The invention relates to the field of interferometric optical imaging. The invention provides a device that combines holoscopic acquisition and reconstruction techniques with a one dimensional confocal gate.
Hillmann et al have described holoscopy, which is similar to digital holography, where a full field Fourier Domain optical coherence tomography (FD-OCT) image is recorded by exposing a 2D sensor array to the light scattered from an object and reference surface from a swept frequency source (D. Hillmann et al, “Holoscopy—holographic optical coherence tomography” Optics Letters 36(13): 2390 2011 hereby incorporated by reference). In holoscopy, unlike classic OCT, the sensor array is typically imaged to the far field of the object rather than to the object. As a result, each sensor element on the 2D array records light scattered in a different direction, encoding a lateral spatial frequency, rather than light from a different position on the object. The axial spatial frequencies are encoded by the optical frequency of the source, just like in swept-source OCT. The data may be reconstructed to a volume by three-dimensional Fourier transform. The reconstruction is notable in that high lateral resolution and power efficiency can be achieved far from the equivalent Rayleigh range calculated from the numerical aperture of the system. Full field systems may also allow a greater amount of light (Maximum Permissible Exposure) than point scanning systems on sensitive tissues such as the eye as the illumination is not focused to a point on the retina. By detecting at a defocused plane, the reconstruction combines information from a larger number of sensor elements to a single output pixel. This method requires a very fast camera and a high power swept source. Full field systems are also particularly susceptible to multiply scattered photons, and light scattered or reflected from surfaces far from the object of interest because they have no confocal light restriction.
Nakamura et al described a line-field spectral domain OCT system that could acquire spectral domain data for a full B-scan in a single exposure of a 2D array (Y. Nakamura, et al, “High-speed three dimensional human retinal imaging by line-field spectral domain optical coherence tomography” Optics Express 15(12):7103 2007 hereby incorporated by reference). A vertical line of light was imaged onto the retina of an eye and reimaged onto the entrance silt of a spectrometer. Each point at the entrance slit of the spectrometer corresponded to a portion of the retina object under observation. The spectrometer decomposed the light from each point on the entrance slit of the spectrometer into a spectrum represented as a column on the 2D sensor. A standard SLD and common 2D image sensor could be used; effectively achieving a motion artifact free B-scan without very high speed electronics Like full field systems, the laterally distributed light allows a much greater exposure on sensitive tissues. Unlike the full field holoscopy systems described earlier, the sensor array in this design is imaged to the retina, rather than to the far field of the retina. The simple one-dimensional data reconstruction along the axial direction suggests that the system has the same limits of lateral resolution typically experienced by flying-spot OCT systems. The confocal gate in one lateral dimension, created by the narrow line illumination and the spectrometer entrance slit, is partially effective at eliminating light scattered from out of focus planes. Because of system aberrations, likely in the spectrometer, the system suffered from a strong SNR roll-off at the edges of the line profile.
The subject invention is a special form of holoscopy that contains the advantages of both of the above techniques but avoids the disadvantages of each. A line of light is focused on an object to be investigated through an imaging system. The light scattered from the investigated object is combined with reference radiation and collected by a sensor array. An aspect of the invention is the introduction of astigmatic optics into the return path such that the narrow axis of the line focus is substantially imaged directly to the narrow axis of the detector and such that the line focus is defocused along its long axis on the sensor array. This results in a line of light at the sensor plane corresponding to the line of light on the retina, but where the lateral spatial frequency, or angular distribution of scattered light, is encoded along the line of detection. In a preferred embodiment, this line of light at the detection plane is projected onto the entrance slit of a spectrometer, such that at the 2D array sensor of the spectrometer, the spectral information along one axis of the sensor caries the axial frequency spectrum of the object, while the orthogonal direction encodes the lateral frequency information. The data may be reconstructed to a B-scan by two-dimensional Fourier transform with or without combination of more complex algorithms. In another embodiment, this line of light is projected onto a 1D sensor array and the illumination optical frequency is swept over time, resulting in the spectral information being encoded as a function of time, as in a swept source OCT system. A useful swept source produces a very narrow instantaneous bandwidth, which shifts with time such that a wide bandwidth is covered in short period of time, and sequential recordings at a detector, over the same time period, each contain a measurement of a very narrow optical frequency band.
