Low nOISE supercontinuum sources for ultra-high resolution 800nm optical coherence tomography for glaucoma diagnosis

The project LOISE will develop two new low-noise supercontinuum (SC) systems with >400nm bandwidth around 800nm, based on photonic crystal fibers (PCFs) pumped by ytterbium picosecond (ps) and femtosecond (fs) MHz fiber lasers. For ps pumping noise reduction is done by undertapering (factor 2) and increasing the repetition rate by 4 (factor 2). For fs-pumping all-normal dispersion is used to decrease the noise by a factor of 10 and to achieve passive Gaussian spectral shaping. The SC sources are developed to be optimum for ultra-high resolution 800nm optical coherence tomography (OCT) systems for imaging of the eye. To guide fiber design state-of-the-art numerical methods will be implemented using modern massively parallel programming paradigms, such as MPI and OpenCL, to run efficiently on modern and emerging parallel computing many-core hardware, including accelerators such as graphical processing units (GPUs). Both sources will be tested for OCT performance at Univ. Kent.

Project period: May 2010 – Jan. 2014. Partners: DTU Fotonik, NKT Photonics, Université de Franche-Comté (Dudley). Very successful project involving 2 PhDs and a postdoc, during which NKT developed their current draw-tower fiber taper facility to be used in LOISE. This grant was about moving the blue edge to as short wavelengths as possible, which lead to a common patent on asymmetric tapering. Because of the short wavelengths ITRUS had a significant component on degradation, but since LOISE is aimed at the OCT wavelength 800nm the degradation will not be a problem. Since LOISE will not be using the blue edge, but instead try to give the 650-950nm component a Gaussian shape, the ITRUS patent is not directly related. ITRUS discovered that seeding does not reduce SC noise at high power and concluded that seeding is not applicable, so LOISE will not consider seeding. ITRUS ended with a common patent on noise reduction by so-called undertapering to fiber diameters smaller than what is optimum for obtaining the shortest possible blue edge. This patent will be public in January 2015 and will be the foundation of LOISE’s goal of reducing the noise by a factor of two by tapering. So far nothing has been published about this very interesting, but purely experimental finding, which will be one of the fundamental physical studies in LOISE. ITRUS found that existing codes are simply too slow and inefficient to be able to characterize noise in high-power SC sources with several meters of fiber, which is why ITRUS only considered noise at low power and over short lengths, where the patented features were not seen. Consequently no real understanding of why under tapering in some cases improved the noise was obtained, even though a few hypotheses were discussed. This will be the clear task of the picosecond SC part of LOISE.