Veiling Glare in High-Dynamic-Range Displays: Experimental Comparison of Contributions from the Human Visual System and the Display Device using a Double Random Staircase Technique

A major source of image quality degradation in imaging systems is the presence of veiling glare (VG) at each step of the imaging chain. VG is the reduction of contrast due to multiple scattering in the acquisition, display, or visual perception of an image. High-dynamic-range (HDR) displays have intrinsic veiling glare as light scatters in the device multiple layers. There has been significant research to understand and reduce VG influences in the acquisition steps for x-ray and photographic sensors. However, not much is understood at the display and human visual perception level. Our study poses the question: is VG a limiting factor for detection tasks when high contrast and high resolution medical images are displayed using high-dynamic-range devices? To answer this question we used a binary decision task on the presence of a Gaussian target on white noise image backgrounds with known location (indicated with dark hairline markers) in a full-resolution, dual-layer HDR display prototype. The VG source was realized with a ring with varying parameters added to the image. Detection thresholds were estimated using a double random staircase technique with one-image-at-a-time paradigm. In addition, divergence in the reader's fixation from the target location in the center of the image field was recorded in real-time and used to provide auditory feedback to minimize significant changes in adaptation level. Our results suggest that VG in the HDR display does not significantly affect the detection thresholds. The dominant effect is from the VG in the human visual system. The reader's thresholds estimates were lower as we decreased the distance and increased the illuminance of the VG source. Results from three readers were used to formulate a multivariate model of the VG effect for amplitude thresholds. The model can be used to suggest appropriate presentation modes for medical images in instances where VG is a significant factor. The goal was to understand veiling glare effects caused by HDR displays in the human visual system in simple ring and target images before moving onto the VG limitations in detection task in CT images.