Sunteți pe pagina 1din 1

Purpose

We have published a model that maps the visual field to 1 optic nerve head
(ONH) sectors according to individual anatomy (Denniss et al, IOVS 2012).
Clinical measurements of input anatomical parameters are subject to error that
causes variability in mapping. We aimed to quantify this variability in order to
determine minimum feasible ONH sector sizes for practical use.
Methods
Variability in fovea-ONH distance and angle measurements made from retinal
nerve fibre layer OCT was estimated from 10 repeat scans of 10 healthy
participants (Figure 1).
Errors in estimating axial length from refractive error were determined from the
95% prediction interval of a linear fit to published population data (Atchison et
al IOVS 2004, Mallen et al OPO 2005) (Figure 2).
Structure-function maps (24-2 pattern) were generated using a variety of
simulated plausible anatomical parameters under two conditions:
1. axial length measured accurately (error only in ONH position)
2. axial length estimated from refractive error (error in both)
Maps (n=200) were generated for each unique parameter set (n=210), each
with added error sampled from the distributions above. Variability due to
measurement error in mapped 1 ONH sectors was calculated.
#966
Support: Australian Research Council, Heidelberg Engineering (F), Haag-Streit (F), Victorian Life Sciences Computation Initiative jonathan.denniss@nottingham.ac.uk
Conclusions
For general clinical use, our data suggest that 30 ONH sectors capture the likely effects of measurement error in
anatomical parameters. For research purposes, minimum ONH sector size varies across the visual field from 3 to 18.
Figure 1: Variability in repeated fovea-ONH angle (a) and distance (b) estimates from
OCT images (10 participants,10 repeats).
Individualised Structure-Function Mapping: Influence of Variability in Clinical
Measurements of Anatomy on Mapping Resolution
Jonathan Denniss
1,2
, Allison M. McKendrick
1
, Andrew Turpin
2
1
Optometry & Vision Sciences,
2
Computing & Information Systems, The University of Melbourne, Australia.
Figure 3 (a,b) The 1 ONH sectors mapped from each 24-2 location when axial length is (a) measured accurately or (b) estimated from
refractive error. (c,d) The 90% range widths of normalised mapped sectors across all anatomical parameters when axial length is (c)
known or (d) estimated from refractive error. Based on (d), we suggest a simple scheme of 30 ONH sectors for clinical use.
Figure 2: Relationship
between axial length and
refractive error in two
published studies.
Dashed lines show 95%
prediction interval of
linear fit (solid line).
Results
Figure 4: Two glaucoma cases showing possible use of individualised structure-function mapping. In each case: (a) high-density 3x3
visual field plot showing ONH position, (b) individualised structure-function map with 30 ONH sectors, (c) retinal nerve fibre layer
thickness from OCT with overlaid corresponding visual field sensitivities (colour-coded), (d) structure-function plot using overlapping 30
ONH sectors centred on the mapped 1 sector, (e) structure-function plot using fixed adjoining 30 ONH sectors as in (b). In panels (d,e)
points colour-coded for visual field sensitivity are surrounded by rings colour-coded for corresponding retinal nerve fibre layer thickness.
AcknowledgemenL: Shonra[ 8allae Caneshrao collecLed Lhe daLa shown ln llgure 4.
normalised SD = 0.97 normalised SD = 61#m

S-ar putea să vă placă și