Early experience with automated B-mode quality assurance tests.

Although quality assurance programmes have been recommended for many years, there is limited evidence of their efficacy. This study aimed to assess whether an automated image quality analysis method could demonstrate changes in scanner performance in a quality assurance programme. Test object images were analysed, measuring lateral resolution, low contrast penetration, slice thickness, anechoic target visibility and grey-scale target contrast and visibility. Known and suspected scanner faults were investigated and routine results were reviewed. At least one variable changed in response to each known or suspected scanner fault. Resolution and grey-scale target visibility changed due to image shadowing. Slice thickness, lateral resolution and grey-scale target contrast were affected where users reported deterioration in image quality. A single probe fell out of tolerance on routine testing, due to an unrecorded change to the default preset by the supplier's representative. Interpretation of individual results is not always intuitive, observed changes depending on the shape of the grey-scale transfer curve and on target and background echo levels. Our results have provided evidence for the efficacy of this method of performance testing. Further experience is required to evaluate this method for prospective detection of faults and further work is required to determine optimum scanner settings and test object properties to maximise fault detection and to reduce the dependence of results on confounding factors.