Abstract
Abstract: :
Purpose: To evaluate a new statistical technique for detecting topographic changes in series of optic nerve head images acquired by scanning laser tomography [Heidelberg Retinal Tomograph (HRT)] by comparison to ‘Topographic Change Analysis’ (TCA) superpixel method and linear regression of Rim Area (RA) over time. Methods: Proven techniques, collectively referred to as Statistic Image Mapping (SIM), are widely used to quantify images of the brain from MRI and PET scanning instrumentation. These techniques are applied to HRT images: a pixel by pixel analysis of topographic height over time yields a statistic image, which is processed to reduce noise and account for multiple testing across the whole image. This is achieved by using permutation testing which also accounts for within–series variability obviating the need for normative data. The new technique was evaluated on 20 normal control subjects and 30 ocular hypertensive patients, selected retrospectively, each with a longitudinal series of 7 images (controls: range 2.8 – 7.1 years; ocular hypertensive range 2.8 – 7.3 years). Each of the 30 patients had normal visual fields at baseline and had developed reproducible visual field loss (AGIS) over the course of the follow–up. TCA progression was defined as a cluster of 20 or more progressing (depressed) superpixels inside the contour line; RA progression was defined as a statistically significant negative slope (P < 0.05.) Results: SIM identified 21 (70%) of the patients as having structural progression, whereas the TCA identified 16 (53%) and RA identified 15 (50%) over the course of the follow up. SIM flagged 2 (10%) false positives in the normal controls whereas TCA identified 3 (15%) and RA identified 2 (10%). Conclusions: In this group of 50 patients and normal subjects SIM had better power in detecting structural change in series of HRT images compared to current quantitative techniques (without flagging more false positives). The clinical utility of these techniques will be established on further longitudinal data sets.
Keywords: imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • imaging/image analysis: clinical • optic disc