Purchase this article with an account.
Zhihong Hu, Muneeswar Nittala, Srinivas Sadda; Comparison and normalization of retinal reflectivity profiles between spectral-domain optical coherence tomography devices. Invest. Ophthalmol. Vis. Sci. 2013;54(15):5492. doi: https://doi.org/.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To compare the retinal layer reflectivity profiles from two different spectral-domain optical coherence tomography (SD-OCT) devices and to further define a general solution to “homogenize” the reflectivity across different devices.
A graph-based segmentation approach was used to identify 11 retinal surfaces (from the internal limiting membrane to the choroid-sclera junction) in SD-OCT single line images from Spectralis and Cirrus devices. For each device, two different scan acquisition protocols were used (20x and 40x ART for Spectralis, 20x with and without tracking for Cirrus). B-scans were obtained from 34 eyes of 17 normal subjects. The absolute reflectivity, as well as the reflectivity normalized against the mean retinal pigment epithelium (RPE) and vitreous reflectivity for the 11 segmented layers (spanning from the vitreous, through the retina and choroid), were compared across devices and scan protocols using a paired t-test.
When comparing between devices/protocols, the average reflectivity for the 11 retinal layers prior to normalization was statistically significantly different (p=0.03), but was no longer different following RPE normalization (p=0.37). Similarly, the liner regression of the average reflectivity demonstrated an average R2 of 0.94 before the normalization, but improved to 0.98 after the normalization. Plotting the reflectivity profiles (Figure 1) better illustrated the similarity between devices and the further improvement with normalization.
Retinal layer reflectivity profiles between different SD-OCT devices and acquisition protocols showed similarities which could be further "homogenized" using normalization strategies. Normalization of reflectivity profiles between devices could potentially facilitate the development of universal automated segmentation protocols.
This PDF is available to Subscribers Only