TOMOGRAPHY

The present invention relates to the field of optical coherence tomography, and in particular to a method of and apparatus for producing a tomogram of a sample.

Optical coherence tomography (OCT) is widely used in medical applications such as ophthalmology and optometry to produce scans of structures such as the human eye, and in particular the retina. A known frequency domain optical coherence tomography (FD-OCT) system is illustrated schematically at 10 in Figure 1. The FD-OCT system 10 of Figure 1 will be familiar to those skilled in the art, and comprises a low coherence light source 12, such as a femtosecond laser, which produces a beam of light E ₁ . The beam E ₁ is incident on a beam splitter 14, where part of the beam E ₁ is diverted towards a reference mirror 16 as beam E _r and part of the beam E ₁ proceeds towards a sample to be analysed, such as an eye, as beam E _s . The beam E _s passes through a lens 20 or other suitable optical arrangement where it is focused on a part of the sample 18.

The reference beam E _r is reflected by the reference mirror 16 and the reflected reference beam E _r (τ) passes through the beam splitter 14 to form part of an analysis beam Ed. Part of the sample beam E _s is reflected by the sample 18 as beam E _s , and the intensity of the beam E _s is dependent upon the reflectivity of the sample 18, or more specifically of different layers of the sample 18. The beam E _s is diverted by the beam splitter 14 to form part of the analysis beam E _d .

The analysis beam E _d is incident upon a diffraction grating 22, which causes it to split into separate beams at the component wavelengths of the analysis beam Ed, and these beams are incident on a detector 24 such as a one dimensional CCD array. Each CCD element of the array produces an output signal whose amplitude varies according to the intensity of the component beam which is incident upon it, and these outputs are processed in a post processor 26, such as a personal computer, processor IC or similar device which performs an inverse transformation using a method such as the Inverse Fourier Transform (IFT) or a filter bank to produce a depth scan, which is a vector of reflectivity values of the part of the sample 18 on which the sample beam E ₃ is focused by the lens 20. The lens 20 can cause the beam E ₃ to be focused on other parts of the sample 18 and by combining depth scans taken at a plurality of parts of the sample 18 a tomogram can be constructed which may be displayed on a suitable display device 30 such as a computer screen.

FD-OCT exhibits an effect known as the complex conjugate artefact, which is caused by the symmetrical nature of the IFT of the real value measured signal used in the post processor 26. The IFT produces a complex signal with components in both positive and negative ranges, with the components in the negative range being referred to as "mirrored images". Additionally, an "aliasing artefact" appears due to limited sampling if the sample to be imaged is not entirely situated in either the positive or negative image range. In this case the images of the parts of the sample which lie outside the positive or negative image range will fold back or alias into the image range and will thus overlap the actual image signal.

Essentially the negative range is equal to the positive range if the sample 18 causes no dispersion of light, and there is no dispersion in the reflected reference beam E _r (τ). However, in structures such as the human eye which contain dispersive material (the aqueous humour, in the case of the human eye), this is not the case, and blurring of the final tomogram occurs. It will be appreciated by those skilled in the art that other effects such as attenuation caused by absorption of light or scattering can alter the sample beam, although these effects are typically much weaker and less predictable than dispersion effects.

Various techniques have been proposed to remove the dispersion blurring effect . For example, dispersion matching techniques are known, in which a dispersive element is placed in the path of the reflected reference beam E _r (τ) to counteract the dispersing effect of the sample 18. Dispersion compensation can also be achieved numerically by appropriate processing of the signals produced by the detector 24. This numerical dispersion compensation will sharpen the full range image and blur the mirrored images in both the positive and negative image ranges. Additionally, techniques have been developed to measure the complex valued signal (i.e. both the positive and negative ranges produced by the IFT) to reconstruct the full imaging range which can be used to construct a tomogram. However, these techniques all require at least two depth scans to be made to construct a tomogram which uses the full imaging range and suppresses the artefacts mentioned above.

According to a first aspect of the present invention there is provided a method of generating a tomogram of a sample wherein an analysis beam is subject to dispersion distortion, the method comprising: a) detecting the analysis beam; b) generating a signal representative of the intensity of wavelengths in the analysis beam; c) generating a first spatial domain signal representative of the signal so generated; d) detecting a desired signal component of the first spatial domain signal and outputting the desired signal component for use in generating a depth scan which can be used to create a line of the tomogram; e) subtracting the desired signal component from the first spatial domain signal to generate a second spatial domain signal; f) generating an intermediate signal which does not include the desired signal component; and g) calculating the complex conjugate of the desired signal component in the first spatial domain signal, subtracting it from the intermediate signal and creating a signal from which a new first spatial domain signal can be generated for use in steps c) to g).

