Evaluation of radiolucency condition in total hip arthroplasty: a statistical comparison of the diagnostic capability of digitised image vs. conventional X-ray film.

Abstract

This work analyses the diagnostic capability of radiographic images taken from patients with total hip arthroplasty and visualised on monitor. Images were obtained with digital acquisition of conventional X-ray films. The investigated pathology is the absence of direct contact between bone and prosthesis (radiolucency). Three senior orthopaedists defined the diagnostic "truth" on well-defined regions of interest on 22 conventional X-ray films of total hip arthroplasty, obtaining a total of 110 reference ratings. Films were digitised by use of an X-ray scanner. Four readers evaluated the X-ray images, applying conventional and monitor visualisation. To show any difference between ratings on film and ratings on monitor a sensitivity, specificity and accuracy study jointly with a receiver operating characteristics (ROC) study were performed for each reader and for all combined readings. The intra-observer reproducibility of the radiographic protocol was equal to 87% and the inter-observer one was in the range 85-92%. The sensitivity, specificity and accuracy study together with the ROC analysis did not show significant differences between the two evaluation modes. The evaluation of radiolucency from digitised X-ray films visualised on a monitor resulted statistically comparable with the conventional evaluation on X-ray films.