Precision evaluation of 2 CT-based radiostereometric analysis systems in a cadaver study

Abstract

Background and purpose: To address the limitations of radiostereometric imaging and to eliminate the need for intraoperative marker placement, CT-based radiostereometric analysis (CT-RSA) software systems have been developed. We aimed to evaluate the precision of a novel CT-RSA software system, V3MA, against an established CT-RSA software system, CTMA, while also examining the impact of CT scanner model on precision.
Methods: 7 CT scans per scanner (Siemens SOMATOM Force and GE Revolution) of a porcine cadaver with a knee implant were used for pairwise comparisons. By aligning paired CT scans, the translation (mm), rotation (°), and maximum total point motion (MTPM, mm) of the tibial implant with respect to the bone were computed. V3MA aligned the scans using the voxel gray values of the bone and implant, whereas CTMA aligned the surface points of the bone and implant. The precision of both software systems and the effect of both scanner models were investigated using paired data in a linear mixed model.
Results: Both software systems resulted in a similar MTPM (contrast V3MA–CTMA –0.002 mm, 95% confidence interval [CI] –0.015 to 0.011; V3MA: 0.09 mm; CTMA: 0.10 mm), indicating comparable precision using a minimal important difference of 0.10 mm. Using Siemens scanner data resulted in a higher estimated MTPM than using GE scanner data (contrast Siemens–GE 0.046 mm, CI 0.024–0.067; Siemens: 0.12 mm; GE: 0.072 mm).
Conclusion: The precision of the new CT-RSA software system, V3MA, is comparable to that of CTMA under zero-motion assumptions. Minor, clinically irrelevant, inter-scanner differences in CT-RSA precision exist for both software systems.