Fourier domain mode locking: new lasers for optical coherence tomography
Optical coherence tomography (OCT) generates micron-scale resolution, cross-sectional images of tissue by measuring the echo time delay of backscattered light (see Figure 1).1 Recent work showed that rapidly wavelength-swept narrowband laser sources can be used for this technique.2 This approach, called swept source OCT (ss-OCT), is similar to the frequency modulated radar used in police speed guns, but uses light instead of radio waves. In the technique, one depth scan is generated for each sweep of the laser. The system performance depends on the laser source, so optimizing it could dramatically improve ss-OCT.
Tunable or wavelength-swept lasers have long been used in sensing and metrology. Standard ones consist of a gain medium with a broad amplification spectrum and a tunable optical bandpass-filter in the cavity for active wavelength selection (see Figure 2, left). However, ss-OCT requires performance that standard systems cannot provide. Typical ss-OCT applications seek a wavelength sweep range of more than 100nm around 1300nm, an instantaneous linewidth of 0.1nm or less, and sweep repetition rates of 100kHz or more. Tuning speeds in standard setups are limited by the time required to build up lasing from amplified spontaneous emission (ASE) background at each new spectral position while the laser is tuned (Figure 2, right).
To overcome the tuning speed limitations of standard lasers, we developed Fourier domain mode locking (FDML).3 In FDML, the cavity length is increased up to several kilometers, enabling the synchronization of the optical round trip time with the tuning rate of the intra-cavity bandpass filter. Light of a certain wavelength passes through the filter and propagates through the long chamber. Meanwhile, the filter sweeps over the entire range. At the exact time when light of this wavelength arrives back at the filter, it is tuned to this wavelength-position again, and then transmitted. Thus, the entire sweep field is optically stored within the cavity and lasing does not have to build up from ASE repetitively. Each sweep is seeded by the previous one (see Figure 3, left).
FDML can be considered as a third stationary operating regime of lasers. The ideal exhibits a constant output power level, like a continuous wave laser. It also has a broad, comb-like output spectrum, like a conventionally mode locked type (see Figure 3, right). In spite of the 7km long delay fiber, FDML is robust and can be built with a small footprint in a 19” enclosure (see Figure 4).
Due to their stationary operation, FDML lasers exhibit dramatically improved coherence and noise properties, compared to other rapidly swept lasers. Furthermore, they have no fundamental limitation in sweep rate. We achieved high quality OCT imaging at record speeds of up to 370,000 lines per second. We demonstrated the superior performance of FDML based OCT systems in numerous time-critical applications. Figure 5 shows the imaging capability of our system. In the future, we will explore novel imaging, ranging, and sensing applications using FDML lasers. We will further improve their output performance and provide a comprehensive theoretical description of their operation.
We would like to acknowledge support from Wolfgang Zinth at the LMU Munich. This research is sponsored by the Emmy Noether program of the German Research Foundation (DFG) HU 1006/2-1 and the European Union project FUN OCT (FP7 HEALTH, contract no. 201880).
Robert Huber leads an independent junior research group in the physics department at the LMU. He received his PhD in 2002 for work on ultrafast electron transfer processes at dye-semiconductor surfaces in wet solar cells. From 2003 to 2006 he worked at the Massachusetts Institute of Technology as a postdoctoral researcher in J. G. Fujimoto's group, where he studied wavelength swept laser sources for optical coherence tomography.
