Validation of a digital image processing software package for the in vivo measurement of wear in cemented Charnley total hip arthroplasties

Computer-generated images were used to assess image processing software employed in the radiographic evaluation of penetration in total hip replacement. The images were corrupted using Laplacian noise and smoothed to simulate different modulation transfer functions in a range associated with hospital digital radiographic systems. With no corruption, the penetration depth measurements were both precise and accurate. However, as the noise increased so did the inaccuracy and imprecision to levels that may make changes in the penetration observed clinically difficult to discern between follow-up assessments. Simulated rotation of the wire marker produced significant bias in the measured penetration depth. The use of these simulated radiographs allows the evaluation of the software used to process the digital images alone rather than the whole measurement system.