Non-mydriatic hyperspectral ocular fundus camera

Abstract

This document describes an ocular base imaging device (11) comprising an illumination module (140) and an imaging module (141). The illumination module (140) includes light sources (103, 104) configured to generate light at wavelengths within a desired spectral range. A first optical assembly is provided to shape and direct the light onto the subject's eye (102). A tunable bandpass filter (109) selects wavelength sub-intervals within the desired spectral range. The imaging module (141) includes a second optical assembly for collecting light reflected from the subject's eye (102) and projecting the reflected light onto an image sensor (113). The second optical assembly includes one or more optical elements capable of compensating for ocular variations. The image sensor (113) is configured to image the reflected light to generate an image of the ocular base at the wavelength sub-intervals.

Description

Technical Field

[0001] This disclosure relates to the field of optical substrate imaging. Embodiments of this disclosure relate to an ocular substrate imaging apparatus, a method for compensating for the heterogeneity of images recorded by the ocular substrate imaging apparatus, and a method for recovering spectral information of a sample from multiple independent spectral measurements.

[0002] Some embodiments of this disclosure provide systems and methods for recording non-mydriatic hyperspectral images of the base of a subject's eye. However, it should be understood that this disclosure has other applications, including multispectral imaging and mydriatic imaging. Background Technology

[0003] The Importance of Color: An ocular basement camera is a low-power microscope capable of imaging the base of the eye at high resolution. Conventional basement cameras illuminate the base using a bright flash of broad-spectrum "white" light, illuminated by optics that guide and shape the light, and by an imaging system that receives the light reflected from the eye. The imaging sensor of a conventional color basement camera is designed to use a combination of three color channels (red, green, and blue) to resolve retinal structures (spatial details). These color channels are tuned to mimic the sensitivity of the three visual pigments in human cone photoreceptors. As a result, the images captured by this camera consist of a combination of these color channels and are remarkably similar to those directly observed by the human eye during clinical examinations of the retina using an ophthalmoscope or slit lamp and a handheld basement lens.

[0004] For specific spatial features (including anatomical structures and disease characteristics), the spectral characteristics of the light source affect the sensitivity of a given substrate camera. For example, using a blue-shifted light source or a blue-green filter can provide a "red-free" image to highlight retinal vessels and major internal hemorrhages. This demonstrates that altering the spectral composition of the illumination light in substrate imaging can provide clinically useful information that is not available via conventional color substrate radiography.

[0005] Hyperspectral eye base cameras share many features with conventional base cameras, but they have the added ability to record the intensity of light reflected from the eye at many (typically more than 20) discrete wavelengths (spectral sub-intervals). Therefore, hyperspectral base cameras offer both high spatial and spectral resolution. Multispectral base cameras also acquire images of light at different wavelengths, but these images are fewer in number, have unequal spectral widths, and may overlap.

[0006] Multispectral and hyperspectral cameras are defined as spatially and spectrally resolved images acquired using substrate cameras that can detect biomarkers that are undetectable using conventional color substrate cameras. Hyperspectral generally refers to an imaging system that acquires a series of images using a narrow range of non-overlapping or minimally overlapping, isosampling light bands within the dominant wavelength range. The term multispectral imaging is typically used for imaging systems that acquire a series of images using a smaller number (usually between 3 and 15) of bands within the dominant wavelength range. These multispectral bands often overlap and have unequal bandwidths. In the context of hyperspectral and / or multispectral substrate imaging of the eye, the dominant wavelength range typically spans the visible and near-infrared wavelengths or portions thereof.

[0007] While hyperspectral and multispectral cameras provide more spectral information than traditional color substrate cameras, the main advantage of hyperspectral cameras is that, due to their narrow imaging bands, each band is independent of the next, making them more suitable for visualization and analysis. However, hyperspectral cameras are more complex to design than multispectral cameras.

[0008] Multispectral / Hyperspectral Substrate Cameras: Several solutions have been proposed for multispectral and hyperspectral imaging. Pixel scanning is a method that records individual spectral pixels and moves the sample across two spatial dimensions. An alternative to two-dimensional scanning is pushbroom acquisition. In pushbroom imaging, a narrow strip of the substrate is imaged, dispersing the spectral components of the reflected light onto the sensor in a direction perpendicular to the scan axis. While this method is widely used in remote sensing, it is more difficult to apply to substrate imaging because the satellite's movement is the scanning direction, and the rapid movement of the eye (scanning) requires complex post-processing alignment.

[0009] Snapshot-based hyperspectral cameras use extended Bayer arrays or optical techniques on the sensor to record spectral and spatial dimensions in a single frame, segmenting different spectral bands into different parts of the camera sensor. This technique offers high acquisition rates, but at the cost of spectral and spatial resolution. Currently, the only suitable solution for multispectral or hyperspectral imaging without compromising spatial resolution is scanning along the spectral dimensions. However, rapid eye movement during this scanning process can affect the measurements.

[0010] The challenge of optical power: For patient acceptance, basal imaging needs to be performed within seconds. For non-mydriatic imaging (without pharmacological pupillary dilation assistance), this interval is reduced to 300 milliseconds or less, as this corresponds to the delay in pupillary light reflection, which subsequently impairs image quality due to pupillary constriction. Achieving high-quality images across multiple wavelengths within short frames is technically challenging because basal cameras are extremely inefficient (a small amount of light from the illumination source reaches the base of the eye, and only a fraction of that light is reflected from the base to the camera sensor).

[0011] Any discussion of the background art throughout the specification should not be construed as an admission that such art is well-known or constitutes part of the common knowledge in the field. Summary of the Invention

[0012] Noting the above limitations, the inventors have identified the need for improved equipment and methods to acquire non-mydriatic hyperspectral images of the eye base.

[0013] In particular, the inventors have determined that achieving good chromatic aberration compensation with a conventional color camera is complex because all photons of different wavelengths arrive at the sensor simultaneously. This can typically only be corrected by using expensive lens systems designed to correct those aberrations to a certain extent. However, as the camera's objectives become more complex, it becomes increasingly difficult to compensate for some of the camera's internal back reflections (stray light).

[0014] The following describes a device for acquiring substrate images with high spectral and spatial resolution using the full resolution of a camera sensor. The spectral band can be continuously tuned within the spectral range of interest (e.g., 420 nm to 760 nm) using a linearly variable bandpass filter.

[0015] Furthermore, a method is described for synchronizing the movement of the linear variable bandpass filter with illumination power and image acquisition to achieve a high SNR across the entire spectral range of interest. The device described herein is fast enough for high-quality non-mydriatic substrate image acquisition (image capture within 300 ms). In another embodiment, a method / system for spectral information recovery is described, wherein images acquired using at least partially overlapping bands with variable spectral widths are processed to obtain an accurate representation of the hyperspectral curves as if they were acquired using narrow, independent, and isosampled bands. In another embodiment, a method / system for reducing chromatic aberration in recorded images is disclosed. Some embodiments of this disclosure can be implemented with low-cost components, including a broadband diffuse light source (light-emitting diode, LED) and a variable bandpass filter.

[0016] One embodiment of this disclosure provides an apparatus capable of providing spectrally and spatially resolved images of the base of a subject's eye under non-mydriatic conditions. The apparatus includes an optical lens assembly (referred to as an illumination path) that projects light from a spectrally tunable light source onto the subject's base. This assembly also shapes the light onto a torus in the subject's pupillary plane to minimize reflections from the cornea. The tunable light source is configured to produce light with a variety of spectral profiles. The spectral profiles do not need to be monochromatic or limited to the minimum bandwidth required to resolve the spectral features of interest, nor do they have high out-of-band suppression, as these parameters are compensated for after acquisition using the spectral information recovery methods disclosed herein. The apparatus also includes imaging optics to project light reflected from the subject's base onto a camera sensor. The camera is synchronized with illumination power and a variable bandpass filter position to improve the signal-to-noise ratio (SNR). The apparatus also includes a gaze alignment system to assist the subject's fixation. The gaze alignment system can be turned off during image capture so that it does not contribute to the recorded image.

