Statistical biases correction in channelized Hotelling model observers

<i>Objective.</i>Channelized Hotelling model observers are efficient at simulating the human observer visual performance in medical imaging detection tasks. However, channelized Hotelling observers (CHO) are subject to statistical biases from zero-signal and finite-sample effects. The point estimate of the<i>d'</i>value is also not always symmetric with exact confidence interval (CI) bounds determined for the infinitely trained CHO. A method for correcting these statistical biases and CI asymmetry is studied.<i>Approach.</i>CHO<i>d'</i>values and CI bounds with hold-out and resubstitution methods were computed for a range of 200 × 200 pixels images from 20 to 10 000 images for 10, 40 and 96 channels from a set of 20 000 images with gaussian coloured simulated noise and simulated signal. The median of the non-central<i>F</i>cumulative distribution (<i>F'</i>), which is the CHO underlying statistical behaviour for the resubstitution method, was computed, and compared to<i>d'</i>values and CI bounds. A set of experimental data was used to evaluate<i>F'</i>median values.<i>Main results.</i>The<i>F'</i>median allows to get accurate corrected simulated<i>d'</i>values down to zero-signals. For small<i>d'</i>values, the variation of<i>d'</i>values with the inverse of number of images is not linear while the<i>F'</i>median allows a good correction in such conditions. The<i>F'</i>median is also inherently symmetric with regards to the CI. With experimental data,<i>F'</i>median values in a range of about 1-10<i>d'</i>values were within -0.8% to 4.7% of linearly extrapolated values at an infinite number of images.<i>Significance.</i>The<i>F'</i>median correction is an effective simultaneous correction of the zero-signal statistical bias and finite-sample statistical bias, and of CI asymmetry of CHO.