advertisement
Editors Selection IGR 17-3
Optical Coherence Tomography: RNFL measurement time domain and spectral domain OCT
Comment by Kouros Nouri-Mahdavi on:
24086 Comparison of retinal nerve fiber layer measurements using time domain and spectral domain optical coherent tomography, Knight OJ; Chang RT; Feuer WJ et al., Ophthalmology, 2009; 116: 1271-1277
Knight and colleagues (1098) recently compared the performance of Stratus OCT (a time domain machine) to that of Cirrus OCT, one of the newer spectral domain OCTs. The SD-OCTs are slowly gaining popularity across the world. With each successive generation of disc/RNFL imaging devices, the question arises whether the measurements from the older machine (here Stratus OCT) are comparable to those of the newer machine (SD-OCT). The authors enrolled glaucoma patients with a wide spectrum of glaucoma severity in addition to a group of normal subjects. This allowed the authors to compare the two machines across a wide range of RNFL thicknesses. Interestingly, the difference between the two machines was directly related to the RNFL thickness. The authors found that SD-OCT measurements were systematically smaller than those of the Stratus OCT for the normal group, glaucoma suspects and eyes with mild to moderate glaucoma; whereas the Cirrus measurements tended to be larger than those of Stratus OCT in eyes with severe glaucoma. This could be related, as the authors also point out, to a different floor effects for the two devices with a lower floor and therefore possibly more accurate measurements near the floor values for Cirrus OCT.
Direct comparison of the measurements from the Stratus OCT to that of Cirrus OCT in the same patient is better avoided
The variability of Cirrus minus Stratus measurements was found to be higher than the intersession variability for Stratus OCT. The authors provided regression formulas to calculate Cirrus data from those of Stratus OCT. However, there is a moderate amount of scatter around the regression lines especially for RNFL thicknesses in the superior and inferior quadrants. As a result, any estimation would entail a large amount of uncertainty. The authors did not report on the distribution of image quality and number of eyes excluded because of inadequate quality. It would have been interesting to know the number of eyes with poor images so as to be able to compare the overall quality of images and ease of obtaining images with the two different versions of OCT. How can we explain the results and what are we to make of them? The trend towards overall thinner RNFL measurements with each successive generation of OCT makes one wonder if, with increasing resolution of the machine, the segmentation algorithms are doing a better job of delineating only the RNFL as opposed to including parts of the ganglion cell layer. The distribution of RNFL thickness in postmortem human eyes has been reported to be lower than current OCT measurements in the normative database. It is interesting that results of the newer OCT machines are approaching those obtained by GDx Scanning Laser Ophthalmoscopy. More accurate measurements of RNFL thickness in vivo will probably have to wait for more advanced versions of the OCT. In the mean time, as Knight and colleagues have suggested, direct comparison of the measurements from the two devices in the same patient is better avoided.
Comments
The comment section on the IGR website is restricted to WGA#One members only. Please log-in through your WGA#One account to continue.Log-in through WGA#One