[0017] In another embodiment, a spectral information recovery method / system is provided, wherein images acquired using at least partially overlapping bands with potentially variable spectral widths are processed to obtain an accurate representation of hyperspectral curves acquired using narrow, independent, and isosampled bands.

[0018] According to a first aspect of the present invention, an ocular base imaging device is provided, comprising:

[0019] The lighting module has the following features:

[0020] One or more light sources are configured to produce light at wavelengths within a desired spectral range;

[0021] A first optical component for shaping and directing the light onto the subject's eye; and

[0022] A tunable bandpass filter for selecting wavelength sub-spacing within the desired spectral range; and

[0023] The imaging module has:

[0024] A second optical component is used to collect light returning from the subject's eye and project the light returning from the eye onto an image sensor. The second optical component includes one or more optical elements capable of compensating for changes in eye position.

[0025] An image sensor is configured to image the returned light to generate an image of the eye base at the wavelength sub-intervals;

[0026] The tunable bandpass filter and the image sensor are synchronized to capture images at different wavelength sub-intervals within the desired spectral range.

[0027] In some embodiments, the tunable bandpass filter is tunable between an infrared wavelength range and a blue wavelength range. In some embodiments, the tunable bandpass filter is configured to tune from the infrared wavelength range to the blue wavelength range, such that the image sensor captures one or more first images in the infrared wavelength range and subsequently captures one or more second images in the visible wavelength range.

[0028] In some embodiments, the tunable bandpass filter is configured to be tuned at a preset speed and a preset step size.

[0029] In some embodiments, the power of the one or more light sources is controlled to provide a preset power level for each spectral sub-spacing.

[0030] In some embodiments, a preset power level is selected for each spectral sub-space to compensate for spectral unevenness caused by one or more of the illumination module and / or the imaging module.

[0031] In some embodiments, the power of the one or more light sources is controlled to achieve a threshold signal-to-noise ratio for the tissue being imaged.

[0032] In some embodiments, the power of the one or more light sources is controlled to obtain a target digital count value on the image sensor for a reference surface. In some embodiments, the reference surface is derived from the retinal reflectance of the overall sample.

[0033] In some embodiments, the power of the one or more light sources is controlled to compensate for the optical absorption of the illumination module and / or imaging module.

[0034] In some embodiments, the power of the one or more light sources is controlled based on the sensitivity of the imaging sensor.

[0035] In some embodiments, the second optical component includes a focusing lens subsystem having one or more focusing lenses that are movable along an optical axis, wherein the axial movement of the one or more focusing lenses is synchronized with the wavelength filter movement of the tunable bandpass filter to provide improved focusing at the image sensor for each of a plurality of spectral sub-spacings to compensate for chromatic aberration.

[0036] In some embodiments, the focusing lens movement is non-linear with respect to the wavelength tuning of the tunable bandpass filter. In some embodiments, the focusing lens movement is quadratic with respect to the wavelength tuning of the tunable bandpass filter.

[0037] In some embodiments, the one or more light sources are LEDs having a spectral bandwidth covering at least the range from 450 nm to 720 nm.

[0038] In some embodiments, the tunable bandpass filter has a spectral bandwidth greater than the step size between the wavelength sub-intervals.

[0039] In some embodiments, the tunable bandpass filter is a linear variable bandpass filter.

[0040] In some embodiments, the illumination module includes a toroidal surface disposed after the tunable bandpass filter for shaping the light at the pupillary plane of the subject's eye.

[0041] In some embodiments, the device includes an optical diffuser disposed between the tunable bandpass filter and the toroidal surface.

[0042] In some embodiments, the device includes a homogenizing rod disposed between the tunable bandpass filter and the toroidal surface. In some embodiments, the optical diffuser is integral with or attached to the homogenizing rod.

[0043] In some embodiments, the illumination module includes a black dot optical mask configured to reduce light reflected back to the image sensor.

[0044] In some embodiments, the one or more light sources include a first LED with output power in the infrared wavelength range and a second LED with output power in the visible range.

[0045] In some embodiments, the device includes a perforated reflector disposed at the junction of the illumination module and the imaging module, the perforated reflector including an external reflective region for reflecting light from the illumination module to the eye and a central aperture for transmitting light returning from the eye to the imaging module.

[0046] The device according to any one of the preceding claims, wherein the image is captured under non-mydriatic imaging conditions.

[0047] According to a second aspect of the invention, a method is provided for compensating for the inhomogeneity of images recorded by an ocular basal imaging device, the basal imaging device comprising a tunable bandpass filter for selecting wavelength sub-intervals within a desired spectral range and a focusing lens capable of being moved along the optical axis of the device, the method comprising the steps of:

[0048] A baseline image is recorded at predetermined wavelength sub-intervals and at a predetermined focusing lens position to image internal reflections from components within the device, wherein the baseline image is an image captured by the device at the target imaging position without eye and external light;

[0049] The device captures images of the eye's base at the predetermined wavelength sub-intervals and at the predetermined focusing lens position to generate a raw image of the base; and

[0050] The baseline image is subtracted from the original image to generate a first corrected image to compensate for the internal reflections of the device.

[0051] In some embodiments, the method includes the step of recording a baseline image of multiple wavelength sub-spaces and the position of the focusing lens.

[0052] In some embodiments, the method includes the following steps:

[0053] An eye model is placed at the target imaging location of the device, the eye model comprising a retinal surface with known reflectivity;

[0054] Images of the eye model are recorded at predetermined wavelength sub-intervals and at predetermined focusing lens positions to generate a reference image;

[0055] Subtracting the first corrected image from the reference image to generate a compensation image to compensate for the intensity inhomogeneity of the device; and

[0056] Divide the first corrected image by the compensated image to obtain the second corrected image.

[0057] In some embodiments, the method includes the step of recording images of the eye model for multiple wavelength sub-spacings and focusing lens positions to generate multiple reference images.

[0058] According to a third aspect of the invention, a method is provided for recovering spectral information of a sample from multiple (K) independent spectral measurements performed under K spectral illumination curves, wherein at least two of the spectral illumination curves at least partially overlap, the method comprising the steps of:

[0059] Determine the spectral filter curve for each of the k spectral illumination curves;

[0060] A filter matrix that fills the K filter curves;

[0061] Invert the filter matrix to generate an inverse filter matrix; and

[0062] The K spectral measurements are multiplied by the inverse filter matrix to calculate the spectral information of the sample.

[0063] In some embodiments, the K spectral illumination curves are generated based on the light source spectral curves passing through filters tuned across K filter positions.

[0064] In some embodiments, the k spectral illumination curves are within the spectral band of 450 nm to 750 nm.

[0065] In some embodiments, the method includes the following steps:

[0066] The calibration matrix is ​​filled with information from one or more of the image sensor sensitivity, the light source illumination spectrum, and the optical system spectral transfer function; and

[0067] Multiply the K spectral measurements by the calibration matrix.

[0068] In some embodiments, the spectral illumination curve is measured or estimated in the P>=K band. In some embodiments, the K spectral measurements are multiplied by the calibration matrix to produce a P-dimensional vector that needs to be downsampled. In some embodiments, the downsampling is performed using a uniform Gaussian downsampling matrix.

