Enhanced 3D retinal vascular network reconstruction from high‐definition SD‐OCT

Abstract

AbstractPurpose To compute the 3D vascular network of the human retina from high‐definition spectral‐domain optical coherence tomography (OCT).Methods A method able to compute a fundus reference from SD‐OCT scans of the human macula to an extent similar to what can be achieved with color fundus photography was recently proposed by our research group. We resort to this new method and to an automatic supervised‐learning classification algorithm (both submitted for publication) to identify OCT A‐scans that cross the vascular network. A multi‐scale wavelet analysis is performed on each of these A‐scans to identify the transition between the hyper‐reflectivity due to the vessel and the shadow cast due to light absorption by hemoglobin. This transition is set as the vessel depth location, and the respective diameter estimated as the width from the computed fundus reference.Results Eyes of healthy controls (N = 10) and eyes (ETDRS levels 10 to 35) from type 2 diabetic patients (N = 20) were imaged with the Cirrus HD‐OCT (Carl Zeiss Meditec, Dublin, CA, USA) and photographs were obtained with the Topcon TRC 50DX (Topcon Medical Systems, Inc., Oakland, NJ, USA) or the Zeiss FF 450. The vascular tree from the OCT reference presents an extension similar to that of the color fundus photography, 99.7(10.68)% (average(SD)). The 3D reconstruction of the vascular network allows to clearly distinguish between crossovers and bifurcations and the vascular network distribution is in agreement with the known anatomy of the human retina.Conclusion The herewith presented method allows the visualization of the vascular network in 3D and compares favourably to the previous method both in terms of quality and extension of the vascular tree.