Purchase this article with an account.
M. Balasubramanian, C. Bowd, G. Vizzeri, P. A. Sample, R. N. Weinreb, N. O'Leary, D. P. Crabb, D. F. Garway-Heath, L. M. Zangwill; Localized Glaucomatous Change Detection Using a New Permutation Test Strategy. Invest. Ophthalmol. Vis. Sci. 2008;49(13):3652.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To evaluate a new superpixel progression event analysis based on re-sampling (SPEAR) to detect localized glaucomatous changes in HRT topographs.
Topographic change analysis (TCA) and statistic image mapping of the retina (SIM) methods are available for detecting localized changes. Unlike TCA, SIM controls for family-wise type-I errors, but requires ≥ 2 follow-ups. SPEAR addresses each of these limitations. Topograph locations were grouped into superpixels (SP) as in TCA. Change significance in each SP between one baseline and each follow-up was estimated from a re-sampling distribution of normalized mean differences. Topograph level changes were detected using a supra-threshold cluster test and the largest cluster size within the optic disk margin (cszSPEAR) and the associated change probability (pSPEAR) were computed. SPEAR parameters were compared with the largest red-pixel cluster size within the optic disk margin in TCA (cszTCA) and evaluated using two independent datasets: 1) from the UCSD Diagnostic Innovations in Glaucoma Study (DIGS) and 2) from the Moorfields Eye Hospital (MF). Of the 240 eyes from 143 DIGS glaucoma patients and suspects with ≥ 4 HRT-II exams, 36 eyes progressed by stereophotos or showed likely progression on visual field guided progression analysis (digsProg). All other eyes were considered to be non-progressing (digsNonProg; n = 204). HRT exams from 20 normal eyes (mfNonProg) and 32 ocular hypertensive eyes that developed visual field changes (mfProg) from the MF were also analyzed.
This PDF is available to Subscribers Only