[0069] According to a fourth aspect of the invention, a method is provided for recovering spectral information of a sample from a plurality (K) of hyperspectral or multispectral images comprising the sample, wherein each image is captured under illumination by one or more light sources at different wavelength sub-intervals in a desired spectral range by moving a tunable filter between k filter positions and recording the corresponding image at a digital image sensor, the method comprising:

[0070] Determine the spectral response of the image sensor at each of the K filter positions;

[0071] Determine the filter transmission spectrum at each of the K filter positions;

[0072] Determine the spectral curve of the light source at each of the K filter positions;

[0073] Determine a calibration matrix representing the combination of the image sensor's spectral response, the filter's transmission spectrum, and the light source's spectral curve;

[0074] The K images of the sample are multiplied by the inverse of the calibration matrix to calculate the spectral information of the sample. Attached Figure Description

[0075] Exemplary embodiments of this disclosure will now be described by way of example only, with reference to the accompanying drawings, in which:

[0076] Figure 1 This is a schematic diagram of a retinal imaging device used to image the base of a subject's eye;

[0077] Figure 2A This is a front view of a perforated mirror;

[0078] Figure 2B This is a side view showing a perforated mirror that reflects the incident illumination beam and transmits the returning beam.

[0079] Figure 3 This is a close-up schematic diagram of a linear variable filter, diffuser, and homogenizer; and

[0080] Figure 4 This is a flowchart illustrating the main steps in a method for compensating for the heterogeneity of images recorded by a retinal imaging device;

[0081] Figure 5 The graphs showing the illumination spectrum, sensor spectral sensitivity, and target sample spectrum as a function of wavelength in the spectral range of 450 nm to 700 nm are presented.

[0082] Figure 6A The graph shows the filter response and the corresponding detected image sensor intensity of a spectral filter with a passband centered at 490 nm.

[0083] Figure 6B The graph shows the filter response and the corresponding detected image sensor intensity of a spectral filter with a passband centered at 590 nm.

[0084] Figure 6C The graph shows the filter response and the corresponding detected image sensor intensity of a spectral filter with a passband centered at 640 nm.

[0085] Figure 7 It is an intensity versus wavelength graph, showing the actual intensity curve (straight line) of the desired spectral characteristics, the image sensor readings (squares) for measurements captured at multiple locations on the desired spectral band, and the recovered spectral readings (circles) obtained after applying the spectral recovery method;

[0086] Figure 8 This is a schematic diagram of a portion of the illumination module of a retinal imaging device according to a second embodiment;

[0087] Figure 9 This is a schematic plan view of a portion of the illumination module of a retinal imaging device according to a third embodiment;

[0088] Figure 10 yes Figure 9 A schematic perspective view of the third embodiment of the retinal imaging device; and

[0089] Figure 11 This is a schematic diagram of a retinal imaging device according to the fourth embodiment. Detailed Implementation

[0090] System Overview

[0091] First refer to Figure 1 This image illustrates an ocular base imaging device 100 for imaging the base of a subject's eye 102. Ocular base imaging systems such as device 100 are configured to detect and monitor various diseases of the eye and body, including diseases of the central nervous system and circulatory system. The base camera is a low-power microscope designed to illuminate the subject's base and simultaneously image the light reflected from the subject's base. Device 100 is generally divided into an illumination module 140 (composed of...). Figure 1 The larger dashed line in the middle indicates) and the imaging module 141 (made by Figure 1 (The smaller dashed line indicates this).

[0092] Light is generated by one or more light sources 103 and 104 (e.g., LEDs), which project light with a desired spectral range into the forward illumination path. Preferably, the desired spectral range covers at least from 450 nm to 720 nm, and optionally covers some infrared wavelengths (e.g., 850 nm) used for camera / eye alignment. The power generated by light sources 103 and 104 is controlled such that the total power incident on the eye 102 is within safe levels according to International Organization for Standardization (ISO) / American National Standards Institute (ANSI) standards. Preferably, one light source is configured to illuminate in the infrared range, and another light source is configured to illuminate as white light in a broadband spectrum in the visible range. Illumination in the infrared range can be used to perform initial alignment of the instrument with the subject, and illumination in the visible range can be used for standard imaging of the substrate. Although in Figure 1 Two light sources are shown, but it should be understood that a single light source or more than two light sources can be combined into device 100 to span different spectral ranges of interest.

[0093] Light from light sources 103 and 104 is at least partially collimated by respective collimating lenses 105 and 106 and combined using beam splitter 107. Collimating lenses 105 and 106 are positioned relative to their respective light sources 103 and 104 to optimize collimation. The collimating lenses should have high numerical apertures and be positioned at distances close to their focal lengths from light sources 103 and 104. In the illustrated embodiment, light sources 103 and 104 are arranged perpendicular to each other, and beam splitter 107 is oriented at 45 degrees toward the optical axes of light sources 103 and 104. In this configuration, a portion of the light from light source 103 is guided through beam splitter 107 and combined with a portion of the light from light source 104 reflected from beam splitter 107. As an example, the beam splitter may be a 50 / 50 beam splitter, configured to reflect 50% of the incident light and the remaining 50% of the incident light. However, it should be understood that other configurations of beam splitters and / or other optical elements can be implemented to combine light from more than one light source. In some embodiments, beam splitter 107 may include a glass plate or a removable reflector (actuated by an actuator connected to a controller).

[0094] At the output of beam splitter 107, the light from sources 103 and 104 is spatially combined and propagates together in a collimated or partially collimated manner. Focusing lens 108 focuses the combined collimated beam onto a tunable bandpass filter in the form of a linear variable bandpass filter 109. The spot size of the combined collimated beam is preferably as small as possible so that most of the partially collimated beam can pass through lens 108. Thus, focusing lens 108 is preferably a high numerical aperture lens. Filter 109 is preferably located at a distance from focusing lens 108, close to the focal length of focusing lens 108. To facilitate optical aberration correction, focusing lens 108 preferably has the same or similar characteristics and dimensions as collimating lenses 105 and 106. The passband of filter 109 is selectively controlled by controller 150 to filter the incident light, producing a filtered beam. The operation of filter 109 is described below. In other embodiments, other types of tunable bandpass filters may be used instead of linear variable bandpass filter 109.

[0095] After passing through filter 109, the light does not tend towards spectral homogeneity in one or more spatial dimensions. For example, if filter 109 is configured to have a passband centered at 525 nm with a bandwidth of 50 nm, the left portion of the light will contain more power at 500 nm, while the right portion will contain more power at 550 nm. The filtered beam in the illumination path passes through a homogenizing rod and a diffuser 115, which shapes the filtered beam into a sharp output beam of a predetermined size with a more uniform energy distribution across the beam. Light passing through the homogenizing rod is internally reflected and exits the rehomogenized rod. A diffuser is positioned at the end of the rod to facilitate this homogenization process. At the end of the diffuser 115, the filtered beam is not collimated but is essentially diffused.

[0096] The homogenized filtered beam is then shaped by a toroidal surface 110 located at or near the end of the diffuser 115 to produce a toroidal surface of light. The size of the toroidal surface 110 depends on the optics in the illumination path, but the inner / outer diameter of the toroidal surface 110 is proportional to the desired toroidal shape to be formed at the pupil plane for imaging the eye 102.

[0097] After passing through the toroidal surface 110, the homogenized and filtered light passes through a series of lenses (e.g., relay lenses 111a and 111b) and at least one field stop 112, creating a toroidal surface of a predetermined size at the pupillary plane of the eye 102, and ensuring that the light on the subject's substrate is relatively homogeneous across the desired field of view. The field stop 112 includes a circular aperture of a predetermined diameter, allowing light of the appropriate field of view to enter the eye 102. A black dot 125 can be inserted after the field stop 112 to perform a masking effect. The black dot is an optical masking element comprising a central black dot surrounded by a transparent region. The black dot is located at the optical conjugate position of the objective lens 122 described below and reduces light reflected back from the center of the objective lens 122 onto the image sensor 113. This helps to avoid saturation of the image sensor 113. The size of the black dot 125 needs to be large enough to remove unwanted reflections, but not too large to block too much light that needs to be on the substrate of the eye 102.