The proposed solution, holoscopic line-field OCT, combines holoscopic acquisition and reconstruction techniques with a one dimensional confocal gate. It thus achieves a happy medium between the ‘wide open’ situation in full field holoscopy, which collects scatter even from surfaces very far from the object of interest; and the precisely confocal situation in point scanning flying-spot OCT systems. Allowing light to be detected at significant distances away from the focused region allows the system to better take advantage of digital refocusing techniques made possible by holoscopic reconstruction, without introducing the difficulties introduced by overwhelming reflections from surfaces such as the cornea of the eye. A preferred embodiment will possess the reconstruction advantages demonstrated by Hillmann with the economy of the Nakamura approach: high lateral resolution and power efficiency at a greater range of depths, with a simple SLD source and low speed 2D sensor array. Note that the amount of power allowed to be simultaneously incident across a line or area of the retina is far greater than the amount allowed focused to a point, allowing for far greater potential sensitivity than can be achieved by flying-spot OCT systems, particularly for objects where light exposure is limited such as the retina of the eye. Additionally the arrangement may have a SNR advantage due to the Fourier processing method as proposed by Blazkiewicz et al.
It should be noted that although the description above places the sensor plane at the far field of the retinal plane, one could also consider an intermediate case, where the astigmatic optics defocus the light along the long axis of the line focus, such that the interference for a single point on the retinal surface will be detected by multiple sensor elements, without defocusing all the way into the far field. In general, one would like to defocus the sensor by at least one Rayleigh length. In determining Rayleigh length we are concerned with the light as it hits the sensor and with dimension of concern in the long axis of the line focus. The light as emitted from a point in the scattering object, and transmitted through the limiting apertures of the optical system should be considered. In this case one maintains advantages conferred by detecting information from a single lateral location within the object plane upon a plurality of sensor pixels. Computational wavefront propagation from a generically defocused plane to a focused plane requires more complexity than the simple Fourier transform described in our preferred embodiment, however a plurality of methods for this task are well understood in the field of digital holography.
a and 1b illustrate background concepts of conjugate planes and Fourier planes useful for understanding aspects of the present invention.
a, 2b and 2c illustrate one embodiment of the present invention wherein
a and 3b illustrate an alternative and preferred embodiment of the present invention wherein
a and 4b illustrate an embodiment of the invention using a swept-source and linear sensor array wherein
a and 5b illustrate the data orientation on the sensor wherein
In the simplified optics described below, spaces are described with their imaging relationship to the object plane, which contains the object being investigated. Spaces within the system are identified as either optically conjugate to the object plane or at a Fourier plane with respect to the object plane.
The combination of plane waves at different incident angles results in sinusoidal interference which is well described by the spatial frequency of interference. We therefore describe different locations within a Fourier plane as containing the information about different spatial frequencies of the object plane. Pupil plane and ‘far field’ are used synonymously with a plane that is located at a Fourier plane relative to the object plane. Planes located between the object and Fourier plane, or between the Fourier and image plane can be described as defocused, but not entirely into the far field. The Rayleigh length denoted R in
In real cases it may be practical to design a system without strict maintenance of conjugate and Fourier planes at all interfaces, however it greatly simplifies the discussion of the important concepts to use these extremes. We therefore illustrate lenses such that, with an object plane at one focal distance away from the lens on a first side, the Fourier plane is located at one focal distance away from the lens on a second side. Because of astigmatic, asymmetrical elements such as cylindrical lenses, it is possible that a plane can be conjugate in one dimension, while it is a Fourier plane in the other dimension.
