The effect of nonlinear human visual system components on performance of a channelized Hotelling observer in structured backgrounds

Linear model observers based on statistical decision theory have been used successfully to predict human visual detection of aperiodic signals in a variety of noisy backgrounds. However, some models have included nonlinearities such as a transducer or nonlinear decision rules to handle intrinsic uncertainty. In addition, masking models used to predict human visual detection of signals superimposed on one of two identical backgrounds (masks) usually include a number of nonlinear components in the channels that reflect properties of the firing of cells in the primary visual cortex (V1). The effect of these nonlinearities on the ability of linear model observers to predict human signal detection in real patient structured backgrounds is unknown. We evaluate the effect of including different nonlinear human visual system components into a linear channelized Hotelling observer (CHO) using a signal known exactly but variable (SKEV) task. In particular, we evaluate whether the rank order of two compression algorithms (JPEG versus JPEG 2000) and two compression encoder settings (JPEG 2000 default versus JPEG 2000 optimized) based on model observer signal detection performance in X-ray coronary angiograms is altered by inclusion of nonlinear components. The results show: 1) the simpler linear CHO model observer outperforms CHO model with the nonlinear components; 2) the rank order of model observer performance for the compression algorithms/parameters does not change when the nonlinear components are included. For the present task and images, the results suggest that the addition of the nonlinearities to a channelized Hotelling model may add complexity to the model observers without great impact on rank order evaluation of image processing and/or acquisition algorithms

I. Introduction

Medical images are typically visually scrutinized by physicians searching and classifying morphological features of interest that might predict the state of disease. In this context, medical image quality has been objectively defined in terms of physicians' performance in clinically relevant visual tasks. Thus, evaluating and optimizing medical image quality entails conducting psychophysical studies measuring human performance in these visual detection or classification tasks. Because conducting such psychophysical studies with physicians is time consuming there has been a rationale to develop model observers (also known as numerical observers) that can accurately predict human performance across a range of conditions. Advantages of using these model observers are that they can assist in evaluating, rank ordering, and optimizing image quality in a time efficient manner and also allow for investigation of a range of conditions that is simply not feasible using psychophysical studies. There is now a large literature of successful use of linear model observers to predict human visual detection performance in noise with varying power spectra [1]–[8]. Model observers have also been used to optimize single photon emission computed tomography, lens aperture, and image compression of coronary angiograms [7]–[11].