[0098] Preferably, the image sensor 113 images only light from the substrate of the eye 102. Any other light from the cornea, iris, lens, or inside the system is considered stray light. Additional internal black dots and light shields can be added to remove additional stray light from the system.

[0099] The toroidal surface 110 is imaged on the reflecting surface of the perforated mirror 120 described below and projected toward the objective lens 122.

[0100] In the imaging path (the light return path), a series of lenses (e.g., lenses 114 and 118) and an aperture stop at least partially defined by a perforated mirror 120 are used to project light reflected from the subject's substrate onto the imaging sensor 113.

[0101] refer to Figure 2A The perforated mirror 120 includes a plane mirror 121 having a centrally located aperture 123. The perforated mirror 120 is positioned along an optical path such that, in a forward illumination direction, a ring of light is reflected from the outer portion of the mirror toward the eye 102. Figure 2B The solid line in the diagram illustrates this. In the direction of returning to the image, light reflected from the base of the eye 102 passes through the central aperture of the perforated mirror 120, as... Figure 2B As shown by the dashed lines. In this way, the perforated mirror 120 facilitates the coupling of the forward illumination path and the return imaging path. (As shown...) Figure 1 As shown, the perforated mirror 120 is preferably tilted to guide the incident and returned light along different trajectories. In other embodiments, this coupling can be achieved by using a beam splitter instead of the perforated mirror 120.

[0102] The surface of mirror 120 reflects light from the illumination path toward the eye, and light leaving the eye 102 in the imaging path passes through an aperture in mirror 120. Objective lens 122, positioned between mirror 120 and the eye 102, is used by both the illumination and imaging paths. In the imaging path, an aerial image of the substrate is formed behind objective lens 122.

[0103] An aperture stop located behind or inside the perforated mirror 120 limits the amount of light directed towards the sensor 113. The aperture stop is ideally positioned as a conjugate to the pupil of the eye 102, such that it allows rays to exit the substrate only from within the illumination torus. The spatial image is re-imaged on the sensor 113 using a series of lenses 114 and 118.

[0104] like Figure 1 As shown, the perforated mirror 120, objective lens 122 and eye 102 share the illumination module 140 and imaging module 141.

[0105] Imaging sensor 113 is used to capture an image of eye 102 at a time synchronized with the passband wavelength of filter 109 centered on a preset wavelength. This corresponds to the preset filter position of the tunable filter element as described below. In some embodiments, a single controller 150 is used to synchronously control the filter passband of filter 109 and the sensor integration time or shutter period of image sensor 113. However, it should be understood that more than one controller may be implemented within device 100. Controller 150 and / or other controllers may be implemented as a digital processor, integrated circuit, microcontroller, system-on-a-chip, or other conventional hardware having appropriate hardware drivers mounted as hardware and / or software, and associated memory for storing data.

[0106] A focusing / zoom lens 118 can be used to adjust defocus caused by variations in axial length and eye refraction errors, refocusing the light to provide a clear image on sensor 113. Lens 118 is mounted to move linearly axially via an actuator configured to adjust the position of lens 118 along the optical axis. Focusing lenses 114 and 116 can also move along with the movement of lens 118. Control of the movement of lenses 114, 116, and 118 can be performed via corresponding control signals from controller 150.

[0107] A gaze alignment / gaze target 124, such as an LED or Liquid Crystal Display (LCD) screen, is positioned conjugate to the subject's base and is used to facilitate gaze alignment, thereby facilitating eye position, when the subject views the target 124. For gaze alignment, the subject is instructed to view and gaze at the target 124 through the objective lens 122. Illumination using infrared light can be used for alignment to avoid distracting the subject's attention. The position / orientation of the device 100 is then controlled in three dimensions to align with the eye 102. The axial position of the focusing lens 118 is then adjusted so that the eye focuses for subsequent imaging. The alignment process can be repeated each time a different area of ​​the eye 102 is to be imaged. A beam splitter 126, a flip mirror, or a transparent glass or plastic sheet can be used to integrate the gaze alignment system with the rest of the device. The gaze alignment / gaze target is preferably activated during the initial alignment and calibration routine and subsequently deactivated during normal image acquisition to avoid unwanted reflections on the image sensor 113.

[0108] A power meter 128 (or spectrometer / spectrometer) can be used to measure light for verification or calibration during acquisition. A beam splitter 130, positioned between relay lens 111 and field stop 112, can be used to tap off a portion of the collimated and filtered optical ring to integrate the power meter 128 with the rest of the device 100. Folding mirrors 132 and 134 can be used to reduce the spatial area of ​​the device 100 when integrated into the device. One or more additional folding mirrors can be implemented to change the overall spatial shape of the device, such as... Figure 11 As shown in the alternative embodiment, the power meter 128 can be located anywhere in the illumination path after the filter 109, but the closer it is to the field stop 112, the more likely the light is to be spatially homogeneous and comparable to the light reaching the substrate.

[0109] As described above, one or more light sources can be used in combination to span different spectral ranges of interest. For non-pupil imaging, a tunable filter 109 is used to perform gaze alignment and focusing; this tunable filter 109 is configured to allow infrared light to pass through or out of the optical path to avoid pupil constriction. Preferably, one light source is configured to illuminate in the infrared range for alignment, while another light source illuminates broadband white light in the visible range for actual imaging of the eye 102.

[0110] Device 100 is capable of non-mydriatic hyperspectral ocular substrate imaging using a combination of one or more high-power LEDs 103 and 104, a linearly variable bandpass filter 109, and a light mixing component to homogenize the light. For example, since light from each side of the tunable bandpass filter 109 will leave the filter with a spectral gradient, the system for mixing the light needs to ensure homogeneous illumination of the substrate.

[0111] In one embodiment of device 100, device 100 may use an aperture of 2 mm or larger, thereby allowing a greater amount of light to be used for imaging eye 102. In other embodiments, an aperture smaller than 2 mm is used to provide a narrower wavelength band, the side effect of which is lower power compared to an aperture of 2 mm or larger. The required illumination power density can be achieved with a spectral band as narrow as 50 nm. Therefore, sufficient power is available for high-quality non-mydriatic imaging with an acquisition time of 300 ms or less.

[0112] Wavelength selection

[0113] Now for reference Figure 3A close-up of a device 100 centered on a linearly variable bandpass filter 109 (e.g., a Delta optical continuously variable filter) is shown. The bandpass filter 109 is positioned at the focal point of a focusing lens 108. The bandpass filter 109 is preferably a combination of a low-pass filter 109a and a high-pass filter 109b. Each filter 109a and 109b is formed by an optical element having spectral characteristics that vary linearly along its length. The light beam passes through the low-pass and high-pass filters and exits the filters with a spectral shape or wavelength sub-interval corresponding to the bandpass filter. The width and center frequency of the spectral sub-interval are defined by a linear offset between the two filters. In another embodiment, the low-pass and high-pass filters are mounted optically in parallel, each having its own light source. The high-pass and low-pass filters are then recombined using a beam splitter before passing through the focusing lens 108.