The method of the present invention takes advantage of the fact that the image components produced by the IFT in a normal FD-OCT system contain information which can be used to construct a full range tomogram without having to perform additional depth scans. Thus, using the method of the present invention a full range tomogram can be acquired more quickly with individual independent depth scans than in prior art systems, albeit at the cost of increased complexity in post processing. Although the term "tomogram" usually refers to two- dimensional images, it will be appreciated that the method of the present invention can be used to generate a single line of a tomogram, or even a single pixel of a tomogram. Thus, it is to be understood that the term "tomogram" as used herein is intended to encompass any single or multiple depth scan forming a tomogram or volumetric image.

Numerical dispersion compensation may be performed on the signals generated in step b) and g)-

The complex conjugate of the desired signal component in the first spatial domain signal may be subtracted from the intermediate signal at a position in the intermediate signal at which a mirrored image of the desired signal component appears

Step f) may comprise performing a Fourier transform on the second spatial domain signal, applying inverse dispersion compensation and performing an inverse Fourier transform on the resulting signal.

Alternatively, step f) may comprise performing a Fourier transform on the second spatial domain signal, applying inverse dispersion compensation and passing the resulting signal through a filter bank.

Creating a signal from which a new first spatial domain signal can be calculated may comprise performing a Fourier transform on the signal produced by the subtraction and applying dispersion compensation to the Fourier transformed signal.

The analysis beam may be split into separate beams at each of the component wavelengths of the analysis beam, the separate beams being detected in step a).

The analysis beam may be split into its component wavelengths using a diffraction grating.

In an alternative embodiment, light from a frequency-scanning light source may be split into sample and reference beams which combine to form the analysis beam. The dispersion distortion in the analysis beam may result from an imbalance of dispersive elements in reference and sample beams.

For example, the dispersion distortion in the analysis beam may result from dispersal which is introduced into one or more of a reference beam, a reflected reference beam, a sample beam and a reflected sample beam by means of a dispersive element.

The depth scan may be generated by adding the desired signal component to a signal containing a full range depth scan of the sample and taking the magnitude of the resulting signal.

According to a second aspect of the present invention there is provided a computer program for performing the method of the first aspect.

According to a third aspect of the present invention there is provided apparatus for generating a tomogram of a sample wherein an analysis beam is subject to dispersion distortion, the apparatus comprising a beam detector for detecting the analysis beam and generating a signal representative of the intensity of wavelengths in the analysis beam; a transformation unit for generating a first spatial domain signal representative of the signal so generated; a component detector for detecting a desired signal component of the first spatial domain signal and outputting the desired signal component for use in generating a depth scan which can be used to create a line of the tomogram; a subtractor for subtracting the desired signal component from the first spatial domain signal to generate a second spatial domain signal; a transformation unit for generating an intermediate signal which does not include the desired signal component; a calculator for calculating the complex conjugate of the desired signal component in the first spatial domain signal and a subtractor for subtracting the complex conjugate from the intermediate signal and creating a signal from which a new first spatial domain signal containing a desired signal component can be generated. The apparatus according may further comprise a compensator for performing numerical dispersion compensation on the signal representative of the intensity of the wavelengths in the analysis beam and on the signal from which the new first spatial domain signal can be generated.

The subtractor may subtract the complex conjugate of the desired signal component in the first spatial domain signal from the intermediate signal at a position in the intermediate signal at which a mirrored image of the desired signal component appears

The apparatus may further comprise a transformation unit for performing a Fourier transform on the second spatial domain signal, a compensator for applying inverse dispersion compensation and a transformation unit for performing an inverse Fourier transform on the resulting signal.

Alternatively, the apparatus may comprise a transformation unit for performing a Fourier transform on the second spatial domain signal, a compensator for applying inverse dispersion compensation and a filter bank for processing the resulting signal.

The apparatus may further comprise a transformation unit for performing a Fourier transform on the signal produced by the subtraction of the complex conjugate and a compensator for applying dispersion compensation to the Fourier transformed signal.

