IODA-shoulder®, a 3-dimensional automated software for glenoid bone loss quantification in shoulder instability. A validation study

Glenoid bone loss (GBL) calculation in patients with shoulder instability has a wide variability in methods and their reliability. The purpose of this study was to describe and validate a new semi-automated software developed to improve GBL calculation using a 3D imaging modeling (IODA-shoulder) and to compare the method to the PICO area method. A semi-automated software to assess GBL was preliminarily developed and validated on 7 fresh frozen specimens (scapulae with artificially created glenoid defect), using water displacement method. Afterwards, the software was retrospectively used on CT images of 20 patients affected by recurrent shoulder dislocation. Inclusion criteria were: unilateral dislocation, minimum 2 dislocation episodes. Exclusion criteria were bone reconstruction of the glenoid, failed shoulder stabilization, bilateral dislocation, shoulder arthritis. Three-dimensional computed tomography images of bilateral shoulder were retrieved for each patient. Two methods to determine GBL were compared. The PICO surface area method and the new IODA method. We assessed the intra- and inter-rater reliability of the two methods with the Intraclass Correlation Coefficient (ICC), the Bland-Alman analysis, and Lin's concordance correlation coefficient (CCC). We did not find a statistically significant difference between the mean volumes calculated with PICO and IODA methods, respectively 914 vs. 815 mm3, p=0.155. The analysis carried out by using the traditional PICO method showed a lower concordance rate among four observers than the higher concordance found using IODA method, regardless of the size and the location of the defect. The ICC agreement with PICO was significantly lower than with IODA (0.76 vs. 0.97). We found a poor CCC with PICO (from 0.65 to 0.81) and a substantial one with IODA (from 0.96 to 0.98). The intra- and inter-rater reliability using IODA method is significantly better than PICO method. The assessment of GBL using IODA method is time saving, avoiding significant inter- and intra-observer variation, mainly due to individual skill and experience in the method. IV, experimental study.