Perceptual thresholds for differences in CT noise texture.

Purpose: The average ($f_{\mathrm{av}}$) or peak ($f_{\text{peak}}$) noise power spectrum (NPS) frequency is often used as a one-parameter descriptor of the CT noise texture. Our study develops a more complete two-parameter model of the CT NPS and investigates the sensitivity of human observers to changes in it.

Approach: A model of CT NPS was created based on its $f_{\text{peak}}$ and a half-Gaussian fit ($\sigma$) to the downslope. Two-alternative forced-choice staircase studies were used to determine perceptual thresholds for noise texture, defined as parameter differences with a predetermined level of discrimination performance (80% correct). Five imaging scientist observers performed the forced-choice studies for eight directions in the $f_{\text{peak}}/\sigma$-space, for two reference NPSs (corresponding to body and lung kernels). The experiment was repeated with 32 radiologists, each evaluating a single direction in the $f_{\text{peak}}/\sigma$-space. NPS differences were quantified by the noise texture contrast ($C_{\text{texture}}$), the integral of the absolute NPS difference.

Results: The two-parameter NPS model was found to be a good representation of various clinical CT reconstructions. Perception thresholds for $f_{\text{peak}}$ alone are $0.2\mathrm{lp}/\mathrm{cm}$ for body and $0.4\mathrm{lp}/\mathrm{cm}$ for lung NPSs. For $\sigma$, these values are 0.15 and $2\mathrm{lp}/\mathrm{cm}$, respectively. Thresholds change if the other parameter also changes. Different NPSs with the same $f_{\text{peak}}$ or $f_{\mathrm{av}}$ can be discriminated. Nonradiologist observers did not need more $C_{\text{texture}}$ than radiologists.

Conclusions: $f_{\text{peak}}$ or $f_{\mathrm{av}}$ is insufficient to describe noise texture completely. The discrimination of noise texture changes depending on its frequency content. Radiologists do not discriminate noise texture changes better than nonradiologists.