Cooled Cameras for In Vivo Bioluminescence Imaging

October 30, 2026

Preclinical optical imaging asks a camera to see almost nothing, for a long time, without adding noise of its own. Light from a luciferase reporter or a near-infrared fluorescent probe has to escape through tissue before it reaches the lens, and most of it doesn't. What's left can be a few photons per pixel per second.

That makes the camera the heart of an in vivo imaging system. This post covers what instrument designers should look for in a cooled camera for bioluminescence and fluorescence imaging, using Finger Lakes Instrumentation (FLI) cameras as examples.

Why in vivo optical imaging is a low-light problem

Tissue scatters and absorbs light strongly, especially at blue and green wavelengths, where hemoglobin absorbs most. A bioluminescent signal from a few millimeters deep is spread into a dim, diffuse glow by the time it reaches the skin.

Instrument builders respond in three ways: a light-tight imaging chamber, a fast lens, and long exposures. Exposures of tens of seconds to several minutes are common in bioluminescence work. Over that time, the camera's own noise sources matter as much as its sensitivity.

The noise budget: signal, dark current and read noise

Three things set the noise in a long, low-light exposure:

  • Shot noise from the signal itself, equal to the square root of the number of detected photons. You can only reduce it by collecting more light.
  • Dark current, the thermal signal a sensor produces with no light at all. It grows with exposure time, and its shot noise grows with it.
  • Read noise, added once each time a pixel is read out.

In a short exposure, read noise usually dominates. In a long one, dark current can take over. A good camera choice starts from your expected signal level and exposure time, then checks which noise term wins. Our post on acceptance testing for gain, noise and linearity shows how to measure each term on a real camera.

Cooling and exposure length

Dark current falls steeply as a sensor gets colder, so cooling is the main lever for long exposures. FLI cameras are cooled with thermoelectric (Peltier) coolers, and cooling is specified relative to ambient. Aurora and Revelation cameras cool to about 50 °C below ambient, and ProLine CCD cameras are specified at 60 °C below ambient.

Liquid circulation can take you further. With liquid at about 20 °C, expect roughly 10 °C deeper cooling than with air, as long as the liquid stays above the dew point. Our post on air vs liquid cooling for OEM cameras covers the tradeoffs inside an instrument.

Exposure length also shapes which sensor type fits:

  • Aurora and Revelation use Sony back-illuminated CMOS sensors with very low read noise, and FLI recommends them for long-exposure work as well as life science.
  • The ProLine PL23042 CCD has lower dark current and is easier to calibrate for multi-minute exposures, but it reads out slowly.
  • Kepler cameras are built for fast, continuous, short exposures. FLI doesn't recommend single Kepler exposures longer than about five minutes; stacking shorter frames works better.

Pixel size, binning and the light budget

In a dim, diffuse scene, spatial resolution is rarely the limit. What matters is how much light each effective pixel collects. That's why many in vivo systems bin pixels, trading resolution for signal.

How you bin depends on the sensor. A CCD can generally sum charge on the chip before readout, so the read noise is paid once per binned pixel. The Sony sensors in Aurora and Revelation cameras don't support on-sensor binning, so binning is done digitally after readout. With very low read noise, that penalty is often small, but it belongs in your noise budget.

Native pixel size matters too. The PL23042 has 15 µm pixels and the KL400BI has 11 µm pixels, both in a 2048 × 2048 format, while the Sony sensors use smaller pixels with more of them. Our post on pixel size and sampling walks through how to match pixel size to your lens and field of view.

Red and near-infrared response

Because tissue absorbs blue and green light, the light that escapes is shifted toward the red. Firefly luciferase emission peaks near 560 nm but shifts toward about 610 nm at body temperature. Fluorescent probes for deep imaging are often chosen for the near-infrared window, roughly 650 to 900 nm, where hemoglobin and water absorb least.

So check the sensor's quantum efficiency curve in the red and near-infrared, not just its peak. The camera window matters too: FLI's F116 window transmits farther into the near-infrared, and the window has to be chosen before the camera enters the clean-room build. For fluorescence, emission filters on a filter wheel block excitation light and tissue autofluorescence; see FLI filter wheels and accessories.

Darks, flats and cosmic rays

Quantitative in vivo imaging depends on calibration. Plan for:

  • Dark frames at the same exposure and temperature as your images. Aurora and Revelation cameras don't use a mechanical shutter, so take darks with the chamber dark and the lens covered, or with a blocked filter-wheel position.
  • Flat fields to correct for lens falloff and pixel-to-pixel response, which matters when you compare signal across the field.
  • Cosmic-ray removal. Long exposures pick up occasional bright hits. Taking several shorter frames and combining them with a median removes most of them.

Fitting the camera into the imaging cabinet

An imaging cabinet is a tight space with a lens, a stage, lighting and often an anesthesia line. The camera usually looks down from the top, so its length and its heat matter.

Revelation cameras use the same sensors and electronics as Aurora in a thin, fin-less body. When mounted to your housing as a heatsink, a Revelation is less than 55 mm front to back, and its liquid fittings can exit the side or back. If you're moving an existing design off an aging CCD, our post on replacing the CCD in gel and blot imagers covers a similar transition, and our OEM partner post covers the wider supplier relationship.

Questions to settle before you choose

  1. What signal level and exposure time do you expect from your dimmest target?
  2. At that exposure, does dark current or read noise dominate?
  3. How cold must the sensor run, and can your cabinet support liquid cooling?
  4. Will you bin, and is binning on-chip or digital?
  5. Does the sensor and window response match your reporter or probe wavelengths?
  6. How will you take darks and flats without a mechanical shutter?

FLI can provide a production test summary for each camera by serial number, which helps when you build your noise budget.

Talk to our sales team

Since 2000, FLI has shipped over 20,000 cameras worldwide. FLI cameras are Made in USA, and our sensor chambers are assembled in our 1,000+ sq ft ISO Class 7 clean room. See how on our in-house manufacturing page.

Send our sales team your target wavelengths, typical exposure times and cabinet constraints, and we'll help you pick a sensor, cooling method and form factor for your imaging system. Request a quote or email sales@flicamera.com.