that updates in real time. The x-axis of the graph represents time, and the y-axis represents the relative fluorescence of your indicator. Remember that this raw signal trace is a representation of the bulk fluorescence captured by the camera. The measurement of this signal is based upon the mean pixel value recorded by the camera, or “F.” These are arbitrary units and will vary between subjects and recording sessions. Therefore, conclusions about the absolute strength of your indicator’s fluorescence cannot be drawn based solely on the numerical value reached on the y-axis of this live graph. To make accurate comparisons between recording sessions, experimental conditions, or other variables in your fiber photometry experiment, you’ll need to perform some post-hoc analysis.
Even before the analysis phase, it is possible to make some assessment of the quality of your signal during the recording. For example, if you are recording from a mouse in an open field, you may be able to see signal changes that correlate with events such as rearing or reactions to external stimuli. You may also be able to evaluate the kinetic features of your trace in relation to the known kinetics of your indicator. As a general rule, the kinetics of the these bulk signals cannot exceed the kinetics of the indicator; if you’re recording with GCaMP6s, for example, and see a peak with a rise time of 50ms, this peak exceeds the kinetics of GCaMP6s and is therefore necessarily an artifact rather than true signal. Paying close attention to kinetics and correlations with the behavioral environment can impart some confidence that your signal is “real” and not merely the result of confounding variables, such as motion artifacts.
Recognizing Motion Artifacts
Signals may fluctuate as a result of activity-independent variables, like motion. Such artificial variation can considerably undermine your data. There are some controls in place to mitigate motion artifacts (including the isosbestic signal, a concept we will be reviewing here at the FP Academy in the coming weeks), but you still want to be on the lookout for evidence of such a confound.
When recording in one of the trigger modes, motion artifacts often appear in live traces as steep parallel deflections; the top and bottom of the trace will move in the same direction (either upward or downward) simultaneously. This parallel motion is the result of both your calcium-dependent and calcium-independent signal fluctuating in sync with each other, which by definition means your signal is not activity dependent.
You may also note kinetics that starkly contrast with the expected kinetics of your fluorophore. For example, you may record a signal with a sharp signal decrease followed by a slow increase; the opposite of established GCaMP kinetics. This is the result of an optical fiber that is exceeding its bend radius, which limits its ability to transmit light (both to and from the brain) and ultimately attenuates your signal.
If you suspect that you are recording some motion artifacts, a good place to start troubleshooting is to practice good patch cord management. The most common cause of motion artifacts is not a result of the brain moving relative to the implant, but rather the patch cord bending. Smooth out any sharp angles or kinks in the patch cord, and let it drape or hang in a way that bends as little as possible. Take care to use a patch cord that is just the right length for your setup — not so short that it impedes the animal’s movement, but not so long that it gets tangled up. If you’re following good patch cord management practices, it goes a long way towards mitigating motion artifacts. Even if you think your signal looks great, take care to practice good patch cord management; no amount of isosbestic control will compensate for a poorly managed patch cord.
One of the nice things about photometry data is that it generally has a high signal-to-noise ratio (SNR) which is interpretable before you process the data. A quick way to evaluate the SNR is to look at the “smoothness” of your live trace on the auto-scaled graph. A clear and crisp trace that appears as a single line likely has good SNR.
If you’re looking at a very rough (“noisy”) trace, the SNR may be relatively poor. You are probably looking at the read noise of your camera — the same thing you would see if the camera was recording but all of the LEDs were off. This “roughness” takes the form of fluctuations that are usually less than one pixel. If the trace is “smooth,” the auto-scaling is probably hiding this read noise.
Understanding Live Traces in Trigger Modes
While most people use constant mode when checking for signal, your actual experimental recordings will typically utilize one of the trigger modes. These trigger modes will have unique patterns of LED on/off cycles, and it is important to know which LEDs are on at any given time when interpreting your signal. Refer to the user manual to familiarize yourself with the different trigger mode LED patterns prior to recording.
When you first turn your driver box on, one thing will always be true: the 470nm LED will be first in the sequence and therefore be your starting point. Since the recorded signal is an interleaved combination of up to three unique signals, the primary raw data trace will take on a zig-zag pattern as the LEDs alternate. By incorporating a slice node into your Bonsai workflow, you can separate the live traces so that you are visualizing each real-time signal on its own.
This guide is intended to serve as a primer for understanding raw fiber photometry data; the best way to have a comprehensive understanding of what to look for while you’re recording is to practice! The Neurophotometrics team is always available to help you evaluate the quality of your raw signal traces and understand the components of the trace more clearly — just use our contact form to get in touch.
This post was written by Caroline Sferrazza and Gemma Deegan.