Methods and algorithms for optical coherence tomography-based angiography: a review and comparison

Abstract

Optical coherence tomography (OCT)-based angiography is increasingly becoming a clinically useful and important imaging technique due to its ability to provide volumetric microvascular networks innervating tissue beds in vivo without a need for exogenous contrast agent. Numerous OCT angiography algorithms have recently been proposed for the purpose of contrasting microvascular networks. A general literature review is provided on the recent progress of OCT angiography methods and algorithms. The basic physics and mathematics behind each method together with its contrast mechanism are described. Potential directions for future technical development of OCT based angiography is then briefly discussed. Finally, by the use of clinical data captured from normal and pathological subjects, the imaging performance of vascular networks delivered by the most recently reported algorithms is evaluated and compared, including optical microangiography, speckle variance,phase variance, split-spectrum amplitude decorrelation angiography, and correlation mapping. It is found that the method that utilizes complex OCT signal to contrast retinal blood flow delivers the best performance among all the algorithms in terms of image contrast and vessel connectivity. The purpose of this review is to help readers understand and select appropriate OCT angiography algorithm for use in specific applications.

Fig. 1

The $k\text{-}\omega$ space of optical coherence tomography (OCT) data. (a) The Doppler shifts due the movement of mirror in sample arm. (b) The Doppler shifts of the static tissue background and moving particles.

Fig. 2

Configurations of incident sample beam and blood vessel: (a) general case; (b) case where blood vessel is perpendicular to the sample beam.

Fig. 3

The performance comparisons. (a)–(e) The blood vessel network in normal human retina visualized by optical microangiography (OMAG), speckle variance, phase variance, split-spectrum amplitude-decorrelation angiography (SSADA), and correlation mapping. (f) The capillaries selected as in yellow to evaluate the connectivity of the angiogram.

Fig. 4

The performance comparisons. (a)–(e) The blood vessel network in the inner retina layer in normal subject visualized by OMAG, speckle variance, phase variance, SSADA, and correlation mapping.

Fig. 5

The performance comparisons. (a)–(e) The blood vessel network in the outer retina layer in normal subject visualized by OMAG, speckle variance, phase variance, SSADA, and correlation mapping.

Fig. 6

The performance comparisons using the dataset captured from a subject diagnosed with diabetic retinopathy. (a) Fluorescein angiogram where the scanned area is marked with dashed square box, (b) zoomed fluorescein angiography image corresponding to the area for OCT angiography. (c)–(g) The retinal blood vessel network visualized by OMAG, speckle variance, phase variance, SSADA, and correlation mapping.