Imagine being a patient hooked up to a number of sensors intended to measure brainwave activity. At the end of the test, your diagnostic technician makes mention that she observed a lot of ‘noise’ in your images. What does that mean? More importantly, do you have a serious health problem? Sometimes the noise picked up by medical sensors makes it difficult for doctors to figure out what is going on.
At Rock West Solutions in Goleta, California, they refer to the noise conundrum as finding a needle in the haystack. They develop technologies and algorithms based on signal detection theory, technologies and algorithms that help them separate the noise from valid signals.
Signal detection theory is commonly spoken of in terms of psychology. There’s good reason for this. Every sensory organ in the body of a human being is constantly being inundated by, and responding to, environmental stimuli. Most of these stimuli consists mainly of noise. The legitimate informational signals are what are needed by sensory organs to perform correctly.
The Definition of Noise
The temptation for the non-trained expert is to think of ‘noise’ in terms of heavy traffic or large crowds of people. In the physiological sense, noise is somewhat similar but distinctly different. Noise can be any kind of stimuli that is not a legitimate signal.
Actual, physical noise could be one example. Let’s say you are in the midst of a group of 50 people all engaging in a myriad of conversations. You are trying to have a conversation with two other people occupying a very small space. While your brain is trying to parse the words spoken by your two friends, your ears are also hearing the murmuring of all the conversations going on around you. Those peripheral conversations are noise; the conversation between you and your friends is the legitimate signal.
For purposes of signal detection theory, noise could be any kind of stimulation separate from the desired legitimate signal. It could be auditory, visual, olfactory, etc. The challenge in medical science is to separate noise from legitimate signals in order to figure out what’s going on inside the patient’s body.
Four Possible Outcomes
Signal detection theory considers four possible outcomes when trying to separate legitimate signals from noise. They are as follows:
- Hit – A stimulus is present and responded to.
- Miss – A stimulus is present but not responded to.
- False Alarm – A stimulus is absent, but a false hit is registered.
- Correct Rejection – A stimulus is absent and correctly recognized as such.
As experts in sensor technology, the job of the engineers at Rock West Solutions is to find the needle in the haystack using their own equipment and their ability to interpret the four possible outcomes of signal detection. How do they do it? Through a combination of things.
First and foremost are their state-of-the-art sensors capable of compiling huge data sets. Second are applied statistical and analytical techniques that help technicians begin parsing the data. Next are hardware and software tools for signal processing, followed by engineering design that applies the optimal solution to given stimuli and signals.
Finding that needle in the haystack is, by definition, a challenging task. This is why engineers study signal detection theory. Remember that it is a theory rather than an exact science. The more we learn about it though, the better we will get at separating noise from legitimate signals. Eventually, noise detection technologies will have all sorts of innovative applications – from medical science to scientific instrumentation to environmental control.