Automated performance assessment of ultrasound systems using a dynamic phantom.

Abstract

Quality assurance of medical ultrasound imaging systems is limited by repeatability, difficulty in quantifying results, and the time involved. A particularly interesting approach is demonstrated in the Edinburgh pipe phantom which, with an accompanying mathematical transformation, produces a single figure of merit for image quality from individual measurements of resolution over a range of depths. However, the Edinburgh pipe phantom still requires time-consuming manual scanning, mitigating against its routine use. This paper presents a means to overcome this limitation with a new device, termed the Dundee dynamic phantom, allowing rapid set-up and automated operation. The Dundee dynamic phantom is based on imaging two filamentary targets, positioned by computer control at different depths in a tank of 9.4% ethanol-water solution. The images are analysed in real time to assess if the targets are resolved, with individual measurements at different depths again used to calculate a single figure of merit, in this case for lateral resolution only. Test results are presented for a total of 18 scanners in clinical use for different applications. As a qualitative indication of viability, the figure of merit produced by the Dundee dynamic phantom is shown to differentiate between scanners operating at different frequencies and between a relatively new, higher quality system and an older, lower quality system.