Abstract
Background
Radiological and clinical results of total shoulder arthroplasty are dependent upon ability to accurately measure and correct glenoid version. There are a variety of imaging modalities and computer-assisted reconstruction programmes that are employed with varying degrees of success. We have compared three freely available modalities: unformatted 2D CT; formatted 2D CT; and 3D CT reconstructions.
Methods
A retrospective analysis of 20 shoulder CT scans was performed. Glenoid version was measured at the estimated mid-point of the glenoid from unformatted 2D CT scans (Scapula body method) and again following formatting of 2D CT scans in the plane of the scapula (Friedman method). 3D scapula reconstructions were also performed by downloading CT DICOM images to OSIRIX 6 and plotting ROI points on Friedman's axis to most accurately define glenoid version. Both measurements taken from 2D CT were compared to those from 3D CT. Eleven CT scans were of male patients, 9 female. Mean age was 55.2 years (Range: 23–77 years). Fourteen scans were performed for trauma, 6 for arthroplasty. Twelve scans were of the left shoulder.
Results
Mean glenoid version as measured on: unformatted 2D CT was −4.51 degrees (−29.67 – 7.22 degrees); formatted 2D CT was −2.04 degrees (−36.96 – 9.72 degrees); and on 3D reconstructions was −3.01 degrees (−32.57 – 14.33 degrees). Sixty percent of measurements taken on formatted 2D CT were within 3 degrees of those taken on 3D reconstructions, with 85% within 5 degrees. This proportion fell to 30% and 50% respectively on unformatted 2D CT.
Discussion
In this small study measurements of glenoid version taken on formatted 2D CT demonstrated greater accuracy than unformatted 2D CT when comparing to 3D reconstruction measurements as the gold standard. Although we demonstrated no significant statistical difference between measurements in this pilot study we believe significance will be obtained as we increase our sample size.