[0114] like Figure 3 As shown, the passband width of filter 109 is defined by the relative displacement of the two filters 109a and 109b in the lateral dimension perpendicular to the optical axis. The tunability of filter 109 at different center wavelengths is achieved by simultaneously moving the two filters 109a and 109b together in the lateral dimension. Control of filter 109 tuning is performed by controller 150 or a separate controller. Filter 109 can be mounted on a translational mount with linear actuators, allowing the position of filter 109 to be gradually increased. As an example, filter 109 can be controlled to move vertically downwards to tune filter 109 at the center wavelength. In this way, the common movement of the cascaded filter elements directly translates into a change in the wavelength sub-spacing of interest (center wavelength tuning).

[0115] The tuning of filter 109 is performed such that the step-defined wavelength sub-interval of the filter positions of filter elements 109a and 109b occurs in time synchronized with the integration time of image sensor 113. The filter can be configured to be tuned continuously or in steps. When the filter position is controlled to move continuously, the wavelength sub-interval will vary slightly during the integration time of image sensor 113 (image frame). This means that from the start to the end of integration at image sensor 113, the center wavelength of filter 109 shifts slightly, and therefore the overall wavelength sub-interval is slightly larger than an equivalent fixed interval.

[0116] Because the initial non-collimated light source is difficult to focus precisely, the bandwidth is very large (typically >20 nm). To minimize light loss, the linear displacement of the two filters 109a and 109b is controlled such that this linear displacement is close to the diameter of the focused light source produced by the focusing lens 108. In another embodiment, this part of the system 100 can also be implemented by separating the filters and adding another focusing position, or by using one or more cylindrical lenses to focus the light onto only a single axis, potentially resulting in a slightly narrower bandwidth for a given power.

[0117] Although the transmission system using lenses has been described and illustrated, it is understood that some or all of the various lenses in device 100 may be replaced by equivalent optical elements (e.g., mirrors or prisms).

[0118] Dynamic power compensation

[0119] Most broadband light sources are not perfectly flat spectrally, and the additional effects of the transmission, reflectivity, and sensitivity of different optical elements in the device reduce the SNR for each recorded band. Because LEDs respond very quickly to power changes, this system is configured to compensate for spectral flatness by dynamically adjusting the LED power for each spectral band as needed. Spectral non-flatness can be attributed to characteristics of the imaged eye, such as retinal reflectivity, and / or to characteristics of the device 100.

[0120] In these embodiments, controller 150 or another controller is fed a set of calibration data to compensate for this spectral unevenness. Specifically, controller 150 can store or access data corresponding to the relative power levels of light sources 103 and 104 controlled at different wavelength sub-intervals. As an example, controller 150 can store or access the following lookup table:

[0121] Wavelength sub-space Compensation factor LED drive current 450nm 1.0 20A 500nm 0.6 12A 550nm 0.3 6A 600nm 0.3 6A 650nm 1.2 24A

[0122] Table 1

[0123] Table 1 is for illustrative purposes only; in practice, more wavelength sub-spacings are typically used.

[0124] Using the exemplary data in Table 1, when filter 109 is transmitting wavelength sub-intervals of 500 nm, 550 nm, and 600 nm, controller 150 is configured to reduce the drive current of one or two light sources to reduce their output power. Conversely, when filter 109 is transmitting a wavelength sub-interval of 650 nm, controller 150 is configured to increase the drive current of one or two light sources to increase their output power. This output power control is dynamic because it occurs dynamically as filter 109 selectively scans the center wavelength across different wavelength sub-intervals within the desired spectral range.

[0125] Using the dynamic power control described above, the power of one or more light sources can be dynamically modulated according to the filter wavelength, providing a preset spectral power for each of the spectral sub-spacings. This can be used to compensate for spectral unevenness (e.g., optical absorption of the illumination and / or imaging modules) caused by one or more of the illumination and / or imaging modules. The power of one or more light sources can also be modulated to achieve a threshold SNR for the imaged tissue.

[0126] The power of one or more light sources can be modulated based on the sensitivity of the imaging sensor. The power of one or more light sources can be modulated to obtain a target digital count value on a reference surface on the image sensor. The reference surface can be derived from retinal reflectance, which is measured from a population sample of people (e.g., the average retinal reflectance of 1000 people).

[0127] In addition to compensating for spectral unevenness, device 100 can also compensate for spatial aberrations caused by the characteristics of device 100 and the imaged eye. This can be achieved through one or more calibration routines executed by controller 150, as described below.

[0128] Now for reference Figure 4 This paper illustrates a method 400 for compensating for the non-homogeneity of an image recorded by a substrate imaging device (e.g., device 100). In step 401, a baseline image is recorded with predetermined wavelength sub-intervals and predetermined focusing lens positions to image internal reflections from components within device 100. Here, an optimal focal point is found in the IR to compensate for eye refraction errors, and then a predefined shift of the focal position relative to that initial position is applied. The baseline image is an image captured by the device without the eye 102 and with external light blocked at the target imaging position. This can be achieved by covering the objective lens 122 or placing device 100 in a darkroom or dark cover / container, or by placing a beam trap with a dark-absorbing surface at the focal plane of the torus (where the eye 102 should be).

[0129] Step 401 can be repeated for any or all different wavelength sub-spacings and any or all different focusing lens positions available in device 100. A baseline image is generated for each of the different wavelength sub-spacings and focusing lens positions. The baseline images can be stored in a database or memory accessible to controller 150, or stored on an external computer or in the cloud.

[0130] In step 402, the device captures an image of the substrate of the eye 102 at one of the same predetermined wavelength sub-intervals and at the same or similar predetermined focusing lens position used in step 401 to generate an original image of the substrate.

[0131] In step 403, a baseline image corresponding to the same wavelength sub-interval and focusing lens position is subtracted from the original image to generate a first corrected image to compensate for internal reflections of the device. The first corrected image may be stored in a database or memory accessible to the controller 150, or stored in an external computer or in the cloud.

[0132] Steps 402 and 403 can be repeated for multiple wavelength sub-intervals and focusing lens positions corresponding to the normal hyperspectral scan of the eye 102 by the device. The first corrected image can be stored and used to calibrate future original images taken by the device 100 at different wavelength sub-intervals and different focusing lens positions. The first corrected image at least partially compensates for internal reflections of the internal components of the device 100 that may have a greater effect on some wavelengths than on others, or be more prevalent at certain locations of the focusing lens 108.

[0133] Method 400 may optionally include further steps 404 to 407. Step 404 includes placing an eye model at a target imaging location of device 100. The eye model includes a retinal surface with known reflectivity. In step 405, device 100 records images of the eye model at one or more predetermined wavelength sub-intervals and at predetermined focusing lens positions to generate corresponding reference images. These reference images may be referred to as white images because the retinal surface can be modeled as a white surface.

[0134] In step 406, a first correction image is subtracted from a first reference image with corresponding wavelength sub-spacing and focusing lens position to generate a compensation image. Next, in step 507, the first correction image is divided by the compensation image to obtain a second correction image that at least partially compensates for the intensity inhomogeneity of the device.

[0135] Steps 405 to 407 can be performed for multiple wavelength sub-spacings and the position of the focusing lens that captures the original substrate image.

[0136] The process is described in more detail below regarding the spectral information recovery method.

[0137] In another embodiment of the device, the camera's integration time is dynamically adjusted to compensate for variations in the SNR ratio between spectral bands.

[0138] In some embodiments, a custom filter in the optical path can be used to flatten or shape the spectral curve. This can be used in place of or in combination with the dynamic spectral power control described above.

[0139] In some embodiments, using the dynamic power compensation method described above, the device 100 is able to image the substrate with spatial field flatness within a 30% deviation.

