What determines contrast resolution in SPECT and PET?

Contrast resolution in SPECT and PET is determined primarily by count statistics, noise, scatter, attenuation effects, and reconstruction parameters. It reflects the ability to detect differences in tracer uptake between adjacent regions.

Because radionuclide imaging is a photon-counting process governed by Poisson statistics, contrast resolution improves as the number of detected counts increases. However, image processing choices (particularly filtering and iterative reconstruction settings) also strongly influence the balance between contrast and noise.

Contrast resolution in SPECT and PET is governed primarily by count statistics and noise, with scatter, attenuation, and reconstruction settings significantly influencing lesion detectability.

Even with adequate spatial resolution, poor contrast resolution can obscure clinically significant lesions.

Understanding the physics

Contrast in radionuclide imaging arises from differences in radiotracer concentration. However, the visibility of this contrast depends on noise.

Noise in nuclear medicine images is statistical in nature. The standard deviation of counts in a region is proportional to the square root of the number of detected events. Therefore, the signal-to-noise ratio increases with the square root of total counts.

The most important determinant is signal-to-noise ratio (SNR). Increasing administered activity, acquisition time, or detector sensitivity improves count statistics and therefore contrast resolution. However, these improvements are gradual because of the square root relationship.

Scatter reduces contrast resolution by introducing mispositioned counts into the image. Compton-scattered photons contribute background signal that reduces contrast between lesions and surrounding tissue. Scatter correction methods improve contrast by removing this contamination.

Attenuation also affects contrast. If attenuation is not corrected, deeper structures may appear falsely reduced in activity, altering apparent contrast relationships.

Reconstruction parameters further influence contrast resolution. Increasing the number of iterations in iterative reconstruction enhances contrast but increases noise. Smoothing filters reduce noise but can decrease lesion contrast and obscure small structures.

Contrast resolution reflects the interplay between count statistics, physical correction modelling, and reconstruction choices.

Where this matters clinically

A lesion may be missed not because of inadequate spatial resolution, but because insufficient contrast resolution makes it indistinguishable from background activity. Optimising acquisition time, activity, and reconstruction parameters is therefore critical for improving lesion detectability.