Statistical biases correction in channelized Hotelling model observers.

Abstract

Objective.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 thed'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.Approach.CHOd'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-centralFcumulative distribution (F'), which is the CHO underlying statistical behaviour for the resubstitution method, was computed, and compared tod'values and CI bounds. A set of experimental data was used to evaluateF'median values.Main results.TheF'median allows to get accurate corrected simulatedd'values down to zero-signals. For smalld'values, the variation ofd'values with the inverse of number of images is not linear while theF'median allows a good correction in such conditions. TheF'median is also inherently symmetric with regards to the CI. With experimental data,F'median values in a range of about 1-10d'values were within -0.8% to 4.7% of linearly extrapolated values at an infinite number of images.Significance.TheF'median correction is an effective simultaneous correction of the zero-signal statistical bias and finite-sample statistical bias, and of CI asymmetry of CHO.