[0140] Color difference correction

[0141] The focal length of a lens varies with wavelength (which causes axial chromatic aberration). Therefore, with conventional lenses, longer wavelengths focus at longer distances compared to shorter wavelengths. One solution to axial chromatic aberration is to construct lenses made of different types of glass; however, this is expensive and complex and can lead to problematic back reflections, thus degrading image quality. Embodiments of the present invention provide an alternative solution that compensates for chromatic aberration by adjusting the position of one or more imaging lenses 114 and 118 for each wavelength band during image acquisition.

[0142] Refer again Figure 1 In one embodiment of the device including a simple objective lens 122 (e.g., a single glass element), the focusing lens 108 moves linearly along the optical axis in sync with the band selection of the relative lateral movement of the filter 109 to dynamically compensate for chromatic aberration. In another embodiment of the device including an achromatic objective lens 122 (e.g., having two glass elements), the focusing lens 108 moves secondaryly in sync with the band selection of the filter 109 to compensate for axial chromatic aberration. (Compared to the movement required for refractive error compensation) chromatic aberration requires only minor adjustments to the focusing lens 108.

[0143] The movement for color correction relative to the corrected refractive error position of a given eye.

[0144] Spectral information recovery methods

[0145] The device 100 is capable of illuminating a substrate using a spectral profile with wavelength sub-spacings for each given location of a linearly variable bandpass filter 109. Each wavelength sub-spacing can have a bandwidth wider than the step size between the spacings (and therefore partially overlap), which is not optimal for conventional substrate hyperspectral imaging. A method for effectively compensating for this spectral overlap and recovering information is described below, as if illumination were using a narrow, non-overlapping bandwidth.

[0146] The key requirements for this compensation are: (1) the ability to determine or measure the spectral profile of the illumination light at each position of filter 109 or the filter 109 used for imaging; and (2) the measured spectral profiles for each position of filter 109 being linearly independent of each other. By design, device 100 ensures that these requirements are met. Specifically, (1) complete control of the illumination and synchronization with the filter position enables accurate measurement of these parameters, and (2) the transition from one position of filter 109 to the next ensures the independence of the measured spectral bands, as no two spectral bands will completely overlap.

[0147] Although device 100 and filter 109 have been described, it will be understood that this spectral information recovery method can be performed with other systems and different tunable filters.

[0148] Figure 5 Data related to the spectral recovery method is shown. The top panel shows the illumination spectrum, which can correspond to the spectrum of the LEDs 103 or 104 used to illuminate the eye 102. The middle panel shows the sensor spectral sensitivity, which can represent the wavelength sensitivity or spectral response of the image sensor 113. The bottom panel shows the ideal substrate (or retina) reflectance spectrum to be extracted.

[0149] Figure 6A The filter response or spectral transmission is shown when the passband center of the filter (e.g., filter 109) is located at 490 nm. Figure 6A The location of the filter reaching the image sensor is also shown, and integration is performed based on the sensor's spectral sensitivity to produce a corresponding intensity spectrum with a single digital count. Figure 6B and Figure 6C The data are shown when the filter passband center is at 590nm and 640nm, respectively.

[0150] For all desired positions k of the tunable filter, the following is obtained: Figure 7 The spectral readings are shown in the square. However, due to the overlap of spectral bands, the information recovered at this stage does not directly correspond to the actual spectral characteristics of the substrate, such as... Figure 7 The lower straight line curve is shown. By applying the spectral information recovery method described below, the target spectral properties of the substrate (such as reflectance) can be accurately recovered, such as... Figure 7 As shown by the black circle in the image, it can be seen that the target spectral characteristics match the true spectral characteristics of the substrate very well.

[0151] The digital counts collected from the sensor each time using a tunable filter can be written as follows:

[0152]

[0153] Where dc is the digital count, k∈[1,K] is the filter position number, f(λ) is the filter spectral transmittance at position k, s(λ) is the sensor sensitivity, l(λ) is the light source spectral curve, and t ill (λ) is the spectral transmittance of the lighting system, t im (λ) is the spectral transmittance of the imaging system, and r(λ) is the substrate reflectance representing the useful information sampled across K bands. Other terms for transmittance and reflectance of other optical components may also be considered without loss of generality.

[0154] This continuous-scale integral can approximate any desired level of accuracy of the sensor value as the sum of P discrete wavelengths, such that:

[0155]

[0156] At this point, assume that P is at least as large as the number of independent bands K to be recovered. This summation can be rewritten as a matrix equation such that:

[0157] DC=F·diag(s)·diag(l)·diag(t ill )·diag(t im )·r

[0158] Where DC is a K-dimensional vector containing the digital counts measured on the sensor for each filter location, F is a K×P matrix containing the spectral transmittance at each filter location, and s,l,t ill ,t im These are P-dimensional vectors representing sensor sensitivity, light source spectral curve, illumination system spectral transmittance, and imaging system spectral transmittance, respectively. The operator `diag(·)` places the vector elements into a diagonal matrix of corresponding dimension P×P. Vector `r` is a P-dimensional reflection vector that is expected to approximate in dimension K, such that:

[0159]

[0160] Where D is a user-defined downsampling matrix of dimension K×P. For example, if the first row of the downsampling matrix is ​​d1 = [1 / 3, 1 / 3, 1 / 3, 0, ..., 0], it will take the uniformly weighted average of the first three elements of r and place them in... The first element is in the matrix. A uniform or Gaussian distribution is an example of such a downsampling matrix.

[0161] Since only a K-dimensional vector DC can be accessed, it is necessary to find a calibration matrix C of dimension K×K such that:

[0162]

[0163] Substituting the above formula for DC, we obtain:

[0164]

[0165] Alternative get:

[0166] D·r=C·F·diag(s)·diag(l)·diag(t ill )·diag(t im )·r

[0167] Reduced to D = C·F·diag(s)·diag(l)·diag(t) ill )·diag(t im )

[0168] Assuming the elements of the diagonal matrix are non-zero, then:

[0169] D·diag(t im ) -1 ·diag(t ill ) -1 ·diag(l) -1 ·diag(s) -1 =C·F

[0170] Multiplying the right side by the transpose of matrix F yields:

[0171] D·diag(t im ) -1 ·diag(t ill ) -1 ·diag(l) -1 ·diag(s) -1 ·F T =C·F·F T

[0172] As mentioned earlier, suppose matrix F contains K linearly independent rows, therefore matrix F·F T It is reversible. Multiply the right side by (F·F) T ) -1 The following formula is given for the calibration matrix C:

[0173] C = D·diag(t) im ) -1 ·diag(t ill ) -1 ·diag(l) -1 ·diag(s) -1 ·F T ·(F·F T ) -1

[0174] In summary, the reflectance value can be estimated using the following formula based on the measured digital count (DC):

[0175]

[0176] Some parts of the diagonal matrix may be impossible to measure or estimate. In such cases, ignoring one or more parts of the diagonal matrix will lead to... The bias estimation. Assuming a surface with known reflectivity can be imaged under the same conditions as the retina, the bias can be recovered as follows. Unbiased estimation:

[0177]

[0178] The term r can be derived from its list of values ​​by first resampling the wavelength to the wavelength used by the filter matrix F, and then multiplying it by the downsampling matrix F. known In the case of a white reference, r known For all values ​​of 1, and which can be omitted, this yields a simplified equation:

[0179]

[0180] This represents the estimated retinal base reflection that can be recovered from a series of partially spectral overlapping images obtained from device 100.

[0181] The spectral information recovery method includes the following steps:

[0182] a) Determine the filter transmission spectrum at K locations of filter 109. This may include obtaining actual measurements of the illumination spectrum or determining its approximation.

[0183] b) Optionally determine or measure other spectral characteristics of the system, such as sensor sensitivity, light source spectrum, and optical transmission.

[0184] c) Optionally, the spectrum from step a) or b) above may be resampled at P more nearby wavelengths.

