Anatomical plane and transverse axis of the scapula: Reliability of manual positioning of the anatomical landmarks

Background The aim of our study was to evaluate the accuracy of manual determination of the three key points defining the anatomical plane of the scapula, which conditions the reliability of planning software programs based on manual method. Method We included 82 scapula computed tomography scans (56 pathologic and 26 normal glenoid), excluding truncation and major three-dimensional artifact. Four observers independently picked the three key points for each case. Inter- and intra-observer agreement was calculated for each point, using the intraclass correlation method. The mean error (mm) between the observers was calculated as the diameter of the smallest sphere including the four chosen positions. Results Lower inter-observer agreement was found for the trigonum superoinferior position and for the glenoid center anteroposterior position. The mean positioning error between the four observers was 6.9 mm for the trigonum point, and error greater than 10 mm was recorded in 25% of the cases. The mean positioning error was 3.5 mm for the glenoid center in altered glenoid, compared to 1.8 mm for normal glenoid. Discussion Manual determination of an anatomical plane of the scapula suffers from inaccuracy especially due to the variability in trigonum picking, and in a lesser extent, to the variability of glenoid center picking in altered glenoid.