A key aspect of the invention is the introduction of astigmatic optics into the system such that the narrow axis of the line focus is substantially imaged directly to the narrow axis of the detector and such that the line focus is defocused along its long axis on the sensor array. This results in a line of light at the sensor plane corresponding to the line focus on the object, but where the lateral spatial frequency, or angular distribution of scattered light, is encoded along the line of detection. In general, one would like to defocus the sensor by at least one Rayleigh length. In determining Rayleigh length we are concerned with the light as it hits the sensor and with dimension of concern in the long axis of the line focus. The light as emitted from a point in the scattering object, and transmitted through the limiting apertures of the optical system should be considered. In this case one maintains advantages conferred by detecting information from a single lateral location within the object plane upon a plurality of sensor pixels. Computational wavefront propagation from a generically defocused plane to a focused plane requires more complexity than the simple Fourier transform described in our preferred embodiment, however a plurality of methods for this task are well understood in the field of digital holography (see for example Schnars et al. Digital Holography: Digital Hologram Reporting, Numerical Reconstruction, and Related Techniques Springer 2005).
Broadband light 102 passes through a collimating lens 103 and through optics 104 to generate a line shaped focus in a beam. A superluminescent diode (SLD) would be a typical light source, but any broadband light source could be used in the present invention and the line shape could be a property of the source instead of requiring additional optical elements. The source could also be swept in optical frequency as will be described further below. The line of light has extent in the y or vertical direction for this embodiment. The line of light is split into two paths of an interferometer at beam splitter 105. The light in one path, the reference arm, passes through a cylindrical lens 106 before reflecting off a minor 107. The purpose of the cylindrical lens is to substantially collimate the beam in both the x and y directions to make the arm relatively insensitive to changes in length, such that the arm may be modified to introduce a variable optical delay. Light in the second path, the sample arm of the interferometer, passes through spherical lens 108, which collimates the light in the x-dimension and focuses the line of light in the y-dimension to have a beam waist at the pupil plane. After passing the pupil plane, the light in the sample arm is focused to a line on the sample using one or more lenses 109. In the case of imaging the eye, focusing could be achieved all or in part using optical properties of the anterior segment of the eye itself. Light reflected from the sample and light reflected from the reference minor is recombined at beamsplitter 105 and directed towards the spectrometer entrance slit 113. The entrance slit is elongated in the y dimension as illustrated in
In fundamental contrast to the arrangement by Nakamura et al., positions across the x-direction on the 2D sensor correspond to optical frequencies and positions across the 2D sensor in the y-direction correspond to positions in the imaging pupil which in turn correspond to spatial frequencies in the line image (in the y-direction); rather than spatial positions along the line image. This is achieved by the insertion of astigmatic optics in the return path of the light towards the sensor.
Nakamura et al used an optical chopper to illuminate the detector with pulses of light in order to reduce the effect of phase washout that may occur if the object moves during the exposure time of the sensor. (Y. Nakamura, et al, “High-speed three dimensional human retinal imaging by line field spectral domain optical coherence tomography” Optics Express 15(12):7103 2007). A lower cost way to achieve this effect is to use an electronic shutter feature as provided by even many low cost 2D arrays. Global shuttering may provide a snapshot with the least sensitivity to motion, while rolling shutting may provide specific sensitivity to different types of motion. For example if a rolling shutter is introduced such that a range of optical frequencies across all lateral angles is exposed in sequence, axial motion will be encoded as a dispersion in the optical frequency information. An exposure efficient way to limit light without the complexity of mechanical chopping could include a pulsed electrical drive system to the illumination source.
After the interferometric data is collected, reconstruction of the data in the spatial frequency domain may be performed in the processor by “multiplication with a conjugated reference wave and propagation of the light waves by the angular spectrum approach” as described by Hillmann before or after reconstruction of the data in the spectral frequency domain. (D. Hillmann et al, “Holoscopy—holographic optical coherence tomography” Optics Letters 36(13): 2390 2011 hereby incorporated by reference). Ralston et al. propose alternative reconstruction techniques in U.S. Pat. No. 7,602,501. Propagation and reconstruction techniques are well understood in the field of Digital Holography and could be easily extended to the system described herein by someone skilled in the art. Because the data is in the spatial frequency domain in only in the y and z axes, and is already in the spatial domain in the x axis, methods are modified accordingly. That is, the conjugated reference wave has variations only in the y direction, and propagation should be performed in a 2 dimensional rather than 3 dimensional space, approximating that the beam phase is constant across the x dimension. A remapping function could be applied to shape the optical frequency spectrum and angular direction to a well behaved grid of spatial frequencies prior to two-dimensional Fourier transform. After reconstruction, various images can be generated of the data.