[0185] d) Fill the downsampling matrix D;

[0186] e) Using the filter transmission spectrum with K spectral sub-intervals and optionally other measured or derived spectra, calculate the calibration matrix C (K×K). As mentioned above, this filter transmission matrix should be in good working order for most practical applications, and its right pseudo-inverse exists. Other P-dimensional matrices are diagonal, and these matrices can also be inverted if no entries on the diagonal are zero. For the purposes of the claims, the right pseudo-inverse operation or other type of matrix inverse will be referred to as the inverse operation. The calibration matrix may also include information about the light source spectrum of the illumination and imaging system, the wavelength sensitivity of the image sensor, and the spectral transmission.

[0187] f) Record hyperspectral images under the same acquisition conditions as the calibration matrix. The hyperspectral images comprise multiple (K) images of light returning from the subject's eye, wherein at least one image is captured at each of the K locations of filter 109. Each filter location corresponds to a wavelength sub-spacer with a finite bandwidth and may partially overlap with one or more adjacent sub-spacers. That is, the bandwidth of the spectral sub-spacer may be wider than the spacing between the center wavelengths of adjacent sub-spacers.

[0188] g) Register K images so that corresponding pixels in the images are aligned with the same spatial regions of the eye.

[0189] h) Convert the (K-dimensional) digital count DC of each pixel (i,j) in the hyperspectral image into a (K-dimensional) reflectance value r1 by multiplying the DC vector by the calibration matrix C, as follows:

[0190] r1(i,j)=C×DC(i,j)

[0191] To compensate for illumination homogeneity and eliminate internal camera reflections, the same acquisition settings (filter position and focusing lens position) can be used to image both the white target (W) and the baseline target (BL). In this case, the equation for recovering the reflection data R² is as follows:

[0192] r W (i,j)=C×DC W (i,j)

[0193] rr BL (i,j)=C×DC BL (i,j)

[0194] r2(i,j)=(r1(i,j)-r BL (i,j)) / (r W (i,j)-r BL (i,j))

[0195] The image acquisition and spectral information recovery process can be summarized as follows:

[0196] 1) Measure the transmission spectrum at each location of the filter using a spectroradiometer (or otherwise estimate the illumination spectrum);

[0197] 2) [Optional] Resample the measured spectrum to the desired band;

[0198] 3) [Optional] Use a spectroradiometer to measure (or estimate) the source spectrum, image sensor wavelength sensitivity, and spectral transmittance of the illumination and imaging system;

[0199] 4) Fill the downsampling matrix D;

[0200] 5) Calculate the calibration matrix D;

[0201] 6) [Optional] Record digital count (DC) images of an artificial eye (typically white) with known reflectivity that replaces the retina in each spectral interval;

[0202] 7) [Optional] Record a DC image of a well-absorbing black surface located at the pupil plane to record the light reflected back from within the system in each spectral interval;

[0203] 8) [Optional] Record dark current images;

[0204] 9) Record the DC image of the eye at each spectral interval;

[0205] 10) Multiply each spectral pixel of each DC image by the calibration matrix C to obtain the corresponding hyperspectral reflectance image; and

[0206] 10) [Optional] Use a white image, a baseline image, and a dark image to correct the reflection of the eye image;

[0207] The steps for acquiring hyperspectral images of the subject's eyes can be summarized as follows:

[0208] 1) The subject is positioned in front of the device;

[0209] 2) Focus on the target (ON);

[0210] 3) Activate the main light source with a filter at the infrared (IR) position (almost invisible to the patient);

[0211] 4) Align the camera to provide a proper visualization of the patient's basal body;

[0212] 5) Focusing and tuning to improve image resolution;

[0213] 6) Target OFF;

[0214] 7) Start acquiring sequences (modulation power, spectral filter movement, focus movement, camera frame recording);

[0215] 8) Complete image acquisition and save the image; and

[0216] 9) [Optional] Acquire another image at a different location on the base of the same eye (as the target moves), or acquire an image of another eye.

[0217] Other embodiments

[0218] Now for reference Figures 8 to 11 An alternative embodiment of device 100 is shown.

[0219] Figure 8A portion of the illumination module of device 800 is shown, including four LEDs 802-805, representing light sources for illuminating the eye. Device 800 operates in a manner similar to that of device 100 described above. However, including four light sources allows for greater spectral and power control within the desired spectral range. As an example, LEDs 802-805 may have different powers at different peak wavelengths and / or may be controlled to be driven at different power levels. One LED can be used to illuminate the eye in the infrared range for system alignment, while the remaining three LEDs can be used for illumination in the visible range for standard image capture.

[0220] In device 800, LED 802 is positioned perpendicular to LEDs 803-805, and each LED has a corresponding collimating lens 807-810. To combine the beams from each of LEDs 802-805, three beam splitters 812-814 are arranged in the optical path. Beam splitters 807-809 combine the beams so that they collectively pass through focusing lens 108 to the rest of the optical system. For example, beam splitters 807-809 could be 50:50 beam splitters.

[0221] Figure 9 and Figure 10 A portion of the illumination module of device 900 is shown, comprising an array of four LEDs 902-905, representing a light source for illuminating the eye. LEDs 902-905 are positioned to guide light through corresponding lenses 907-910 and onto a parabolic reflector 912. The parabolic reflector 912 sequentially combines the light from each LED 902-905 and directs the combined light onto a focusing lens 108 and a filter 109. An advantage of device 900 is the avoidance of beam splitters at the input path, which increases the overall optical power through the optical system.

[0222] Figure 11 An alternative device 1100 is shown, having an optical path different from that of device 100 described above. Folding mirrors 902-905 are used to fold the light beam around the optical path to objective lens 122 and eye (not shown). Example central and peripheral rays are shown in... Figure 11 The line shown is a dashed line.

[0223] explain

[0224] The spectral widths (such as “bandwidth,” “wavelength sub-spacing,” and “passband”) mentioned in this specification are intended to refer to standard measurements, such as full width at half maximum (FWHM) measurements. FWHM measurements define the spectral width as the width of the spectral peak whose amplitude is equal to half its maximum value.

[0225] In this specification, the term "base" is used to refer to a portion of the eye that includes at least the retina and optionally other parts such as the optic disc, retinal vessels, retinal pigment epithelium, and choroid. Base images are intended to include at least an image of the retina, and optionally information about these other ocular regions.

[0226] Unless otherwise expressly stated, it will be apparent from the following discussion that, throughout the discussion of this specification, the use of terms such as “processing,” “calculation,” “operation,” “determine,” “analysis,” etc., refers to the actions and / or processes of a computer or computing system or similar electronic computing device that manipulate and / or convert data expressed as physical quantities (e.g., electronic quantities) into other data similarly expressed as physical quantities.

[0227] In a similar manner, the terms "controller" or "processor" can refer to any device or part of a device that processes electronic data, such as from registers and / or memory, to convert that electronic data into other electronic data, such as data that can be stored in registers and / or memory. "Computer," "computing machine," or "computing platform" can include one or more processors.

[0228] In this specification, references to "one embodiment," "some embodiments," or "embodiment" mean that a particular feature, structure, or characteristic described in connection with an embodiment is included in at least one embodiment of this disclosure. Therefore, the appearance of the phrases "in one embodiment," "in some embodiments," or "in an embodiment" throughout this specification does not necessarily refer to the same embodiment. Furthermore, in one or more embodiments, as will be apparent to those skilled in the art from this disclosure, particular features, structures, or characteristics can be combined in any suitable manner.

[0229] As used herein, unless otherwise specified, the ordinal adjectives “first,” “second,” “third,” etc., used to describe common objects merely indicate different instances of similar objects being referenced and are not intended to imply that the objects described in this way must be in a given order in time, space, rank, or any other way.