The analysis beam may be split into separate beams at each of the component wavelengths of the analysis beam, the separate beams being detected by the beam detector.

The analysis beam may be split into its component wavelengths using a diffraction grating.

In an alternative embodiment, the apparatus may comprise a frequency-scanning light source which emits light which is split into sample and reference beams which combine to form the analysis beam. The dispersion distortion in the analysis beam may result from an imbalance of dispersive elements in reference and sample beams.

For example, the apparatus may further comprise one or more dispersive elements which can be introduced into one or more of a reference beam, a reflected reference beam, a sample beam and a reflected sample beam.

Embodiments of the invention will now be described, strictly by way of example only, with reference to the accompanying drawings, of which

Figure 1 is a schematic illustration of a known Michelson interferometer based frequency domain optical coherence tomography system;

Figure 2 is a schematic illustration of an exemplary frequency domain optical coherence tomography system for performing the method of the present invention; and

Figure 3 is a schematic illustration showing functional blocks of a post processor used to perform the method of the present invention.

Referring first to Figure 2, a frequency domain optical coherence tomography (OCT) system is shown generally at 40. The system 40 is similar to the system 10 shown in Figure 1 and thus like elements are identified by like reference numerals, although it will be appreciated by those skilled in the art that the principles of the invention can equally be applied to other OCT systems, such as a swept source OCT system.

The system 40 includes one or more dispersive elements 42 which are made of or contain a dispersive material such as water, and can be inserted into or removed from the path of the reference beam E _r or the reflected reference beam E _r (τ) and/or the path of the sample beam E _s or the reflected sample beam E _s to cause an imbalance in dispersive elements in sample and reference arms of the system and thus create dispersion distortion in the analysis beam Ed which can be used in a post processing stage performed by a post processor 44 to retrieve information from the full range of the signals produced by the detector 24. It will be appreciated that where the sample 18 contains dispersive material, an imbalance in dispersive material in the sample and reference arms of the system 40 will exist at the outset and thus the analysis beam E _d will contain dispersal information without the insertion of the dispersive element 42, and thus no additional dispersion may be required. Conversely, if the sample 18 does not contain any dispersive material, there will be no imbalance between the sample and reference arms and the analysis beam Ed will not contain any dispersal information unless the dispersive element 42 is inserted into the path of the reference beam E _r or the reflected reference beam E _r (τ) or the path of the sample beam E _s or the reflected sample beam E _s .

Figure 3 is a schematic illustration showing functional blocks of the post processor 44. It is to be understood that the functional blocks shown in Figure 3 do not necessarily represent actual components of the post-processor 44, but are intended simply to illustrate steps performed by the post processor 44. The post processor 44 may be a personal computer or a microprocessor executing a suitable set of instructions, or may be any other appropriate device, such as a DSP, FPGA or the like.

The post processor 44 receives as its initial input a signal s produced by the detector 24, which provides information on the intensity of each of the separate different wavelength components of the analysis beam E _d . This signal undergoes numerical dispersion compensation in compensation block 46, to compensate for the effect of dispersion caused either by dispersive material in the sample 18 or the dispersive element 42. Suitable numeric dispersion compensation techniques are known in the art, and thus will not be described in detail here.

Following numeric dispersion compensation, the signal is passed to transformation block 48, where it undergoes an Inverse Fourier Transform, to produce a spatial domain signal S (i.e. a depth scan showing reflectivity vs. depth in the sample 18 at the area of the sample 18 on which the analysis beam E _s is focused by the lens 20). A peak detector 50 is used to detect a desired true signal component of the spatial domain signal S, that is to say a peak occurring in the spatial domain signal S. This peak may occur in either the positive or the negative range of the spatial domain signal, and once detected this desired true signal component (which is a complex value) is added in an adder 70 to a signal t which contains a full range complex valued depth scan of the sample 18. The resulting signal t' is used to generate a line of a tomogram, with the magnitude of the signal t' typically forming one line of a tomogram. Subsequent pixels of the tomogram are generated without having to perform further scans of the sample 18 using information contained in the whole of the spatial domain signal S, as will be described below. The pixels generated are not necessarily fixed, but may be adjusted by subsequent iterations of the method as new signal components, which were previously obscured by strong interfering components, become visible as the interfering components are removed by successive iterations of the method.

