Effect of radiographic quality on computer-assisted head penetration measurements.

Even the most sophisticated computer-assisted radiographic techniques of measuring femoral head penetration into the polyethylene liner depend on the quality of the radiograph being evaluated, which varies greatly in clinical settings. The authors of this study sought to determine how the accuracy and reproducibility of three commercially available computer-assisted measurement systems differed when measuring optimal radiographs (with sharply defined component edges) and suboptimal radiographs (with less well defined edges). Using three computer-assisted measurement systems, the authors measured head penetration on simulated and clinical hip radiographs. All systems calculated head penetration as the movement of the head center relative to the cup center. To define the periphery of the prosthetic head and cup, one method (System One) used the human eye and a digitizing tablet, whereas the other two methods (System Two and System Three) used digital edge detection algorithms. For simulated hip radiographs, error was calculated as the absolute value of the difference between the known amount of head penetration, determined by a coordinate measuring machine, and the amount of penetration determined by the software. Three way analysis of variance showed a significant difference in absolute error among the three measurement techniques. System One had a significantly smaller absolute error (0.11 +/- 0.06 mm) than did System Two (0.25 +/- 0.25 mm) and System Three (0.19 +/- 0.13 mm). In addition, three-way analysis of variance showed that optimal radiographs were associated with a significantly lower absolute error (0.14 +/- 0.09 mm) than were suboptimal radiographs (0.23 +/- 0.22 mm). For optimal radiographs, there was no significant difference in error among the three measurement methods; all systems were accurate and reproducible. However, for suboptimal radiographs absolute error increased and varied widely, and a significant difference among the methods existed. These data show the susceptibility of head penetration measurements to radiographic technique and underscore the importance of good quality radiographs for all analyses of head penetration.