Imaging from a patient with idiopathic vitreomacular traction. A, Spectral-domain optical coherence tomography of the macula. Slice-based scrolling through consecutive raster scans reveals multiple cystic retinal changes. However, the relationship of the cysts to one another and the correlation of the cystic volume to the area of traction are not easily ascertained from these 2-dimensional slices. B, A conventional false-color map of retinal thickness (with red representing thicker retina; green, thinner retina) superimposed on an infrared image of the macula. Although the thickened retina is apparent in the area of traction, the individual retinal structures (such as the retinal cysts or the posterior hyaloid) cannot be isolated, manipulated, inspected, or quantified. C, A single 2-dimensional optical coherence tomography slice (the same slice shown in A) superimposed on a 3-dimensional volumetric reconstruction of the posterior hyaloid (dark green), area of traction between the hyaloid and retinal surface (light green), intraretinal cystic spaces (blue), and subretinal fluid (orange). Specific retinal structures can be isolated, examined from any angle, and quantified. Previously unseen tubular connections between cysts can also be appreciated in the 3-dimensional reconstruction.
Three-dimensional volumetric reconstruction of the retina in a patient with vitreomacular traction. A, Posterior hyaloid (green), internal limiting membrane (yellow), and cystic retinal spaces (blue) that result from the traction. Arrows indicate the area of attachment between the vitreous and the retina. B, Isolated cystic retinal spaces from the same 3-dimensional volumetric reconstruction. Inspection of the 3-dimensional image from multiple perspectives reveals a previously hidden tubelike connection (arrow) between the smaller and larger cysts.
Aaker GD, Gracia L, Myung JS, Borcherding V, Banfelder JR, D’Amico DJ, Kiss S. Three-Dimensional Reconstruction and Analysis of Vitreomacular Traction: Quantification of Cyst Volume and Vitreoretinal Interface Area. Arch Ophthalmol. 2011;129(6):805-820. doi:10.1001/archophthalmol.2011.123
Optical coherence tomography (OCT) has made considerable advancements in retinal imaging, especially with the advent of high-resolution, spectral-domain OCT.1 Nonetheless, viewing and analysis of OCT data are limited to 2-dimensional (2D) slice-based scrolling through consecutive scans (Figure 1).2
In a series of eyes with idiopathic vitreomacular traction, we used a method of rendering 2D raster OCT data into 3-dimensional (3D) volumetric objects. By isolating and quantifying distinct retinal structures within these 3D objects, we sought to determine the following: (1) the correlation between cyst volume and area of vitreoretinal adhesion; and (2) the relationship between individual cysts within the retina.
In this Weill Cornell Medical College Institutional Review Board–approved study, OCT scans (Heidelberg Spectralis HRA + OCT; Heidelberg Engineering, Inc, Carlsbad, California) of idiopathic vitreomacular traction were imported into Avizo visualization software (VSG, Burlington, Massachusetts) using a custom plug-in written in C++. This plug-in extracts all B-scans, the infrared image, and automated segmentations provided by Heidelberg Eye Explorer software version 188.8.131.52 (Heidelberg Engineering, Inc), all registered to the same coordinate system. Using manually set, threshold-based computational algorithms, boundaries of the cystic spaces, internal limiting membrane, and posterior hyaloid face were segmented in an automated fashion. From these segmentations, 3D objects were created and compared for accuracy with the original 2D OCT scans.
Surface areas of attachment and intraretinal cystic volumes were calculated and converted to millimeter equivalents using the scale embedded within the raw data. The correlation between the surface area of attachment and the volume of cystic spaces was calculated.
To assess the relationships between retinal structures, 3D reconstructions were inspected in the fully immersive Computer Assisted Virtual-Reality Environment (Christie Digital Systems USA, Inc, Cypress, California).
Seven eyes of 7 patients with incomplete perifoveal vitreous detachment and cystoid foveal thickening were included. The 2D OCT data were rendered as 3D volumetric objects with defined cysts, internal limiting membrane, and posterior hyaloid surfaces (Figure 1 and Figure 2).
The total volume of the cysts ranged from 0.00145 mm3 to 0.647 mm3 (mean [SD], 0.115 [0.218] mm3). The surface area of traction between the posterior hyaloid and internal limiting membrane ranged from 0.0114 mm2 to 1.0226 mm2 (mean [SD], 0.314 [0.316] mm2). There was a strong positive correlation between the tractional surface area and cystic space volume (r = 0.890; P < .01).
Inspection of 3D reconstructions from different viewpoints revealed that several cysts, which appeared separate on conventional OCT, were connected in 3D space (Figure 2 and video.
In vitreomacular traction syndrome, volumetric reconstruction of 2D OCT slices permits isolation and analysis of 3D retinal structures. Quantification of cyst volume and vitreoretinal adhesion area revealed a strong positive correlation. As pharmacological vitreolysis emerges, quantification of the adhesion area may influence the decision between observation, intravitreous injection, or surgery.3
Previous attempts to evaluate vitreomacular traction with OCT were limited to examining individual OCT slices and measuring overall macular volume and retinal thickness with false-color maps, and they did not isolate or individually quantify retinal structures (Figure 1).4 Attempts to calculate drusen volume had similar limitations.5 In these reports, OCT slices were individually segmented rather than reconstructed as 3D objects.5 The technique described here can be used to quantify drusen volume, geographic atrophy area, and cystic changes in age-related macular degeneration.
Analysis of 3D volumetric reconstructions of OCT images may improve our understanding of pathophysiological features of various retinal diseases. Here, 3D reconstructions revealed that some intraretinal cysts were connected via small tubelike channels not obvious on conventional OCT (Figure 2). Furthermore, quantification of 3D structures may provide meaningful parameters in other disorders such as cystoid macular edema, age-related macular degeneration, diabetic macular edema, and retinal vascular occlusion. However, long-term prospective data from larger patient cohorts are necessary to establish accurate predictive models.
Correspondence: Dr Kiss, Department of Ophthalmology, Weill Cornell Medical College, 1305 York Ave, 11th Floor, New York, NY 10021 (email@example.com).
Author Contributions: Mr Aaker and Dr Gracia contributed equally to this work and are considered co–first authors.
Financial Disclosure: Weill Cornell Medical College, Drs Gracia and Kiss, and Mr Banfelder have intellectual property rights to some of the material presented in this article.
Funding/Support: This work was supported by Research to Prevent Blindness.
Role of the Sponsor: The sponsor had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; and in the preparation, review, or approval of the manuscript.
Previous Presentation: This paper was presented at the 27th Meeting of the Club Jules Gonin; November 4, 2010; Kyoto, Japan.