Signal-to-noise ratio

Signal (yellow line color) + Noise (red random line color)

About

Signal-to-noise ratio (abbreviated SNR) is a measure used medical imaging that compares the level of a desired anatomical signal to the level of background noise. A disturbance in background noise of a radio-graphic image can be caused by a quantum mottle.

Conventional radiography

In radiography, the larger number of photons absorbed, equals greater the SNR, that results in less noisy images. The use of high kVp and use of intensifying screens can reduce the number of photons and radiation dose to patients, therefore reducing SNR. Conversely, measures to increase SNR, such as by increasing the mAs, also increases the patient dose.

Calculation

To calculate SNR...

draw x5 ROIs and record the mean and standard deviation of the x5 pixel values

1. draw x5 ROIs and record the mean (M) and standard deviation (SD) of the pixel value.

2. Calculate: $M/ SD$ (divide)

3. Result is SNR

For example; image we have drawn x5 (2x2) ROIs on a flat field image. We lookup the pixel values (i.e. PVIs) on the CR reader and we see the each value listed below:

ROI_1 = $424$
ROI_2 = $565$
ROI_3 = $601$
ROI_4 = $415$
ROI_5 = $555$

Let's begin the SNR calculation

Calculate mean (M):

1. rearrange values from least to greatest. For example:

$415 + 424 + 555 + 565 + 601$

2. calculate sum = $2560$

3. divide $2560 / 5$ [5 is the count]

4. average [mean] = $512$

Next, calculate Variance:

1. calculate variance For example:

$(415 - 512)^2 + (424 - 512)^2 + (555 - 512)^2 + (565 - 512)^2 + (601 - 512)^2$

$29732 / 5$ (divide)

2. variance= $5946.2$

Next, calculate standard deviation (SD):

1. calculate the Square root of the variance. For example:

√$5946.2$

2. standard deviation = $77.113$

Finally, calculate SNR:

1. Divide M/ SD For example: $512$ / $77.113$

2. SNR = $6.640$ [three decimal places]]

Links

• X-Ray Physics: Signal-to-Noise in Fluoroscopy