Appropriate quality assurance (QA) and quality control (QC) procedures have been well developed and validated for dual X-ray absorptiometry (DXA), and are widely applied in multicenter clinical trials to monitor device stability used to check the treatment effects on bone mineral density. This is not yet the case for quantitative ultrasound (QUS) technology, for which no QC approaches have yet been fully tested. The first Achilles (GE-Lunar Corporation, Madison, WI, USA) has been on the market for 10 years (1991). The goal of this study was to develop the QC methodology for the QUS Achilles+ device using its past/current experience (log and maintenance files.) as well as by integrating the progress made over the last years in the ultrasound domain so as to better understand the influence of temperature on ultrasound parameters. Because of the lack of confidence in the external black rubber phantom used in daily QC with the Achilles+, to monitor the device stability, we selected several QC parameters known to be influenced by potential malfunctions as experienced by the maintenance department of GE-Lunar company as well as the physical approach. These are phantom temperature-adjusted speed of sound (PSOS-TC) and broadband ultrasound attenuation (PBUA-TC), water speed of sound error (WSE), water spectrum slope (WSS) and water gain (WG). We used four Achilles+ devices perfectly stable during their entire QC range, to calculate the optimum thresholds (based mostly on 95% confidence interval) for each of these parameters as well as the precision for the in vitro SOS and BUA. An additional not fully stable Achilles device has been used to run a QC procedure example. The precision expressed as the CV was 0.22% and 0.65% for the PSOS-TC and PBUA-TC, respectively. The alarm thresholds used for QC process are +/- 0.6%, +/- 1.9%, +/- 6.8 m/s, +/- 5.3% and +/- 7.3% for the PSOS-TC, PBUA-TC, WSE, WSS and WG, respectively. Applying a logical approach on the impact of each parameter on each other as well as their respective reactivity to malfunctions, we build a QC process flowchart meant to detect real malfunction in the daily QC. We found that in case of real malfunctions, the in vivo SOS should be decreased by 1.33 m/s for each 1 m/s increase in WSE. Unfortunately, in vivo BUA can not be adjusted when real malfunction occurs. Nevertheless, the BUA can be qualified as bad quality data and excluded from the medical interpretation. Using the currently available phantom and parameters, the best possible QC procedures to detect long-term drift in the daily QC of the Achilles+ was developed. To fully validate our approach and gain confidence in the defined limits it is our plan to apply this QC processing to a higher number of QUS devices.
Read full abstract