advertisement
Lin et al. (173) evaluated and compared the longitudinal variability of time domain optical coherence tomography (OCT) and confocal scanning laser ophthalmoscopy (CSLO) in a group of 75 subjects (50 with glaucoma and 25 normal subjects). In addition, this study examined the factors related to longitudinal variability with each technique in multivariable regression models. Their findings indicate that overall longitudinal variability fell within a similar range for CSLO and OCT with intraclass correlation coefficients (ICC) from 0.86 to 0.95 for OCT and from 0.89 to 0.96 for CSLO. However, the ICC for rim area was significantly better for CSLO. In addition, reference plane height for the CSLO and disc area for the OCT where responsible for much of the variability in rim area. This paper adds important comparative information to the growing body of longitudinal data comparing the effectiveness of optic nerve analysis methods. Longitudinal variability is one of the most important factors in any measurement technique.
For the OCT the primary source of variability was the measurement of disc areaGiven similar correlations with pathologic changes, variability will determine the efficiency of a technique to detect progression. This paper illustrates the relative variability of both techniques and show that both have a relatively low variability and thus may be useful for longitudinal progression. In addition, and perhaps most importantly, the study points out the principle components of variability with each technique and hints at way to improve variability with these techniques. For the OCT the primary source of variability was the measurement of disc area. This is not surprising giving the frequency of disc boundary detection artifacts, the influence of beta zone atrophy, and the lack of image registration. Thus methods to improve disc area boundary detection would be useful.
The reference plane height was the critical component of variability for parameter measurements for the CSLOThis may be accomplished through better software algorithms and improved instrumentation with the higher resolution and image registration available on the newer spectral domain instruments. It would be interesting to see if manually corrected OCT scans had less variability. Similarly, reference plane height was the critical component of variability for parameter measurements for the CSLO.