[0230] In the following claims and the description herein, any of the terms “comprising,” “comprised of,” or “which comprises” are open-ended terms that mean at least the subsequently listed element / feature, but do not exclude other elements / features. Therefore, when used in the claims, the term “comprising” should not be construed as limited to the means, elements, or steps listed thereafter. For example, the scope of an expression for a device comprising A and B should not be limited to a device consisting only of elements A and B. Any of the terms “including,” “which includes,” or “that includes” as used herein are also open-ended terms that also mean at least the element / feature following that term, but do not exclude other elements / features. Therefore, “comprising” and “including” are synonyms, meaning to include.

[0231] It should be understood that in the foregoing description of exemplary embodiments of this disclosure, various features of this disclosure are sometimes combined in a single embodiment, figure, or description thereof to simplify this disclosure and aid in understanding one or more of the various aspects of the invention. However, the approach of this disclosure should not be construed as reflecting an intention to claim more features than are expressly stated in each claim. Rather, as reflected in the following claims, aspects of the invention do not lie in all features of a single foregoing disclosed embodiment. Therefore, the claims following the detailed description are expressly incorporated herein by reference, each claim being a separate embodiment of this disclosure in itself.

[0232] Furthermore, as those skilled in the art will understand, while some embodiments described herein include some but not others of features included in other embodiments, combinations of features from different embodiments are intended to be within the scope of this disclosure and form different embodiments. For example, any of the claimed embodiments in the following claims may be used in any combination.

[0233] Numerous specific details are set forth in the description provided herein. However, it should be understood that embodiments of this disclosure can be practiced without these specific details. In other instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this description.

[0234] Similarly, it should be noted that when used in the claims, the term "coupled" should not be construed as limited to a direct connection. The terms "coupled" and "connected," and their derivatives, may be used. It should be understood that these terms are not intended to be synonyms. Therefore, the scope of expressing "device A coupled to device B" should not be limited to devices or systems where the output of device A is directly connected to the input of device B. This implies that there is a path between the output of A and the input of B, which may include other means or tools. "Coupled" may mean two or more elements in direct physical, electrical, or optical contact, or two or more elements that are not in direct contact with each other but still cooperate or interact with each other.

[0235] The embodiments described herein are intended to cover any modifications or variations of the invention. Although the invention has been described and explained with reference to specific exemplary embodiments, those skilled in the art will recognize that additional embodiments within the scope of the invention can be readily conceived.

Claims

1. A non-mydriatic ocular base imaging device, comprising: The lighting module has the following features: One or more light sources are configured to produce light at wavelengths within a desired spectral range; A first optical component is used to shape and guide the light onto the subject's eye; as well as A tunable bandpass filter for selecting wavelength sub-spacings within the desired spectral range; The imaging module has: A second optical component is used to collect light returning from the subject's eye and project the light returning from the eye onto an image sensor. The second optical component includes one or more optical elements capable of compensating for changes in the eye. as well as An image sensor is configured to image the returned light to generate a non-mydriatic image of the base of the eye at the wavelength sub-intervals; as well as One or more controllers are configured as follows: The tunable bandpass filter is tuned between multiple wavelength sub-spacings within the desired spectral range; The image sensor is controlled to generate multiple non-mydriatic images of the ocular base at each of the plurality of wavelength sub-intervals; as well as The power of one or more light sources is dynamically controlled to provide a preset power level for each of the plurality of wavelength sub-spaces; The tunable bandpass filter and the image sensor are synchronized by one or more controllers to capture images at different wavelength sub-intervals within the desired spectral range; and Multiple non-mydriatic images of the eye base are captured within 300 milliseconds.

2. The apparatus of claim 1, wherein, The tunable bandpass filter is tunable between the infrared wavelength range and the blue wavelength range.

3. The apparatus of claim 2, wherein, The tunable bandpass filter is configured to tune from the infrared wavelength range to the blue wavelength range, such that the image sensor captures one or more first images in the infrared wavelength range and subsequently captures one or more second images in the visible wavelength range.

4. The apparatus of claim 2 or claim 3, wherein, The tunable bandpass filter is configured to be tuned at a preset speed and a preset step size.

5. The apparatus of any one of claims 1-3, wherein, Each spectral sub-space is selected with a preset power level to compensate for spectral unevenness caused by one or more of the illumination module and / or the imaging module.

6. The apparatus of any one of claims 1-3, wherein, The power of the one or more light sources is controlled to achieve a threshold signal-to-noise ratio for the tissue being imaged.

7. The apparatus of any one of claims 1-3, wherein, The power of one or more light sources is controlled to obtain a target digital count value on the image sensor for a reference surface.

8. The apparatus of claim 7, wherein, The reference surface is derived from the retinal reflectance of the overall sample population.

9. The apparatus of any one of claims 1-3 and 8, wherein, Control the power of the one or more light sources to compensate for the optical absorption of the illumination module and / or imaging module.

10. The apparatus of any one of claims 1-3 and 8, wherein, The power of one or more light sources is controlled based on the sensitivity of the image sensor.

11. The apparatus of any one of claims 1-3 and 8, wherein, The second optical component includes a focusing lens subsystem having one or more focusing lenses that are movable along the optical axis, wherein the axial movement of the one or more focusing lenses is synchronized with the wavelength filter movement of the tunable bandpass filter to provide improved focusing at the image sensor for each of a plurality of spectral sub-spacings to compensate for chromatic aberration.

12. The apparatus of any one of claims 1-3 and 8, wherein, The second optical component includes a focusing lens subsystem having one or more focusing lenses that are movable along the optical axis, wherein the movement of the focusing lenses is non-linear with respect to the wavelength tuning of the tunable bandpass filter.

13. The apparatus of any one of claims 1-3 and 8, wherein, The second optical component includes a focusing lens subsystem having one or more focusing lenses that are movable along the optical axis, wherein the movement of the focusing lenses is quadratic relative to the wavelength tuning of the tunable bandpass filter.

14. The apparatus of any one of claims 1-3 and 8, wherein, The one or more light sources are LEDs having a spectral bandwidth covering at least the range from 450 nm to 720 nm.

15. The apparatus of any one of claims 1-3 and 8, wherein, The tunable bandpass filter has a spectral bandwidth greater than the step size between the wavelength sub-intervals.

16. The apparatus of any one of claims 1-3 and 8, wherein, The tunable bandpass filter is a linear variable bandpass filter.

17. The apparatus of any one of claims 1-3 and 8, wherein, The illumination module includes a toroidal surface disposed after the tunable bandpass filter for shaping the light at the pupillary plane of the subject's eye.

18. The device of claim 17, further comprising an optical diffuser disposed between the tunable bandpass filter and the toroidal surface.

19. The device of claim 18, comprising a homogeneous rod disposed between the tunable bandpass filter and the toroidal surface.

20. The apparatus of claim 19, wherein, The optical diffuser is integrated with or attached to the homogenizing rod.

21. The apparatus of any one of claims 1-3, 8, and 18-20, wherein, The illumination module includes a black dot optical mask configured to reduce light reflected back to the image sensor.

22. The apparatus of any one of claims 1-3, 8, and 18-20, wherein, The one or more light sources have a spectral bandwidth covering at least from 450 nm to 720 nm, and include a first LED with output power in the infrared wavelength range and a second LED with output power in the visible wavelength range.

23. The device according to any one of claims 1-3, 8 and 18-20, comprising a perforated reflector disposed at the junction of the illumination module and the imaging module, the perforated reflector comprising an external reflecting region for reflecting light from the illumination module to the eye and a central aperture for transmitting light returning from the eye to the imaging module.

Images