January 15, 1992 / Vol. 17, No. 2 / OPTICS LETTERSHigh-speed optical coherence domain reflectometryE. A. SwansonLincoln Laboratory, Massachusetts Institute of Technology, 244 Wood Street, Lexington, Massachusetts 02173-9108D. Huang, M. R. Hee, and J. G. FujimotoDepartment of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139C. P. Lin and C. A. PuliafitoLaser Research Laboratory, New England Eye Center, Tufts University School of Medicine, Boston, Massachusetts 02111Received September 26, 1991We describe a high-speed optical coherence domain reflectometer. Scan speeds of 40 mm/s are achieved witha dynamic range of >90 dB and a spatial resolution of 17 gm. Two applications are presented: the noninva-sive measurement of anterior eye structure in a rabbitinterelement spacing in a multielement lens.Optical coherence domain reflectometry (OCDR)has emerged as an attractive method for obtaininghigh-spatial-resolution (<10 Am) reflectance mea-surements of fiber-optic, integrated-optic, and bio-logical structures.`8Most previous OCDR systemshave achieved high sensitivity and dynamic rangeby using narrow-band heterodyne detection bypiezoelectric path-length modulation and lock-inamplifier techniques. For many applications,present techniques are unacceptably slow. Highspeed is essential for applications to biological andmedical diagnostics, where measurements must beperformed rapidly compared with movement of aliving subject. Speed is also important for processor assembly line diagnostics, as well as in any appli-cation that requires high data acquisition rates. Byusing a high-speed linear translation stage, OCDRmeasurements can be performed at high speedswith significant simplifications in system design.In this Letter we demonstrate a new OCDR tech-nique for high-speed optical ranging measurements.Speeds approaching 40 mm/s with a dynamic rangeof >90 dB and a resolution of -17 ,4m are achieved.The basic schematic of the OCDR setup is shownin Fig. 1. A broadband light source such as a super-luminescent diode (SLD) is coupled to a fiber-opticMichelson interferometer. One arm of the inter-ferometer leads to the sample of interest, and theother leads to a reference mirror. Fiber-optic andintegrated-optic samples can be directly attached tothe sample arm fiber. For bulk-optical or biologicalsamples, a probe module is used to direct the beamonto the sample and to collect the reflected signal.To aid in sample alignment, we use a fiber-coupledvisible aiming laser. The reflected light beams aredetected at the photodetector. They coherently in-terfere only when the sample and the reference pathlengths are equal to within the source coherencelength. Heterodyne detection is performed by tak-in vivo and the characterization of reflections anding advantage of the direct Doppler frequency shiftthat results from the uniform high-speed scan of thereference path length. Recording the interferencesignal magnitude as a function of the reference mir-ror position profiles the reflectance of the sample.The spatial resolution, scan speed, and dynamicrange of the OCDR system can be designed and opti-mized according to the desired application. Ifwe assume a Gaussian line shape, the round-tripFWHM of the source intensity coherence envelope(AL) is related to the FWHM wavelength bandwidth(AA) of the source byAL = ln(2) -A2(1)where A is the optical wavelength. The Doppler fre-quency shift (fD) that results from a uniform velocityscan (v) and the FWHM power bandwidth of the sig-nal (Af ) are given byfD = 2 (2)AAf = ln(4) 4 v-(3)The optimum bandwidth for the bandpass filter isapproximately 2Af Wider bandwidths will decreasesensitivity, whereas smaller bandwidths will de-crease resolution.Figure 2 shows the measured power spectral den-sity of the light source (Laser Diode Model LDT-3201-SMF) and its FWHM value of 17.4 nm.Figure 3 show the measured interference signal atthe photodetector and the demodulated envelopeobtained by scanning the reference mirror at37.5 mm/s. This speed was used in all the mea-surements reported here. Good agreement betweenthe measured coherence envelope FWHM of 17.2 Atmand the theoretical value of 16.7 ,m [Eq. (1)] was0146-9592/92/020151-03$5.00/0 © 1992 Optical Society of America151152 OPTICS LETTERS / Vol. 17, No. 2 / January 15, 1992REFERENCENn D IAFILER|Fig. 1. Schematic of the high-speed OCDR system.analog-to-digital converter.azwI-isA-D,811 nmA tO nmFig. 2. SLD power spectral density.obtained. The Doppler frequency shift was mea-sured at -93 kHz, also in good agreement withEq. (2).The approximate signal-to-noise ratio (SNR) inthe bandpass filter if we assume shot-noise-limiteddetection is given bysurements of anterior chamber depth are useful indetermining the refractive power of intraocular lensimplants'" as well as in the diagnosis of angle-closure glaucoma." Although ultrasound ranginghas been used extensively in ophthalmology,'2it re-quires direct physical contact of the eye by an ultra-sound transducer or saline immersion of the eye inorder to facilitate the transmission of acoustic wavesinto the eye. In contrast, OCDR is a noncontactdiagnostic.In order to demonstrate the clinical potential ofhigh-speed OCDR, measurements of eye structurewere performed in an anesthetized Dutch pigmentedrabbit in vivo. Figures 5(a) and 5(b) show an OCDRmeasurement of the anterior eye along with a sketchof rabbit eye structure for comparison.' The out-put from the sample fiber was collimated with a20-mm focal-length lens and subsequently focusedinto the anterior eye by using a 150-mm focal-lengthlens. The depth of field was measured to be -3 mmFWHM. The scan speed was 37.5 mm/s, and thescan range was 7.5 mm. Reflection peaks at theair-cornea, cornea-aqueous, and aqueous-lens inter-faces as well as scattering from within the corneacan be seen. An echo of the air-cornea interface isalso visible. It results from the imperfect SLD facetcoating mentioned above. The refractive indices ofthe cornea (n = 1.376) and the aqueous (n = 1.336)(Ref. 13) were used to determine the measuredcornea thickness of 396 4m and the anterior cham-ber depth of 2.40 mm. The optical power incidenton the sample eye was 29.4 MW far below the Ameri-SNR= 1 lqPs R42 hp' NEB(4)where f7 is the detector quantum efficiency, P, is thesample signal power, h is Planck's constant, v is theoptical frequency, R, is the sample reflectivity, andNEB is