WO2005121696A1

WO2005121696A1 - Method for evaluating signals in optical coherence tomography (oct)

Info

Publication number: WO2005121696A1
Application number: PCT/DE2005/001011
Authority: WO
Inventors: Peter Koch
Original assignee: Universität Zu Lübeck
Priority date: 2004-06-09
Filing date: 2005-06-08
Publication date: 2005-12-22
Also published as: DE102004028204B3

Abstract

The invention relates to a method for evaluating signals in optical coherence tomography with an optical path length in the reference arm that can be modified over time. Said path length is modified by an iterative phase modulator during a measuring cycle from a first end value to a second end value. The method is characterised by the determination of the modification speed of the optical path length as a temporal function from the OCT measuring signal during at least one measuring cycle and the use of the determined modification speed for demodulating the OCT measuring signals that are recorded during the following measuring cycles.

Description

Process for signal evaluation in OCT

The invention relates to a method for signal evaluation in optical coherence tomography (OCT).

Optical coherence tomography allows the non-invasive display and measurement of structures within a tissue. A detailed example of an implementation is set out in US 5,321,501. The principle consists of first dividing radiation with a short coherence length, feeding the partial beams to the reference and sample arm of an interferometer, in which the light is reflected or backscattered, and finally superimposing the returning partial beams in order to evaluate the interference pattern. In the sample arm, the light is at one depth - e.g. biological - sample scattered, the depth-resolved scattering capacity of the sample being the size to be measured with the OCT. The optical path length is typically continuously changed in the reference arm. A certain sample depth only contributes to the interference signal if the optical path lengths of sample and reference light match within the coherence length of the light. With precise knowledge of the reference arm length, the scattering can thus be determined at a specific sample depth.

If the reference arm length is changed at a constant speed v between a front and rear end position (this defines the time span of the cycle length of a single so-called A-scan), then a frequency with a central wavelength λo of the radiation used results at the light detector

(1) ω ₀ = 2λ ₀ v

oscillating intensity I (t), which is modulated with the depth information.

(2) I (t) = A (t) - cos [ω ₀ t + φ]

The amplitude of the signal A (t) can thus be used to deduce the reflectivity in the sample depth corresponding to the reference arm length. The measured signal is usually processed by first going through a bandpass filter, the passband of which is chosen to be as narrow as possible. The bandwidth of the envelope A (t) is optimal here in order to largely eliminate the broadband noise. The amplitude of the signal is then demodulated by rectification and subsequent low-pass filtering. The detection of the amplitudes A (t) can then take place with a bandwidth which is generally approximately two orders of magnitude smaller than ω ₀ .

A number of devices are known in OCT technology which continuously change the length of the reference arm. In the simplest case, these devices, which are also referred to as phase modulators, are realized by a displaceable reference mirror (e.g. US 6,341,870 B1 or US 5,220,463). In US 6 111 645 an arrangement of pivotable mirrors or gratings between which reflected light travels for different lengths is presented. DE 198 14 068 AI describes phase modulators which are based on rotating mirror arrangements or prisms. Another effective way to create a phase modulator is to stretch an optical fiber (DE 100 35 833 AI).

In the phase modulators mentioned, the change in the reference arm length cannot be linear in time over the entire range, i. H. at constant speed v.

In the case of rotary phase modulators, this is a consequence of the inherently non-linear relationship between the constant angular velocity and the extension of the optical path length. In the case of phase modulators which repeat a linear movement, accelerated movements are necessary, and variable speeds arise from the usually non-linear behavior of the drives. The expansion behavior of the frequently used piezo actuators shows, for example, a pronounced hysteresis. Mechanical systems also have strong resonances at certain excitation frequencies. Therefore, repetitive phase modulators tend to deviate strongly from the desired behavior.

A variation in the speed v at which the optical path length and thus the phase φ of the reference radiation is changed is therefore reflected in the carrier frequency ωo of the measured intensity I (t), which becomes an amplitude and frequency-modulated oscillation. In order to take the frequency variation into account in the evaluation, one is forced to use filters with a much wider bandwidth than is actually desirable, as a result of which the noise component of the signal increases significantly. In addition, errors arise in the assignment of the current reference arm length, ie the depth of the scattering structure in the sample, to the measurement signal if the optical path length is not yet measured using an independent method. This is particularly noticeable when using OCT in profilometry or pachymetry.

