Fully automated software to monitor wear in prosthetic knees using fluoroscopic images

Abstract.

Total knee arthroplasty is now a widely accepted treatment for late-stage arthritis. Wear of the polyethylene layer in prosthetic knees is a known cause of implant failure. Early detection of wear may allow prediction of device failure. In this paper we describe a fully automated image processing algorithm to measure the minimum tibiofemoral joint space width (mJSW) for monitoring prosthesis wear radiographically. The femoral portion and tibial plate were automatically delineated and mJSW was calculated in each compartment. The software also delineated the tip of the prosthesis pin in order to make a magnification correction. The algorithm was tested with a set of triplicate acquisitions of 18 fluoroscopic knee images. The RMS standard deviation (RMSSD) for the triplicate measurements was calculated as a figure of merit. The RMSSD was 0.077 and 0.087 mm for the lateral and medial compartments. The computer successfully found the minimum JSW for both compartments in all 54 images. A single case (2% of total) required user interaction to correct for an obvious failure to delineate the prosthesis pin. We document a robust and precise tool for quantifying mJSW to monitor prosthesis wear.