Single- and dual-energy quantitative CT adjacent to acetabular prosthetic components: a reliability study

Purpose: Density measurements adjacent to acetabular prosthetic components are challenged by artifacts and the complex anatomy. Three-dimensional ROIs are needed to assess the bone stock. The purpose was to test the intraobserver agreement and reliability of custom segmentation software and to compare BMD measurements in single and dual energy CT (SECT and DECT). Methods and Materials: 10 male patients with uncemented hip prosthetics were scanned and rescanned using 120 kVp SECT and DECT with virtual monochromatic images reconstructed at 130 keV. Hemispherical ROIs were defined slice-by-slice and bone mineral density was calculated using a calibration phantom. Median time between ROI drawings was 6 days. Results: The mean BMD for repeated SECT scans was 430 mg/ccm with a between scan difference of 18 mg/ccm, p=0.001. ICC was 0.98 (95%CI: 0.73 to 1). For DECT the mean BMD was 162 mg/ccm with a difference of 11 mg/ccm, p<0.0001. ICC was 0.94 (95%CI: 0.36 to 0.99). Bland-Altman Limits of Agreement were -51 to 15 in SECT and -30 to 7 in DECT. Repeatability coefficients were 33 and 18 mg/ccm for SECT and DECT respectively. The difference was not statistically significant. Conclusion: The intraobserver reliability was high for both scan modes with statistically significant differences between the repeated measurements in SECT and DECT. The limits of agreement were slightly narrower in DECT, but with no statistically significant difference between repeatability coefficients. The results suggest that the intraobserver agreement of the scan modes is equal. BMD cannot be measured interchangeably with SECT and DECT.