Education Icon
Department of Radiology, Brigham and Women's Hospital
Harvard Medical School

IV. Image Noise

Image noise is the degree of variation of pixel values caused by the statistical nature of radioactive decay and detection processes. Even if we acquire an image of a uniform (flat) source on an ideal gamma camera with perfect uniformity and efficiency, the number of counts detected in all pixels of the image will not be the same.

The counting noise in nuclear medicine is Poisson noise, so that the pixel noise variance is equal to the mean number of counts expected in a given region of the image. Standard deviation is the square-root of the variance.

diagram

If a lesion is present, the signal-to-noise ratio (SNR) is a measure of how far (in units of standard deviations of noise) the lesion protrudes above (or below) the average background level. A lesion SNR of about 2 would be marginally detectable. A SNR of about 3 would be fairly easily detectable.

One would always like to obtain as many counts as possible to reduce the percentage of image noise:

% image noise = ( sigma / N ) x 100% = 100/sqrt(N)

The limitations (as always) are scan time and patient dose. Therefore, ideal imaging systems are as efficient as possible.

Next Page

Return to beginning page: Physical Characteristics of Nuclear Medicine Images

Lesson Author: Stephen C Moore, PhD, scmoore@bwh.harvard.edu
See About BrighamRAD for information about this Web site.

Is this a mirrored page?
The official URL is http://brighamrad.harvard.edu/education/online/physics/MooreNM/ImageNoise.html

Contact the BrighamRAD design team at radweb@dsg.harvard.edu
Updated June 1, 1998