In subtractor 52 the true signal component detected by the detector 50 is subtracted from the spatial domain signal S, leaving a spatial domain signal S ^' containing an image of the true signal component, but not the true signal component itself. The signal S ^' undergoes a Fourier transform in transformation block 54 to produce a signal ϊ ^~ which contains information on the intensity of each of the separate wavelength components of the signal s. However, the signal ? ^~ does not contain any information on the true signal component, as this component is not present in the signal S ^' , and thus the signal ? ^~ is different from the signal s.

The signal ϊ ^~ is passed to an inverse compensation block 56, where the inverse of the numerical dispersion compensation applied in compensation block 46 is applied to the signal ϊ ^~ . Thus, the dispersion compensation is removed from the signal ϊ ^~ to produce an uncompensated version of the signal ? ^~ .

In an inverse compensation block 58 the inverse of the original numeric dispersion compensation, as applied in compensation block 46, is applied to the uncompensated version of the signal ? ^~ to produce a negative dispersed compensated signal, and this negative compensated signal undergoes an Inverse Fourier Transform in transformation block 60, thus producing a spatial domain signal S ⁺ , which is different from the spatial domain signal S as it does not contain the desired true signal component which was previously output and added to the signal t.

As the Inverse Fourier Transform has symmetrical properties, the complex conjugate of the true signal component of the spatial domain signal S output by the peak detector 50 is theoretically identical to the true signal component itself. The complex conjugate of the true signal component of the spatial domain signal S is calculated in calculator block 62, and in subtractor 64 this complex conjugate is subtracted from the spatial domain signal S ⁺ , at the position in the spatial domain signal S ⁺ at which the mirrored image of the true signal component appears, to produce a new spatial domain signal S' which contains neither the true signal component output by the peak detector 50, nor the image of that signal component. The new spatial domain signal S' undergoes a Fourier transform in transformation block 66 and dispersion compensation in compensation block 68, to undo the inverse dispersion compensation applied in compensation block 58, thus producing a new signal s ' containing no information relating to the true signal component previously output. This new signal s ' is then used as the input signal of the post processor 44, which generates a new output signal at the output of the peak detector 50 for use in generating a further signal component to add to the tomogram line. This process repeats as many times as there are components in the original signal s, such that a complete tomogram can be created using a single measurement taken from the sample.

Although in the embodiment shown in Figure 3 the complex conjugate is subtracted from the spatial domain signal S ⁺ at the position in the spatial domain signal S ⁺ at which the mirrored image of the true signal component appears, it will be appreciated, however, that the complex conjugate could be subtracted from the spatial domain signal S ⁺ at a different position. Similarly, the complex conjugate need not be subtracted from the spatial domain signal S ⁺ , but could be subtracted from the signal ϊ ^~ , if an appropriate transformation were applied. Those skilled in the relevant art will appreciate that the method of the present invention can be applied using a variety of tomography systems. Thus, although the embodiment described above uses a diffraction grating 22 to split the beam Ed into its component wavelengths, it will be understood that other configurations may be used. For example, a swept source OCT system could be used, in which a frequency-scanning light source emits light which is split into sample and reference beams which combine to form the analysis beam. Similarly, an optical- fibre based system, in which fibres of different lengths are used to introduce dispersion distortion in the analysis beam, could be employed in place of the modified Michelson interferometer described above.

Similarly, a filter bank can be used in place of the inverse transformation block 60 to generate the spatial domain signal S ⁺ . The implementation of such a system will be familiar to those skilled in the art, and thus will not be described in detail here.

Moreover, although the method has been described above in relation to an optical tomography system, it will be appreciated that it is also applicable to optical biometry systems.

As is mentioned above, the post processor can be implemented in a variety of ways, for example as a personal computer executing a suitable set of instructions defined in a computer programme or as a suitably-programmed DSP, FPGA or the like.

The method of the present invention may be modified or optimised, for example by employing a global estimate of the power spectrum of the light source, which is used for pulse shaping of detected desired true signal components. Thereby leakage of symmetric terms can be further reduced.

Although the term "tomogram" usually refers to two-dimensional images, it will be appreciated that the method of the present invention can be used to generate a single line of a tomogram, or even a single pixel of a tomogram. Thus, it is to be understood that the term "tomogram" as used herein is intended to encompass any single or multiple depth scan forming a tomogram or volumetric image.