One solution to demodulating signals with fluctuating carrier frequencies is frequency tracking using so-called “phase-locked loops” (PLLs). For this purpose, a carrier signal with sufficient amplitude is required, which also has no jumps in the phase position φ. Both conditions are not fulfilled in the OCT of scattering objects due to the statistical distribution of the scatterers in the sample. Even with a homogeneous distribution of the spreaders, the signal keeps collapsing to zero and phase jumps occur. This phenomenon results from the interference of radiation with statistically distributed phases and is analogous to the speckles known from coherent optics. Therefore, the usual electrotechnical methods for frequency tracking cannot be used with the OCT.

In order to achieve the greatest possible constancy of the carrier frequency ωo, phase modulators are therefore controlled either by means of signals which compensate for the deviations from the constant speed in the widest possible range of the depth scan. For this purpose, a one-time calibration measurement is carried out in order to calculate a corresponding correction function. It turns out, however, that a one-off calibration is often not sufficient; depending on the operating conditions and aging effects, the user has to record new calibration curves from time to time.

Alternatively, the control is implemented within a closed control loop (control). For this purpose, the optical path or the phase velocity in the phase modulator is precisely recorded using a suitable method. The measured values are then coupled back to the drive of the phase modulator in such a way that the reference arm length changes in good approximation at a constant speed. This regulation also overcomes the drift of the drive, and external influences on its behavior are compensated for.

In both cases, the optical path length covered in the phase modulator must be recorded. This was done, for example, by changing the angle of rotation or the distance between two modules is detected by means of suitable sensors. In the device according to DE 100 35 833 AI, z. B. additional capacitive distance sensors are used.

Methods are also known from the prior art which can directly determine the length of an optical path or a phase velocity using optical means. Interferometers or velocimeters are to be mentioned here in particular, which use a long-coherent laser light source to determine optical path length changes per unit of time between two reflecting surfaces.

Considerable technical requirements are placed on the devices for determining the optical path lengths. For a regulation according to DE 100 35 833 AI it is e.g. necessary to determine the position with an accuracy of less than 0.1 nm with a bandwidth of approx. 10 kHz. Regardless of whether a regulated or controlled operation is selected, drives with high resonance frequencies are required, which require considerable electrical power for control.

For all of these reasons, generating a constant phase velocity adds significantly to the overall cost of the phase modulators. It is complex to achieve repetition rates better than approx. 100 Hz with these methods. For phase modulators that are either based on rotary movements or are operated in mechanical resonance, control or regulation methods cannot be used at all.

It is an object of the invention to obtain measurement signals with a simplified OCT structure, in particular without components for controlling or regulating the phase modulator, which nevertheless allow filtering with a narrow bandwidth and thus evaluation with a good signal-to-noise ratio.

The object is achieved by a method with the features of the main claim. The subclaims indicate advantageous further training.

The invention makes use of the fact that all conventional phase modulators perform a repetitive (or also cyclical) movement. The reference arm length and its rate of change are therefore periodic as functions of time. Knowledge of the speed of the reference arm in the course of a single scan can thus be transferred to later scans for relatively short periods of time. The device according to the invention dispenses with means which force the constancy of the phase velocity in places. Rather, it is designed in such a way that the rate of change of the reference arm length is repeatedly recorded during the ongoing measurement and made available to a process computer. This does not require any additional measurement components, but is possible directly from the measurement data to be collected anyway, which are used primarily to examine the structure of the sample.

It is not immediately apparent to the person skilled in the art that the phase velocity can be measured using the velocimeter principle. Because a velocimeter according to the prior art requires a long-coherent light source (laser), which is not present in the OCT device. In contrast to a velocimeter, in which a reference mirror rests and a second mirror with the sample to be measured is moved, the OCT integrates the backscatter from all layers of the sample with the reflections from different reference arm positions one after the other. With a short-coherent light source, speed measurements are only possible within a measuring range that corresponds to the coherence length of the light source (approx. 10 μm). However, this is sufficient for the “local” or instantaneous measurement of the speed of the reference mirror.

The interference signal I (t) is sampled directly at a sampling rate of at least lωo. A window analysis of the OCT signal is then preferably carried out to determine ωo - on equidistant time base points tj. For example, the measured values of I (t) can each be subjected to a "local" Fourier transformation in a width defined along the time axis (a few fringes / oscillations of the carrier) in order to determine the dominant frequency in this time window Carrier frequency ωo at those points where the signal is sufficient and has no phase jumps can be determined and stored, and a table is obtained

(3) ω ₀ (t _i ) = 2λ ₀ - v (t _l )

on the support points

(4) t ,. = t ₀ + z ^' Δt with i = 0, ..., N

The reference arm lengths Zj, which are present at the times tj, can be determined by simple integration, for example via ι-l

(5) z, = z (t,) = z ₀ + ∑v (t _H ) Δt n = 0

with any offset z ₀ . It can also be advantageous to define an equidistant set z'j of support points, for example by

1 N- \

(6) z '= z ₀ + jAz with j =, ..., N and Δz = - V v (t „) Δt

Interpolating the measured OCT signal from the irregularly spaced coordinates z; to the z ' _j , it is freed from distortions due to the uneven phase velocity. Ideally, only a single carrier frequency occurs.

