Why is noise proportional to √N in nuclear medicine imaging?
In nuclear medicine imaging, noise is proportional to the square root of the number of detected counts (√N) because radioactive decay follows Poisson statistics. For a Poisson process, the variance equals the mean number of events. Since standard deviation is the square root of variance, statistical noise becomes √N.
In nuclear medicine, statistical noise equals the square root of the number of detected counts, meaning image quality improves only gradually as counts increase.
This means that if a detector records N counts, the uncertainty (or noise) associated with that measurement is √N. As the number of counts increases, absolute noise increases slightly, but relative noise decreases.
The key implication is that improving image quality requires increasing the number of detected photons, but, improvements occur gradually, not linearly.
Understanding the physics
Radioactive decay is a random process. Even if the true average count rate is constant, the number of photons detected over a given time interval fluctuates. These fluctuations follow the Poisson distribution.
For a Poisson distribution:
Standard deviation = √N
Standard deviation represents the statistical spread around the mean count value. In imaging, this spread appears visually as noise.
While the absolute noise increases with higher counts (since √N increases as N increases), the relative noise decreases:
This relationship explains why increasing counts improves image quality. However, the improvement follows a square-root relationship. For example:
To halve the relative noise, you must increase counts by four.
Doubling the counts only improves relative noise by a factor of √2.
This diminishing return is fundamental to radionuclide imaging and explains why simply increasing activity or acquisition time does not proportionally improve image quality.
Where this matters clinically
Understanding the √ relationship helps explain why low-dose or short-acquisition nuclear medicine studies appear noisy. It also informs protocol optimisation: improving image quality significantly often requires substantial increases in detected counts, which may affect radiation dose or scanning time.