When we first used optical coherence tomography (OCT) in the early 90s, images were pixelated and even the retina borders
sometimes difficult to distinguish. Technology was time domain, enabling approximately 600 single A-Scans per second. Generating
retina maps at this speed required considerable interpolations. Nevertheless, this technology allowed novel insights into
the retina and thus started to be implemented into clinical routine.
By about 2003, spectral domain OCTs started to improve scan speed to 20000–50000 A-scans per second. This enabled denser sampling
in the X–Y plane and features such as 3D presentation of data cubes. Especially when combined with eye-tracking, this spectral
domain technology allowed for clearly improved repeatability and follow-up examinations. Therefore, this technology has become
standard for clinical evaluation and follow-up today.
Some shortcomings, however, remain: While scan speed is high, it is not high enough to obtain dense OCT cubes of the posterior
pole without lengthy examinations and/or eye-tracking. Moreover, there is still the need to predefine scan protocols for each
individual patient. Many physicians have just accepted dividing the posterior pole imaging into a 'retina' (i.e., the macula)
and 'glaucoma' (i.e., optic disc) world, although there is growing evidence that neurons, for example, are not only affected
at the optic disc. Thus, predefined scanning areas appear to be technically mandated rather than based on disease and anatomy.
Another relevant point appears to be fundus image quality: When obtained by simple infrared camera imaging, image quality
is usually not very good, while adding modalities like a scanning laser ophthalmoscope (SLO) or photo adds to system complexity
and system costs.
There are several reasons, why in the future ophthalmic OCT technology may move from Spectral Domain OCT to systems based
on wavelength swept lasers (swept source OCT: SS-OCT). One argument is the dramatically higher imaging speed that can be achieved
with SS-OCT, especially using so called FDML lasers.
Fourier Domain Locking (FDML) of lasers for swept source was first published in 2006 by Dr R. Huber at J.G. Fujimoto's lab
and he continued to work on this new type of laser. One of the major advantages of this technology is that scan speed in the
megahertz range is feasible today.
Leading a laser and imaging group at the faculty of physics (Institute for Biomolecular Optics, BMO) of Ludwig-Maximilians-University
in Munich, Dr Huber developed several OCT applications based on this swept source FDML approach. In collaboration with the
department of ophthalmology, we jointly defined a system capable of imaging up to 60° × 60° by OCT without pupil dilation.
Given the current scan speed of ~1.68 MHz, this system is >40 times faster than usual spectral domain OCT systems today.
This enables us to image the complete posterior pole — covering both the macula and optic disc — in one 3D dataset and with
dense isotropic sampling in the X–Y plane in ~0.8 seconds. Theoretical X–Y sampling density is ~15.7 μm at this resolution
and speed, which means that the resulting 3D dataset has one complete OCT A-Scan each 15.7 microns on the fundus (Figure 1).
Figure 1: ‘En-face’ fundus imaging and one OCT scan out of 3D data cube.