Abstract
Purpose:
To distinguish between normal status and occluded vascular tree from optical coherence tomography (OCT) scans of the human retina.
Methods:
The vascular tree of the human ocular fundus presents two specific signatures in high-definition OCT: the hyper-reflectivity from the top of the vessel and the shadow cast due to the absorption of the light. While the former is present independently of the vascular tree status (perfused or occluded), the shadow cast over the retinal pigment epithelium exists for the normal status but not for the occluded vascular tree. By independently computing the depth-wise line integral over the top layers of the retina, the vascular tree of the human retina can be identified due to the refraction index change from the retinal tissue to the vessel. On the other hand, computing the line integral for the retinal pigment epithelium, the vascular tree presents itself as the dark region due to light absorption on the vessels above where these exist and blood circulation can be found. In this way, perfused vascular tree can be discriminated to occluded vascular tree based on specific signatures found on OCT data.
Results:
By computing the projection over each A-scan from the OCT through specific retinal layers, it was demonstrated the possibility to discriminate between normal and occluded regions of the human ocular fundus. Additionally, when compared to colour fundus photography, an extended vascular tree is computed for the occluded tree segments. The fact that this process does not require the administration of any dye render it feasible for the noninvasive assessment of occlusion.
Conclusions:
The presented method compares favourably to colour fundus photography in discriminating between normal and occluded vascular tree, within the same scan, and provides functional information noninvasively.
Keywords: 549 image processing •
550 imaging/image analysis: clinical •
688 retina