Improved Detection of Visual Field Progression Using a Spatiotemporal Boundary Detection Method

Glaucoma is the leading cause of irreversible blindness worldwide and requires regular monitoring upon diagnosis to ascertain whether the disease is stable or progressing. However, making this determination remains a difficult clinical task. Recently, a novel spatiotemporal boundary detection predictor of glaucomatous visual field (VF) progression (STBound) was developed. In this work, we explore the ability of STBound to differentiate progressing and non-progressing glaucoma patients in comparison to existing methods. STBound, Spatial PROGgression, and traditional trend-based progression methods (global index (GI) regression, mean regression slope, point-wise linear regression, permutation of pointwise linear regression) were applied to longitudinal VF data from 191 eyes of 91 glaucoma patients. The ability of each method to identify progression was compared using Akaike information criterion (AIC), full/partial area under the receiver operating characteristic curve (AUC/pAUC), sensitivity, and specificity. STBound offered improved diagnostic ability (AIC: 197.77 vs. 204.11–217.55; AUC: 0.74 vs. 0.63–0.70) and showed no correlation (r: −0.01–0.11; p-values: 0.11–0.93) with the competing methods. STBound combined with GI (the top performing competitor) provided improved performance over all individual metrics and compared to all metrics combined with GI (all p-values < 0.05). STBound may be a valuable diagnostic tool and can be used in conjunction with existing methods.