Lesion detection in simulated elastographic and echographic images: a psychophysical study

The image quality of two ultrasonic imaging techniques was studied: conventional echography and the recently introduced elastography. The image quality was assessed by estimating the detectability of disc-shaped lesions of various sizes and contrast levels. The study was designed to verify the hypothesis that elastograms could show lesions at a higher subjective and objective level of detectability than echograms of the same object contrast. This hypothesis was adopted because homogeneous elastograms can present a higher point signal-to-noise ratio than uniform echograms. Both elastograms and echograms were generated by two-dimensional (2D) simulations. The subjective assessment was performed by psychophysical experiments using the staircase up-down method. The threshold contrast of detection for both modalities was determined at different diameters of the disc-shaped lesion. These values were used to construct the contrast-detail curves for both techniques. For identical object contrasts, elastography was found to have significantly higher detectability at all lesion diameters considered. The contrast thresholds were also used for an objective evaluation with the lesion signal-to-noise ratio. The objective measure evaluated at the subjective threshold of detection for both modalities was not found to be identical, nor constant over the range of lesion diameters as expected.