Intrinsic imaging tips

Intrinsic signal optical imaging is a functional imaging modality where the reflectance of red light indicates active portions of cortex. It is used for many applications, including imaging individual barrels in rodent somatosensory cortex (e.g., Polley et al. 2004, Aronoff & Petersen 2007), maps in visual cortex (Blasdel & Salama 1986, Kalatsky & Stryker 2003), and the tonotopic organization in auditory cortex (Nelken et al. 2004, Kalatsky et al. 2005).

Here is a generic procedure:

1. Image the surface vasculature with green illumination and take an image for later alignment.
2. Switch to red illumination and grab frames during with and without stimulation.
3. Average frames from the two conditions, then subtract the baseline image from the stimulated condition to determine the active areas of cortex.

Here are some tips for rigging this up in your own lab.

Use a high bit depth camera

You’ll be measuring a very small modulation, on the order of 10^-4. Since 8-bit cameras have only 256 grey levels, it will be impossible to detect signals in single pairs of frames. However, a large amount of averaging will allow detection even with an 8-bit camera.

Determine the time course of the signal

The time course of the signal will vary a bit among cortical areas and model systems. You’ll want to know the time course so that you can design the appropriate acquisition parameters. For example, you don’t want to grab “baseline” or “blank” frames before the signal has actually returned to baseline. Similarly, you want to ensure that you’re grabbing stimulus evoked frames during the peak of the response.

Give pure oxygen

Being a metabolism-based hemodynamic signal, giving oxygen to the animal makes sense, whether using injectable or inhaled anesthetics. Typically this increases the amplitude of the signal.

Use bright, steady illumination

LEDs have really taken over in many areas of microscopy and imaging. They’re available in many different wavelengths, they’re cheap and easy to use, but best of all: they’re rock solid. Arc lamps and other sources can flicker and add terrible amounts of noise, where LEDs are perfectly flat. LEDs have gotten brighter, but they’re still not especially bright compared to traditional sources. So you might want to go with a bank of lights. This is an imaging application where you are not light limited, and in imaging the noise is lower the more photons you measure. So you want to set the illumination so that the camera is just below saturation.

Use a tandem lens macroscope

This is a classic tool for intrinsic imaging. Use two lenses on your camera, face to face. The ratio of their two focal lengths will be the magnification. This setup is nice because it lets in a lot of light, performs minimal magnification, and has a very shallow depth of field. Focus down from the surface vasculature 100-300 µm.

Special notes for Fourier-based intrinsic imaging (pictured above)

Kalatsky & Stryker (2003) introduced this paradigm to intrinsic imaging, where one variable of the stimulus is varied cyclically and then Fourier analysis is used to pick out the phase of response of each pixel in order to determine what part of the stimulus cycle stimulated that area of cortex. For this imaging, averaging is not used, so you’ll need a 12-bit camera, preferably with as much well depth as possible (Dalsa 1M30s and 1M60s are popular choices). In order to avoid artifacts, you’ll also want a frame scan camera, where the entire image frame is read at once, rather that line-by-line. Once the above requirements are fullfilled, go for the fastest acquisition available. Also, timing is very critical for this paradigm, so you need to ensure the stimulus and acquisition are precisely synchronized.