We would like to thank Drs Azad and Takkar for their interest in the study on 3-dimensional enhanced imaging of the posterior vitreous and vitreoretinal interface in eyes with diabetic retinopathy using swept-source optical coherence tomography (SSOCT) and appreciate their insightful comments. The concern raised is a potential overestimation of vitreoschisis in healthy (control) eyes owing to an effect of motion correction.
Motion during OCT acquisition can potentially cause oversampling of some areas and undersampling of others. The motion correction algorithm overcomes this by merging multiple SSOCT volumes into a single volume with improved signal. When assessing dynamic tissue such as the vitreous, its configuration may change among acquisition of multiple SSOCT volumes. This may cause blurring of certain vitreous features when the motion correction algorithm is applied. The authors defined vitreoschisis as 2 distinct linear hyperreflective bands of the posterior hyaloid separated from each other by an optically clear hyporeflective space on motion-corrected SSOCT. While boundaries of these linear hyperreflective bands may become slightly indistinct if blurring occurs following motion correction, in the authors’ observation, it is unlikely that with adequate motion correction the posterior hyaloid would appear falsely divided. If the algorithm fails or performs suboptimally, especially in cases of exaggerated motion, it is usually apparent, since a duplication of all the features of the posterior vitreous and vitreoretinal interface are visible in such cases. The authors ran the algorithm on a different combination of SSOCT volumes if it did not perform adequately for a given combination of SSOCT volumes so that an optimal motion-corrected SSOCT volume with improved signal was available for analysis of each eye. In addition, SSOCT offers imaging advantages such as high sensitivity, long imaging range, and high acquisition speed that allows wide-field 3-dimensional analysis of the vitreous features. Also, vitreous windowing applied to motion-corrected SSOCT further improves vitreous visualization.
Hence, for the above reasons, we believe that prevalence of vitreoschisis in healthy (control) eyes (27%), although higher than that reported using scanning laser–based OCT (7%), is likely a true finding in our study. Further, although vitreous tissue may be thick and adherent to the inner retinal surface in eyes with diabetic retinopathy, in our observation, motion is expected to affect the vitreous features similarly in all eyes regardless of the disease state. Nevertheless, if a slight overestimation of the prevalence of vitreoschisis did occur owing to motion correction, this effect would have been equally distributed across healthy (control) eyes and eyes with diabetic retinopathy in the referenced study.
In conclusion, we performed 3-dimensional analysis of the posterior vitreous and vitreoretinal interface over a wide field in eyes with diabetic retinopathy when compared to healthy (control) eyes using a vitreous windowing method applied to motion-corrected SSOCT volumes. Although an important limitation, blurring of the dynamic vitreous tissue owing to motion correction, in our opinion, is likely an insignificant contributor to the quantification of the prevalence of vitreoschisis in our study.