it's explanation about principle of optical coherence tomography. it's based on 2 papers related with basic OCT theory.
OCT Summary2014.4.4 Chap 1,41. Introduction OCTAn optical imaging tool which has been used to cross-sectional images of the internal microstructure in biological tissues in vivo. 2High (High resolution-Micrometer resolution(1-15m) & High speed(high scanning speed-It makes possible to observe live cell) Long wavelength light(low coherence)2-D & 3-D image Disadvantage -> CT MRI . 3. General interference law for partially coherent light I(k) (k-domain=2/:wave# photo current) IFFT -> z(depth information)-domain we get interference signal, its -domain. when it is directly converted from wavelength to wave#(k-domain), the interference signal has nonlinear sampling. It makes axial resolution broadening. So we have to need resampling procedure. (1. spline interpolation method 2. a zero-filling interpolation method) * Interpolation method &
TD-OCTFD-OCT2. OCT system main parameter Resolution Penetration depthSensitivity (SNR=power)Image acquisition rate(frame rate)2-(1) Resolution (& Penetration depth) Axial resolution is defined by(1) half of the coherence length(2) center wave length and FWHM(full-width at half-maximum=-3dB bandwidth) of the optical source ->Power/2 bandwidth
Lateral resolution is related to spot size focused on in the sample ( spot size center wavelength, beam size, focal length and depth of field(2*Rayleigh length ) ->spot size short focal length large numerical aperture(NA) . The small spot size the depth of focus sample image contrast . -> so, OCT sample (penetration depth ) low NA .
The Depth range(Zmax) is limited by the resolution of the spectrometer (or sampling resolution in SS-OCT) -> according to the Nyquist sampling theorem the sampling frequency has to be twice as large as the highest frequency in the spectrum for example, interference signal band width > 2*spectrometer resolution(band width) analog signal digitalize nyquist frequency
52-(2) Optical Source An optical source determines performance such as (1) penetration depth (2) resolution (3) image qualityOCT -> the near infrared(NIR) region(700nm~1500nm),called the optical window source . Because it has lower the absorption coefficient of water and blood than the other region.-> Therefore biological tissue information . And the high optical power of the source penetration depth and SNR of the OCT .
OCT image resolution source center wavelength FWHM . Axial resolution shorter center wavelength and broader bandwidth source . 2-(2) Side lobe When the optical source has the shape of non-Gaussian, Ideal gaussian shape Fourier transform gaussian shape. real side lobe Optical source sidelobe mainlobe coherence length axial resolution . (axial resloution=coherence length/2) far ghost image image quality . Ghost image? Mirror image, IFFT . Optical source Ghost image side lobe Main beam DC line , 2 ghost image . Source gaussian shape -> power Full-range technique ( )- . IFFT . image depth .
2-(3) Signal-to-Noise Ratio, Noise source Shot noise : from current fluctuation (: background light and the dark current in the detector)->SD TD noise TD . Thermal noise : from random thermal particle motion 1/f noise : a characteristic noise in active devices, arises in the low frequency region Relative intensity noise(RIN,) : fluctuation in optical power from the source and from the mechanical movement of the optical mounts (real fluctuation .->optical mount power )
* noise .
2-(3) SNR in TD SNR=signal/noise= the signal power from a perfectly reflecting mirror/the weakest sample reflectivity(signal power noise level ) SNR 1/NEB NEB : the noise equivalent bandwidth of the signal bandpass = the electrical bandwidth(f) for detection of the interferometric signal f = 2*Vs/0 (Vs : reference mirror speed, : optical source bandwidth , 0 : the center wavelength of the optical source ) -> so! NEB= f = 2*Vs/0 , NEBVs-> Vs& NEB SNR 2-(3) SNR in FD SD-OCT system -> SNR (Shot noise ?) SD-OCT SNR depends on M(the # of illuminated detector= pixel# of the CCD camera) Ex)1024pixel CCD camera TD OCT 24dB - SNRsd the exposure timeSS-OCT -> SD-OCT SNR exposure time A-line scan rate fA The optical power and the SNR of OCT penetration depth . -> optical power can make images with deeper penetration depth in both TD&FD. The high optical power increases the SNR -> ! In FD the SNR decreases with depth due to the finite resolution of the spectrometer. -> so! High resolution spectrometer 3. PS-OCT
4. Advanced Frequency Domain OCT SD-OCT 800nm wavelength region (1300nm InGaAs line scan camera acquisition )->Recently we have used 1000nm wavelength source. It has higher image quality than 800nm when we observe under retina area. 1300nm wavelength SS-OCT(using a frequency-swept laser,16kHz and 20kHz) . (1) To obtain high-speed imaging and (2) To enable dual-balanced detection to eliminate the DC autocorrelation noise over SD-OCT(signal vertical, horizontal .)Ultra-high-speed swept source(>200kHz)using the Fourier domain mode locking(FDML) method ( ring laser ) -> mode locking , 4-(1) Experimental setup of the SD-OCT SLD(superluminescent diodes/compact&inexpensive) : 0=1310nm, =FWHM=86nm, power=7mW -> z=(2ln2*02)/(3.14* )=270nm
4-(1)-a Diffraction grating
4-(1)-bCCD camera 12.5MHz, 12.5Mclocks/1secFor 1024pixel 1line(A-scan) data minimum 266clocks . 46990lines/sec (line rate) (B-scan) 1image 512 lines -> 1image/10.9ms, 46990/512=91/1sec(frame rate)
x-galvo , trigger T current linear galvo mirrorcurrent linear B-scan Focus spot .
