We repeated the validation experiments and sensitivity analysis reported above with added image noise. Although the absolute values of the model-based measures were shifted upwards, as would be expected, there were no changes in the relative values, nor any systematic or statistically significant changes in sensitivity, even for 10% added noise.