|
Journal of Bone and Joint Surgery - British Volume, Vol 90-B, Issue 7,
839-846.
doi: 10.1302/0301-620X.90B7.20305 Copyright © 2008 by British Editorial Society of Bone and Joint Surgery The determination of linear and angular penetration of the femoral head into the acetabular component as an assessment of wear in total hip replacementA COMPARISON OF FOUR COMPUTER-ASSISTED METHODSC. H. Geerdink, MD, Orthopaedic Surgeon1; B. Grimm, PhD, Research Manager2; W. Vencken, Research Assistant2; I. C. Heyligers, MD, PhD, Orthopaedic Surgeon2; and A. J. Tonino, MD, PhD, Orthopaedic Surgeon2
1 Department of Orthopaedic Surgery, Ikazia Hospital Rotterdam, Postbus 5009, 3008 AA, Rotterdam, The Netherlands. Correspondence should be sent to Dr C. H. Geerdink; e-mail: carelgeerdink{at}hotmail.com
We have compared four computer-assisted methods to measure penetration of the femoral head into the acetabular component in total hip replacement. These were the Martell Hip Analysis suite 7.14, Rogan HyperOrtho, Rogan View Pro-X and Roman v1.70. The images used for the investigation comprised 24 anteroposterior digital radiographs and 24 conventional acetate radiographs which were scanned to provide digital images. These radiographs were acquired from 24 patients with an uncemented total hip replacement with a follow-up of approximately eight years (mean 8.1; 6.3 to 9.1). Each image was measured twice by two blinded observers. The mean annual rates of penetration of the femoral head measured in the eight-year single image analysis were: Martell, 0.24 (SD 0.19); HyperOrtho, 0.12 (SD 0.08); View Pro-X, 0.12 (SD 0.06); Roman, 0.12 (SD 0.07). In paired analysis of the six-month and eight-year radiographs: Martell, 0.35 (SD 0.22); HyperOrtho, 0.15 (SD 0.13); View Pro-X, 0.11 (SD 0.06); Roman, 0.11 (SD 0.07). The intra- and inter-observer variability for the paired analysis was best for View Pro-X and Roman software, with intraclass correlations of 0.97, 0.87 and 0.96, 0.87, respectively, and worst for HyperOrtho and Martell, with intraclass correlations of 0.46, 0.13 and 0.33, 0.39, respectively. The Roman method proved the most precise and the most easy to use in clinical practice and the software is available free of charge. The Martell method showed the lowest precision, indicating a problem with its edge detection algorithm on digital images.
|
|