y-galvo 3D image scan
4-(1)-c wavelength calibration We tilted the reference mirror and obtained the spectrum with an optical spectrum analyzer(OSA)->reference mirror reference mirror the intensity profile .
4-(1)-c wavelength calibration
Grating equation- Simulation results 2.Regression equation
(=Wave length, n=pixel number) 3d order polynomial fitted results4-(1)-c I(k) (k-domain=2/:wave# photo current) IFFT -> z(depth information)-domain we get interference signal, its -domain. when it is directly converted from wavelength to wave#, the interference signal has nonlinear sampling. It makes axial resolution broadening. So we have to need resampling procedure. (1. spline interpolation method 2. a zero-filling interpolation method) * Interpolation method & 4-(1)-d System performances The full wavelength range of our spectrometer = 150nm pixel 0.145nm . 150/1024 -> spectral resolutionThe depth range = 2.9mm The sensitivity of SD-OCT = 108.7dB
4-(1)-dAxial resolution (14.05m/8.8m in air y)-> this difference can be attributed to the errors in the linear rescaling procedure from -domain to k-domain(the high axial pixel size)Frame rate 91fps/15fps (frame rate) the real frame rate calculation time for linear interpolation and fourier transformation was needed. Solution : The multi-thread programming method ( , ,
4-(2) Swept-Source OCTAdvantageThe rise time of a photodiode is faster than the CCD chips. ->improved Data acquisition speed Interference signal grating detector. Simpler system OCT image mirror image . ? ~? A dual balanced detector ( photo diode 2 . . real Signal , noise .)-> 800nm range optical source . Water absorption eye imaging .
Disadvantage Narrow wavelength scanning range FWHM resolution . -> semiconductor optical amplifier(SOA)4-(2)-a External-Line Cavity Swept Source at 850nmOur swept source OCT has the optical wavelength selection filter with Littman configuration Disadvantage of Original Littman configuration -> the output beam is the zeroth-order reflection from the grating at grazing incidence because the diffraction grating works as output coupling in addition to wavelength selection-> so! OCT system alignment too sensitive.
Our system -> 1x2 coupler(30:70). -> PC(polarization controller) Gaussian profile . Polarization sensitivity of the Intracavity componet fiber wavelength-dependent loss
4-(2)-b The optical spectra of the swept source outputSS in Ref(ring-cavity)Our SS(line-cavity)Center wavelength815-870nm850nmFWHN0.085nm0.068nmTuning range55nm68nmSweeping speed43.2kHz(polygon scanner)200Hz(galvometer)
Table. Specification our swept source compared with other groups swept sourceFig (a) 0.063nm with a fixed scanning mirror spectral width (b) center wave length 849nm(c) Optical source (d) t swept source wavelength4-(2)-c Images using the swept-source OCT1000samples/an A-scan Reference arm ND(neutral density/-20dB) filter -> to reduce source intensity noise (RIN?)-> SNR + the sample attenuation constant(40dB) -> ? 1000 point sample-length -> 1024 poing sample-length by zero-padding ->rescaled to the linear frequency domain by using a simple calibration method ->IFFT-> A-scan 512data point
1.03mm depth position/z=5.5m TerminologyCornea Iris Lens Optical nerve blind spot Retina Pupil Chroid Sclera
Eye OCT images, anterior & posteriorOptical coherence tomography : principles and applicationA.F. Fercher2014.4.7~
1. IntroductionTwo fundamental optical tomography technique. (1) Diffuse optical tomography (DOT)(2) Optical diffraction tomography (ODT)-> OCT is physically founded on ODT
* ODT uses single scattered light and derives tomography images by the Fourier diffraction projection theorem.
Ballistic photon : photon may travel without scattering Near-ballistic photon : scattering ballistic path 281.1. Basic schemesLCI is based on the occurrence of fringes if the optical path lengths of reference and sample beams co