Abstract
Purpose: :
To compare a new statistical technique for detecting structural progression in scanning laser tomography acquired images of the optic nerve head (ONH) against monitoring change with summary measures of morphological features of the ONH (stereometric parameters).
Methods: :
Statistic Image Mapping (SIM) is a proven method used in neuroimaging to identify areas of activity in images of the brain. SIM generates a ‘change map’ which highlights statistically significant morphological change and has been used to detect progression in series of Heidelberg Retina Tomograph (HRT) images (Patterson et al IOVS 2005 46: 1659–67). The performance of SIM is compared to longitudinal analysis using Rim Area (RA), Rim Volume (RV), Cup Shape Measure (CSM), Height Variation Contour (HVC) and mean Retina Nerve Fiber Layer Thickness (RNFLT) available on the HRT software. The false–positive progression rate was estimated from test–retest data obtained from 74 patients with ocular hypertension (OHT) or glaucoma, using the change criteria for stereometric parameters recommended in the HRT operation manual. The stereometric parameter change criteria were then altered so that the false positive rate matched that of SIM and the true–positive progression detection rates estimated using a longitudinal dataset of 217 OHT patients, 52 of whom progressed by visual fields. SIM and the parameter analysis were compared by survival analysis.
Results: :
The use of stereometric parameters for detecting change has prohibitively high false–positive rates (RA 8.1%, RV 25.7%, CSM 25.7%, HVC 29.7%, RNFL 21.6%) as compared to SIM (2.7%). In the longitudinal data, with false positive rates for SIM and stereometric parameters matched, SIM detected change significantly earlier than the other methods (P<0.001): SIM identified 50% of the patients as progressing after 3.9 years and the stereometric parameters failed to identify 50% of the patients as progressing.
Conclusions: :
SIM detects change earlier and more reliably as compared to monitoring HRT stereometric parameters over time. SIM also provides a clinically useful visual image of the location, depth, extent and rate of structural change, in series of optic nerve head images.
Keywords: imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • imaging/image analysis: clinical • optic disc