Purpose: This study was undertaken to assess and improve the repeatabi
lity and reproducibility of nerve fiber layer (NFL) measurements using
the Nerve Fiber Analyzer (NFA I), a scanning laser polarimeter. Metho
ds: The repeatability of two subsequent measurements was determined in
100 healthy eyes. Four methods of image evaluation were performed. Po
larimetric measurements of retinal NFL were obtained along a peripapil
lary band at the superior and inferior poles of the optic disk and ana
lyzed in a one-pixel band of absolute retardation values (method 1) an
d relative to the nasal segment (method 2). Method 3 used measurements
obtained in a 10-pixel band along the same peripapillary band. Method
4 comprised measurements in selected areas outside the peripapillary
band, which did not contain blood vessels. The repeatability (r) and r
elative reliability coefficient (Rel) were assessed as indices of agre
ement of repeated measurements. In addition, the reproducibility (R) f
rom day to day for methods 3 and 4 was assessed by taking 10 images on
10 consecutive days from each of 10 subjects. Results: The 95% critic
al value for r in the superior and inferior segments improved from abo
ut 30% for methods 1 and 2, to 23% for method 3, and to 11% for method
4. For the superior and inferior segments, the fourth method of image
evaluation yielded a Rel of 98%, as compared with 54% in method 1, 80
% in method 2, and 87% in method 3. The 95% critical value for R was 2
3% for method 3 but only 15% for method 4. Conclusions: The improved r
eproducibility of measurements has made the NFA I a promising device.
An improved reproducibility (smaller limit for detecting change) does
not necessarily indicate good sensitivity and specificity for any inst
rument. The role of the scanning laser polarimeter in clinical practic
e requires further study.