In order to be able to create a complete course of ωo over the entire measuring range, however, the sample must be expected to have reflections at all depths. If this is not the case, the most complete possible course of ωo can be achieved by varying the distance of the applicator from the measurement object. Advantageously, in the typical OCT examination of living samples, in particular also on or in a patient, the inevitable self-movement of the sample gives the opportunity to obtain speed information for all reference arm lengths from a plurality of successive scans. At those points where the carrier frequency cannot be determined due to a missing signal or phase jumps, previous measurements are then used. Because of the statistical structure of this phenomenon with scattering objects, it can be expected that the phase velocity can be determined and stored in this way for all optical delays over a certain time. The information about the carrier frequency ωo should be updated again and again in the process computer in the course of the OCT measurement.

The course of ωo over the measuring range determined from the current and many previous scans is then used for demodulation, which can either take place mathematically on the digitized form of the interference signal I (t) or in an analog manner (for example variation of the filter frequency, mixing with ωo). For example, when mixing in an intermediate band - preferably to zero differential frequency (lock-in) - the one sought Amplitude modulation can either be recorded directly or at least impressed on a largely constant carrier frequency that can be filtered with a narrow bandwidth.

The direct determination of the carrier frequency from the interference signal described here is also advantageous because an incomplete coupling between the mechanics and the optical path then does not lead to a falsification of the measurement result.

With regard to the concrete implementation of the mathematical determination and especially demodulation, there are certainly many different ways to go. The invention does not want to be understood to be limited to certain algorithms. It is essential in all cases that the course of ωo can be recorded on practically any sample in normal measurement operation without the measurement intended by the end user being impaired in any way.

The above-mentioned ways of using the speed information about the reference arm aim at a subsequent improvement of the OCT measurement in the course of post-processing of the electronically recorded data. However, the recording itself can also be influenced in such a way that a set of measurement data that is easier to analyze is obtained from the outset.

For this purpose, it is proposed to collect the measurement data on the equidistant support points z ' _j according to (6). Corresponding, irregularly spaced times t ' _{j must} be calculated and transmitted to the AD converter so that the latter detects the signal at the specified times.

One way to calculate the _t'j is as follows:

Every value z calculated according to (5); is assigned to any interval of width Δz, ie a list of pairs (z ;, z ' _j ) is created with the property

(7) z '. ≤ z, <z '7 + 1

and calculated

(8) ^{t, J + = t, + Z} v ^j (t) '^ ^{i = 0>} - ^{> N xmά = 0>} - ^{> N} - ¹ from the equidistant measuring times t used first, and the values z and v (t _ι ) determined for this purpose. The measured values raised to the t _'j give an OCT signal curve that is free of motion artifacts of the phase modulator. Further evaluation can take place with a good signal-to-noise ratio according to the prior art.

Claims

Expectations

1.Procedure for signal evaluation in OCT with a time-varying optical path length in the reference arm, which is changed by a repetitive phase modulator during a measuring cycle from a first to a second end value, characterized by - determining the rate of change of the optical path length as a function of time from the OCT measurement signal during at least one measurement cycle and - using the determined rate of change to demodulate the OCT measurement signal recorded during the following measurement cycles.

2. The method according to claim 1, characterized in that the rate of change of the optical path length is determined by a window analysis of the carrier frequency of the amplitude-modulated OCT measurement signal.

3. The method according to claim 2, characterized in that in windows of a few period lengths in width, the OCT signal is subjected to a Fourier transformation.

4. The method according to any one of the preceding claims, characterized in that the determination of the rate of change of the optical path length for at least two sections between the first and the second end value are obtained from different measurement cycles.

5. The method according to claim 4, characterized in that the rate of change of the optical path length for the entire range between the first and the second end value is combined from the measured values for the sections.

6. The method according to any one of the preceding claims, characterized in that the demodulation of the OCT measurement signal is carried out by subsequently calculating a demodulated signal.

7. The method according to claim 6, characterized in that the OCT measurement signal is mixed with the determined time-dependent carrier frequency.

8. The method according to any one of claims 1 to 5, characterized in that the demodulation of the OCT measurement signal takes place in the measurement cycles following the determination of the rate of change of the optical path during the data recording of the OCT measurement.

9. The method according to claim 8, characterized in that points of time are calculated and transmitted to the measuring unit by means of the rate of change of the optical path, at which the OCT measurement signal is recorded.