Several alternative embodiments of the invention are possible, and a larger variety of substitutions might be imagined in a more complex arrangement in particular where the conjugate object plane is reimaged more times. For example, in a preferred embodiment of the present invention illustrated in
The illumination from the reference beam incident upon the confocal limiting aperture of the detector or sensing device may be introduced at an angle relative to the light returning from the sample. This arrangement introduces a phase ramp across the interference data. The phase ramp across the interference data allows for advantages in reconstruction as described by Franke et al. hereby incorporated by reference (Franke et al. High resolution Holoscopy” Proc SPIE 8213 821324 2012).
It is also possible to replace the SLD light source with a light source swept in optical frequency, combined with a linear array detector rather than the 2D spectrometer array as is illustrated in
In this embodiment, time can be considered the axis on which optical frequency is encoded at the detector. As illustrated in
Further, a scanning mechanism may be added such that multiple line images may be acquired in order to build up greater knowledge of a volume of the sample. The line focus described should be interpreted primarily in terms of its one dimensional confocal gating properties, rather than strictly its geometrical shape. For example, an arc shaped illumination and matching detection selection would provide a similar confocal limitation. The amount of confocal restriction can be modified to some degree to control the amount of out of focus light which reaches the detector. This may be achieved by adjusting width of the line focus (e.g. by adjusting the numerical aperture of the illumination beam if diffraction limited), or by adjusting the width of the detection restriction, or a combination of both.
The invention has application in the imaging of scattering objects, especially for biomedical investigation, in particular the eye and retina. In this case, the sample and nearby optics may be provided by the eye itself. The device may also be used for other biomedical imaging applications such as cardiovascular, dental, cellular, intra-surgical guidance of hollow and solid organs, or for industrial applications such as image based materials inspection.
The invention is particularly beneficial if used to image objects where motion sensitivity is especially critical: e.g. objects that are moving quickly such as living specimens, object where small amounts of motion reduce the value of a measurement such as in the case of corneal curvature.
Although various embodiments that incorporate the teachings of the present invention have been shown and described in detail herein, those skilled in the art can readily devise many other varied embodiments that still incorporate these teachings.
The following references are hereby incorporated by reference:
DE Patent No. 4309056 Haeusler et al. “Method for determining the distance and intensity of scattering of the scattering points”
U.S. Pat. No. 7,602,501 Ralston et al. “Interferometric Synthetic Aperture Microscopy”
PCT Publication NO. WO 2012/143113 Hillman et al. “Method for Optical Tomography”
D. Hillmann et al, “Holoscopy—holographic optical coherence tomography” Optics Letters 36(13): 2390 2011.
Y. Nakamura, et al, “High-Speed three dimensional human retinal imaging by line field spectral domain optical coherence tomography” Optics Express 15(12):7103 2007.
Blazkiewicz et al, “Signal-to-noise ratio study of full-field Fourier-domain optical coherence tomography” Applied Optics 44(36):7722 (2005).
Franke et al, “High resolution Holoscopy” Proc SPIE 8213 821324. (2012).
Gaskill, Linear Systems, Fourier Transforms, and Optics John Wiley & Sons, Inc. 1978.
Schnars et al. Digital Holography: Digital Hologram Reporting, Numerical Reconstruction, and Related Techniques Springer 2005
This application claims priority to U.S. Provisional Application Ser. No. 61/589,172 filed Jan. 20, 2012 hereby incorporated by reference in its entirety.
Number | Date | Country | |
---|---|---|---|
61589172 | Jan 2012 | US |