Radiolucent lines in a cementless hip stem: The use of semi-automated measurements to monitor progression over a 10 year period

Radiolucent lines (RLLs) observed adjacent to a cementless stem following Total Hip Arthroplasty (THA) have been linked to stem loosening. Their manual assessment is susceptible to observer error. In this study a semi-automated (Auto) method was developed to characterise RLLs from Anterior-Posterior (AP) and Lateral (Lat) radiographs, evaluate their time progression and effects on clinical outcomes. The study comprised of 237 subjects with postoperative, 1 and 10 Y radiographs. Of these 131 also had 4–7 Y radiographs. All subjects also had accompanying implant, demographic and clinical outcomes data. An algorithm was written to evaluate the incidence of RLLs and post-processing was performed to characterise RLLs as ‘not present’ or ‘benign’, or significant (‘SigRLL’). The criteria for SigRLL were tuned to minimise the number of false negatives compared to a manual characterisation. The incidence of SigRLL using AP/Lat or AP only was compared and the latter was used to monitor their development over time. The relationship with clinical outcomes was evaluated. Incidence of SigRLLs using the Auto and Manual AP/Lat methods agreed in 199/237 (84.0%) of subjects. The discrepancies included seven false negatives. The Auto AP/Lat and AP methods agreed in 221/237 (93.2%) of subjects. Twenty-eight subjects with a 4–7 Y review had SigRLLs at 10 Y. Of these, 23 had SigRLLs at 4–7 Y. Greater pain was reported in the SigRLLs group (p = 0.003). Four out of five stems subsequently revised for loosening exhibited 10 Y SigRLLs.