Best-Fit Circle Missing Area Method Shows Good Accuracy and Interrater Reliability When Assessing Glenoid Bone Loss

Abstract

Background:

The amount of glenoid bone loss (GBL) in anterior shoulder instability helps determine management options. Unfortunately, there is no consensus on how to accurately measure GBL.

Hypothesis:

The best-fit circle missing area method has better accuracy and interrater reliability than the ratio and diameter methods.

Study Design:

Controlled laboratory study.

Methods:

Three-dimensional computed tomography shoulder scans were collected and segmented on 4 healthy male participants (mean age, 22.5 ± 3.4 years). For each scapula, 5 GBL models were created with known levels of GBL (5%-30%) in the anteroinferior glenoid rim. Ground-truth GBL was determined using the glenoid fossa edge and best-fit circle of the anteroinferior edge. Six blinded reviewers with varying expertise measured GBL for 20 randomized models using 3 different best-fit circle methods (missing area, ratio, and diameter). Accuracy was assessed by root mean square error and obtained by comparing measured and ground truth for bone loss for each model. Interrater reliability was assessed using intraclass correlation coefficients (ICCs) with a 2-way random-effects model with consistency.

Results:

On average, the missing area method overestimated GBL by 1%, while the ratio method and diameter method overestimated GBL by 2.4% and 6.3%, respectively. The interrater reliabilities of the missing area (ICC, 0.89), ratio (ICC, 0.91), and diameter (ICC, 0.90) methods were good, excellent, and good, respectively. For all reviewers, the overall root mean square error of the missing area method (3.6%) was better than that of the ratio (5%) and diameter (7.9%) methods. When stratifying reviewer accuracy by level of training, the missing area method remained superior at all levels when compared with the ratio and diameter methods (attending [3.1% vs 3.8% and 8.9%], fellow [4.2% vs 6.3% and 8.4%], and resident [3.5% vs 4.6% and 6.3%]).

Conclusion:

The best-fit circle missing area method was reliable and more accurate at measuring GBL than the other methods. This held true at all levels of surgical experience (resident, fellow, and attending).

Clinical Relevance:

The best-fit circle missing area method is accurate, has good reliability, and is simple. This can enhance preoperative planning for shoulder surgeons.