Accuracy and precision study in vitro for peripheral quantitative computed tomography.

Abstract

We evaluated the accuracy and precision of a peripheral quantitative computed tomography (pQCT) scanner, the Stratec XCT-960, using 12 human cadaveric forearms. The accuracy was determined by comparing the total bone mineral content (BMC) with the ash weight (AW). We scanned and ashed three consecutive slices (thickness 2.5 mm) at the standard position (s-position) and at 2.5 mm both proximal and distal to the s-position. The correlation coefficient between the AW and total BMC using slices at the s-position was r = 0.87 with an accuracy error (random component) of 15.5%. The correlation coefficient using all slices was r = 0.90 with an accuracy error of 14.3%. The correlation coefficient improved to r = 0.95 with an accuracy error of 9.7% after averaging the results of all three slices for each forearm. The short-term precision error expressed as the coefficient of variation (CV) of bone mineral density (BMD) and BMC was determined by measuring the forearms five times either with repositioning or without repositioning. The CVs with repositioning were 2.77 and 1.15 for total BMD and BMC, 1.85 for trabecular BMD; without repositioning they were 0.29, 0.58 and 0.69 respectively. To further evaluate the influence of positioning, additional scans were performed at 1, 2 and 5 mm proximal, and 1 and 2 mm distal to the s-position. BMD and BMC were greatly influenced by the scan location; for example, the percentage differences in trabecular BMD 1 mm distal and proximal relative to the s-position were 2.5% +/- 5.1% and 0.18% +/- 6.3%, respectively. The Stratec XCT-960 appears to be a moderately accurate and highly precise scanner with potential usefulness for evaluating BMC and BMD of ultradistal radius.