Analyzing Microscopy Images? What You Should Know About Dynamic Range

Ever tried to turn the volume all the way up on a small radio or small stereo system? (Hopefully you have not tried it with earphones in!)

Notice how, after some point, the sound didn’t get any louder- it just got more distorted? That’s because you’ve hit the ceiling of your machine’s dynamic range.  It’s called ‘clipping’ distortion, because what should be the peak of sound intensity (amplitude) gets clipped by the inability of the amplifier to produce more current.

After some point, the soft passages of music sound loud, and the loud passages do not sound any louder- they just sound more distorted. It’s basically a horrible thing to do to your music, your equipment and your ears!

Audio signals are not the only ones that can be mistreated like this; you can easily abuse images the same way. You can turn up the gain of your microscope camera, photomultiplier, or whatever detector you are using so high that you get clipping.

In this article, we’ll look at what causes clipping and saturation in microscopy images, followed by definitions of Dynamic Range, Bit Depth, Intrascene dynamic ranges, High Dynamic Range, and the quantification of image data. For broader acquisition and processing advice, see how to get your microscopy mojo back.

What does that noise look like?

So what does clipping actually look like in an image? Flat, featureless colors of the highest intensity.  From a certain level of brightness and upwards, all features look equally bright- ‘saturated’. The same thing can happen at the other end of the scale; below a certain threshold of brightness, all features can be made to look equally dark (black).

These general rules are true whatever the nature of the signal we wish to record, and whatever the means of recording. Whether it be a pulsating needle engraving a phonograph record, a microphone converting sound waves to electrical current to cause magnetic changes on a tape, a photographic emulsion, a digital audio or image recording setup. The issues related with dynamic range and signal-to-noise ratio are not just a function of the user settings (gain setting, for instance, as mentioned above) but also of the hardware involved.

What exactly is Dynamic Range?

Dynamic range can be expressed in decibel, in bits, or as an exponent (e.g. 100 db = 17 bits = 10⁵). The way of digital encoding of different signal levels is directly reflected on the final dynamic range of the recorded data. The dynamic range of digitally encoded data is a measure of the range of discrete amplitudes that the system can resolve. Limitations to this resolution can come from two sources:

1. The first is the detector itself. Different types and different models of detectors (be they CCD, CMOS, PMT, Avalanche, Hybrid etc.) have different dynamic range properties.
2. The second is the analogue-to-digital (A/D) converter. A proper matching between the two has to be made, to allow discrimination of as many gray scale steps as possible. Some detectors can have very high dynamic range values, and the appropriate converter is required, so that this dynamic range is not wasted. While the dynamic range cannot be larger than the effective grayscale level resolution of the A/D converter, it can be smaller if, for example, there is saturation in the photodetector.

Bit Depth

Bit depth is a related concept that refers to the binary range of possible grayscale values utilized by the A/D converter to translate analog image information into discrete digital values capable of being read and analyzed by a computer. For example, 8-bit A/D converters have a binary range of 2⁸ or 256 possible discrete values, while a 12-bit converter has a range of 2¹² or 4,096 discrete values, and a 16-bit converter has 2¹⁶, or 65,536 discrete values. The bit depth of the A/D converter determines the size of the gray scale increments, with higher bit depths corresponding to a greater range of useful image information available from the camera- if the camera is giving this information to begin with, that is.

Intrascene Dynamic Range

The dynamic range which can be detected at the same time in the same field of view (between maximum and minimum intensities) is known as the ‘Intrascene Dynamic Range’. Interscene dynamic range is the range of intensities that can be detected when the detector gain, integration time (photon collection time) and other variables are adjusted for different fields. There is another parameter to consider- that is the amount of noise present along with the signal.

Effective Dynamic Range

The effective dynamic range is often calculated to be the maximum signal that can be accumulated, divided by the noise associated with reading the signal. Signal-to-noise ratio refers to actual signal and noise levels for a given image, so its highest value cannot be more than that of the dynamic range. We mentioned in the first part that we can lose detail if below a certain threshold of brightness, all features look equally dark. We can now explain that these features would be at the most as intense as background noise.

Don’t forget the noise

When the contribution of noise is considered, what we have is essentially a raise of the signal level ‘floor’, below which there is insufficient contrast between specimen features that would enable us to distinguish them as separate. Inevitably, this influences the effective resolution of the system. When plotting contrast-distance functions to demonstrate resolution limits, we often forget that the distance required to provide sufficient contrast to resolve two points in an image is increased when noise is taken into account. We speak about resolution as if noise is zero and the dynamic range is infinite, but in real life a constant noise level limits the useable contrast and point separation range.

A range we can’t see

The human visual system is capable of about 5 to 7 bit discrimination. Compressed digital TV signals evidently are capable of less, as one can clearly see discrete grades in the colors of a sunset or similar scenes. So, why is it that, while we are limited to 7 bit, and most computer screens to 8 bit (256 gray levels), there are high dynamic range image recording systems around?

Larger number of gray levels=accurate data

There are indeed valid reasons. One is related to quantitation. Regardless of whether we can see the difference or not, larger numbers of gray levels allow light intensities to be more accurately determined. Additionally, when multiple image-processing operations are being performed, image data sets which are more precisely resolved into many gray level steps can be subjected to a greater degree of manipulation without the appearance of artifacts.

Isolate and expand

We should not forget that the dynamic range of the image is not necessarily the dynamic range of our region of interest. We may decide to select for display only a part of the recorded image. That part may be a ‘gray’ part, with little visible contrast. When we isolate it, we can expand its contrast levels to occupy all 256 levels of an 8-bit monitor or print. If the image had been taken with low bit depth to begin with, then this expansion would result in artefacts- in visible discernible ‘grades’, such as those seen in digital TV, instead of smooth tonal gradations. Having many gray levels may seem redundant in the original image- but may be just what we need when we focus on a specific region of interest.

High Dynamic Range

And a final note, to avoid any misunderstandings, High Dynamic Range (HDR) in photography does not involve 16-bit detectors or anything that fancy! It involves multiple exposures which are combined to produce a single image that contains the information obtained from all of them.

It can be very useful sometimes- but its generalized use and abuse can result in some aesthetically rather unfortunate photographs, which, none the less, are proudly displayed in the social media by their owners!!! HDR can have a value in microphotography, too, if the range of the dynamic range that one wishes to record in a single field exceeds the dynamic range of the detector.

Dr Nick Oswald

Nick has a PhD from the University Dundee and is the Founder and Director of Bitesize Bio, Science Squared Ltd and The Life Science Marketing Society.

Share this article: