Systems and methods for evaluating ocular conditions

Abstract

A system for evaluating ocular conditions includes an imaging device having at least one lens, a processor, and memory; a hollow elongate member adapted to be positioned around a patient's eye and having a base portion adapted to attach to the imaging device so that the second end is positioned around the lens of the imaging device; and one or more light sources affixed to an inner surface of the hollow elongate member and positioned to illuminate the patient's eye. The memory stores code which when executed by the processor causes the processor to perform: capturing video or at least one image of a patient's eye, which has been instilled with a vital dye, while said one or more light sources illuminate the patient's eye; and detecting an ocular condition based at least in part on a vital dye pattern of the captured video or image.

Description

SYSTEMS AND METHODS FOR EVALUATING OCULAR CONDITIONSCROSS-REFERENCE TO RELATED APPLICATION(S)

[0001] The present application claims priority to U.S. Provisional Patent Application No. 63 / 586,321 , filed September 28, 2023, which is hereby incorporated by reference in its entirety.TECHNICAL FIELD

[0002] The present disclosure relates to systems and methods for evaluating ocular conditions and, in particular, assessing tear film breakup and corneal staining patterns based at least in part on captured images and / or video of the patient’s eye.BACKGROUND

[0003] One of the inherent problems in assessing new treatment modalities for dry eye is the lack of one specific sign or symptom that characterizes the disease. A constellation of ocular surface endpoints such as symptom questionnaires, tear film measurements, surface staining patterns and redness must be graded with accurate and sensitive scales to assess the efficacy of treatment.

[0004] Patients with dry eye disease experience greater tear film instability and more rapid rates of tear film break-up post blinks. Blink may function as a compensator for a dysfunctional and unstable tear film, further worsening the negative consequences on visual function. Conventionally, tear film stability is evaluated by an ophthalmologist during an eye examination using a slit lamp, which combines a microscope with a high intensity slit light source.

[0005] Corneal staining is a diagnostic sign that may indicate damage to the corneal surface due to various factors like dry eye syndrome, contact lens wear, allergies, ortrauma. Corneal staining refers to the presence of irregularities, disruptions, or damage to the outermost layer of the cornea, the corneal epithelium, which can be visualized using a dye like fluorescein, Lissamine Green, or Rose Bengal, during an eye examination.

[0006] Keratitis, on the other hand, is an inflammation or infection of the cornea itself. While corneal staining can be a sign of keratitis, it is not synonymous with the condition. As noted above, corneal staining is a diagnostic sign that reveals disruptions or damage to the corneal epithelium, whereas keratitis is a broader term referring to inflammation or infection of the cornea. Corneal staining can be observed in cases of keratitis, but it can also be seen in other ocular surface disorders that do not involve inflammation or infection of the cornea.

[0007] Keratitis is generally classified into two main types: infectious keratitis and non- infectious keratitis. Infectious keratitis is caused by microorganisms such as bacteria, viruses, fungi, or parasites. Non-infectious keratitis is caused by factors such as injury, chemical exposure, ultraviolet (UV) light, dry eye, autoimmune disorders, or ocular allergies. Some examples of non-infectious keratitis include: exposure keratitis, which occurs when the cornea is inadequately protected due to an inability to close the eyelids fully or issues with tear production; neurotrophic keratitis, which results from impaired corneal sensitivity or innervation, which can be due to various factors such as diabetes, herpes zoster, or surgical procedures, and Superficial Punctate Keratitis (SPK).

[0008] SPK is a condition characterized by inflammation and damage to the outermost layer of the cornea, the corneal epithelium, resulting in the appearance of tiny, discrete areas of damage. SPK can be caused by various factors, including dryeye, exposure to ultraviolet light, viral infections, contact lens wear, or ocular allergies.To diagnose superficial punctate keratitis, an eye care professional will typically perform a slit lamp examination after dye has been instilled onto the surface of the eye.BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Fig. 1 A depicts a system for assessing a patient’s tear film stability, according to disclosed embodiments;

[0010] Fig. 1 B1 depicts a perspective view of the eye cone, according to disclosed embodiments.

[0011] Fig. 1 B2 depicts a case attachable to an imaging device, the case having a receptacle for attaching the eye cone, according to disclosed embodiments.

[0012] Fig. 1 B3 depicts a perspective view of the eye cone from a base end, according to disclosed embodiments.

[0013] Fig. 1 C depicts a cutaway view of the eye cone, including interior components thereof, according to disclosed embodiments.

[0014] Figs. 1 D1 and 1 D2 are views of a diffusion ring, dome, and filter for use as interior components of the eye cone, according to disclosed embodiments.

[0015] Figs. 1 D3 and 1 D4 are views of embodiments of the illumination assembly as seen from the patient end of the eye cone, according to disclosed embodiments.

[0016] Fig. 1 D5 is a view of an annular printed circuit board configured to be positioned in an interior portion of the diffusion ring of the illumination assembly, according to disclosed embodiments.

[0017] Fig. 2 is a diagram depicting multiple instances of the system for evaluating ocular conditions, connected via a network to a remote computer system, according to disclosed embodiments;

[0018] Figs. 3A and 3B are diagrams of methods for evaluating ocular conditions, according to disclosed embodiments;

[0019] Fig. 4 is a diagram of a method for evaluating ocular conditions by determining an ocular protection index, according to disclosed embodiments;

[0020] Fig. 5 is an image, provided to simulate an image captured using the system, showing an example of an eye having a tear film breakup pattern with an inferior vertical streak.

[0021] Fig. 6 is an image, provided to simulate an image captured using the system, showing an example of an eye having a tear film breakup pattern with inferior streaks.

[0022] Fig. 7 is an image, provided to simulate an image captured using the system, showing an example of an eye having inferior SPK corneal staining.

[0023] Fig. 8 is an image, provided to simulate an image captured using the system, showing an example of an eye having corneal SPK.

[0024] Fig. 9 shows an example image in which an area of the inferior cornea is stained with dots which are characteristic of SPK.

[0025] Fig. 10 shows that the stained area of Fig. 9 may be geometrically delineated as a region of interest.

[0026] Fig. 11 is an image, provided to simulate an image captured using the system, showing an example of an eye having an inferior abrasion.

[0027] Fig. 12 is an image, provided to simulate an image captured using the system, showing an example of an eye having inferior confluent staining.

[0028] Where considered appropriate, reference numerals may be repeated among the drawings to indicate corresponding or analogous elements. Moreover, some of the blocks depicted in the drawings may be combined into a single function.DETAILED DESCRIPTION

[0029] In the following detailed description, numerous specific details are set forth in order to provide a thorough understanding of embodiments of the invention.However, it will be understood by those of ordinary skill in the art that the embodiments of the present invention may be practiced without these specific details. In other instances, well-known methods, procedures, components, and circuits have not been described in detail so as not to obscure the present invention.

[0030] Disclosed embodiments are directed to using an imaging system, and associated methods, to measure tear film characteristics, such as, for example, interblink interval (IBI), tear film breakup time (TFBUT), tear film break up area Ocular Protection Index (OPI and OPI 2.0), tear film break up patterns in a real-world dry eye disease population.

[0031] Blink is a contributing factor to dry eye and one of the few objective endpoints available that incorporates both signs and symptoms. The average blink rate is about 8.0 blinks / minute and the average IBI is about 7.5 seconds. To ensure that the ocular surface remains protected, the TFBUT should match or exceed the IBI. The larger the interval between tear film breakup and the following blink, the greater potential for damage to the ocular surface. An Ocular Protection Index (OPI) may be defined by TFBUT divided by IBI. An OPI < 1.0 indicates breakup tear film andexposure of the cornea before blinking, whereas an OP I score > 1 .0 indicates a protected surface, with breakup occurring after blink time. OPI 2.0 is the relationship of tear film break up area of time.

[0032] Disclosed embodiments provide systems and methods to capture keratitis and corneal staining due to dry eye or dry eye related conditions, such as neurotrophic keratitis, or any other condition that creates corneal staining, such as corneal wounds and abrasions, meibomian gland disease, irritation, allergic diseases, environmental exposure related dryness, as well as other keratopathies.

[0033] Figure 1A depicts a system 100 for evaluating a patient’s ocular condition.The system 100 includes an imaging device 110, which may be, for example, a mobile device (e.g., a smartphone belonging to, or supplied to, a patient) or similar electronic device. In embodiments, the imaging device 110 runs software, e.g., an app, to perform various aspects of the disclosed methods and captures images and video of the patient’s eye (or eyes) for use in evaluating ocular conditions, as discussed in further detail below.

[0034] The imaging device 110 has a lens (or lenses) 115, which may be positioned on a back surface, near a top corner, of the imaging device 110. The imaging device 110 includes a processor 120 and memory 125, as well as various other internal components. The memory 125 stores code which is executed by the processor 120 to perform methods disclosed herein. The processor 120 and memory 125 of the imaging device 110, in conjunction with other internal components, also perform the typical functions of the imaging device 110. For example, in the case of a mobile device, the processor may execute code stored in the memory to perform functionssuch as telephone and data communication in a mobile network and the capturing of images and video.

[0035] The system 100 further includes an illumination and support component, referred to herein as an eye cone 128, for facilitating the capture of images and video of the patient’s eye (or eyes). The eye cone 128 illuminates the patient’s eye (or eyes) during capture of images and video and also helps keep the imaging device 110 in the proper position relative to the patient’s eye (or eyes) to help maintain the quality of the captured images and video.

[0036] In embodiments, the eye cone 128 comprises a hollow elongate member 130 having a first end 134 adapted to be positioned around a patient’s eye (not shown) and a second end 136 having a base portion 138 adapted to attach to the imaging device 110 so that the second end 136 is positioned around the lens 115 of the imaging device 110. In embodiments, the eye cone 128 may include a magnification lens (not shown) positioned in front of the lens 115 of the imaging device 110 such that the combined focal length equals the distance from the lenses to the eye. In embodiments, the hollow elongate member 130 may be, for example, substantially frustoconical and may have a substantially circular or substantially elliptical crosssection, among other possible shapes. It is noted that the hollow elongate member 130 (and, more generally, the illumination and support component) may be described in colloquial terms as being “conical” or being a “cone” or part of a cone. However, such terms should not be considered to limit the scope of the disclosed embodiments. In implementations, the hollow elongate member 130 may incorporate components in an interior portion thereof, such as illumination components and / or optical components (e.g., waveguides, lenses, filters, etc.), as discussed below.Thus, the term “hollow” is meant to broadly describe the shape of the hollow elongate member 130 itself.

[0037] In embodiments, there may be a receptacle (see Fig. 1 B2) on the imaging device 110 into which the base portion 138 can be inserted (e.g., a receptacle formed in a case attached to the imaging device 110). In embodiments, the hollow elongate member 130 has a larger circumference at the first end 134 than at the second end 136 and may have a cross-section that is substantially circular or elliptical. The appropriate dimensions for the ends of the hollow elongate member 130 may be determined based, at least in part, on specifications for various mobile devices usable as the imaging device 110 and anatomical reference dimensions for facial features.

[0038] In embodiments, the hollow elongate member 130 may have a length of between about 4 cm and about 16 cm. The length of the hollow elongate member 130 may be determined based on a defined focal distance between the imaging device 110 and the patient’s eye. The hollow elongate member 130 may be formed of plastic and may be produced using various techniques, e.g., by three-dimensional printing. The hollow elongate member 130 may be rigid in the portion extending from the second end 136 along the length of the hollow elongate member 130 and may have an elastic portion or component 135 (e.g., an attached elastic element) at the first end 134, where the elastic portion or component 135 is adapted to be pressed against an area surrounding the patient’s eye. In embodiments, the entire hollow elongate member 130 may be substantially uniform in rigidity.

[0039] One or more light sources, e.g., light emitting diodes (LEDs) 140, are provided in an interior portion of the elongate member 130, e.g., affixed to an inner surface132, or positioned around an inner circumference, of the hollow elongate member 130, and positioned to illuminate the patient’s eye. The LEDs 140 may be formed on a carrier 150, such as, for example, a flexible substrate or annular printed circuit board (PCB) (see Fig. 1A and Fig. 1 D5, respectively). In embodiments, there may be an array of between 24 and 48 LEDs. In embodiments, the LEDs 140 may have an emission wavelength of about 450 nm to about 500 nm. In embodiments, the LEDs 140 may produce a total output of between about 50 lumens and about 1000 lumens.

[0040] The eye cone 128 may include a battery 145 (e.g., a rechargeable battery) connected to power the LEDs 140. Alternatively, or in addition, a power input (not shown) may be provided to accept power from an alternating current (AC) adapter to power the LEDs 140 and / or recharge the battery 145.

[0041] Figs. 1 B1 depicts a perspective view of the eye cone 200. Fig. 1 B2 depicts a case 210 attachable to an imaging device 110 (see Fig. 1 A). Fig. 1 B3 depicts a perspective view of an embodiment of the eye cone from a base end. The case 210 is designed to attach to the imaging device 110 and may also serve to protect the imaging device 110 as with a typical case for a mobile device. The case 210 may be formed of plastic and may be produced using various techniques, e.g., by three- dimensional printing.

[0042] The eye cone 200 comprises a hollow elongate member 230 having a base portion 238 adapted to attach to the imaging device 110. In the example depicted, the base portion 238 has a number of protrusions 212, e.g., four, which extend outward in a direction transverse to a lengthwise direction of the eye cone 200 to facilitate attachment to the imaging device 110. In embodiments, the base portion238 is received in a receptacle 215 such that the protrusions 212 are received in corresponding slots 217 which radiate outward from a central portion. In the example depicted, there are clips 219 which allow the protrusions 212 to snap into place and be held securely. Various other attachment mechanisms may be used between the base portion 238 and the receptacle 215, e.g., magnets. In the embodiment of Fig.1 B3, the base portion 238 includes annular protrusion 239 configured to fit, e.g., with a press fit, against the receptacle 215, which may help to prevent the infiltration of ambient light.

[0043] In embodiments, the first end 234 may be shaped to allow the eye cone 200 to be positioned around the patient’s eye with an inner portion 235 that is adapted to be positioned near or on the patient’s nose and an outer portion 237 that extends further, in a length-wise direction of the eye cone 200 so as to fit a contour of the patient’s face to reduce the influx of ambient light. The first end 234 may be symmetric about a substantially horizontal axis, i.e. , an axis aligned to pass through both of the patient’s eyes, so that the eye cone 200 can be properly positioned on either eye by flipping the eye cone 200 upside down. Thus, a single eye cone 200 may be used for both of the patient’s eyes.

[0044] Fig. 1 C depicts a cutaway view of the eye cone 200, including its interior components. The eye cone 200 comprises an illumination assembly 240, which is positioned within the hollow elongate member 230. In embodiments, an interior surface of the hollow elongate member 230 may include installation elements, such as, protrusions, grooves, etc., to cooperate with similar structures on the illumination assembly 240 to secure the illumination assembly 240 in place.

[0045] The illumination assembly 240 comprises a translucent diffusion ring 245, which extends around an inner circumference of the hollow elongate member 230. The diffusion ring 245 has an approximately square cross section with an interior portion 248 which is surrounded on three sides by the walls of the diffusion ring 245. In embodiments, an inner corner edge 247 of the diffusion ring 245 may be chamfered or rounded. The interior portion 248 of the diffusion ring 245 houses illumination elements (not shown), e.g., an array of LEDs, which are diffused by the translucent properties of the diffusion ring 245. The diffusion of the illumination source helps to reduce the reflection of the discrete LED elements by the patient’s eye. This is beneficial because the reflection of individual LED elements by the eye may produce white spots which may be confused with the symptoms of conditions such as keratitis. Thus, the diffusion ring 245, i.e. , diffusion filter, turns harsh spots of light that would appear similar to SPK on the cornea into a uniform illumination of the entire cornea.

[0046] The diffusion ring 245 is adjacent to a dome 250, which is another component of the illumination assembly 240. The dome 250 has a concave surface facing the open side of the diffusion ring 245 which acts, in effect, as a reflector of the light produced by the illumination elements. The dome 250 may be formed of plastic, e.g., by 3-D printing, and may be made of the same material as the diffusion ring 245. The dome 250 may be translucent or opaque and, in particular embodiments, a reflective material may be applied to the surface of the dome. In embodiments, the dome 250 may be translucent and may act as a diffusion element. As light from the LEDs is absorbed by the dome 250, the resulting illumination will backscatter in the direction of the patient’s eye, thereby functioning as a reflective element to an extent. There isan opening at the center of the dome 250 to allow light to pass through to the imaging device 110 (see Fig. 1A). In embodiments, an annular cavity 255 may be formed behind the dome 250 to receive a filter 260, such as a Wratten filter.

[0047] Figs. 1 D1 and 1 D2 are views of a diffusion ring 245, dome 250, and filter 260 for use as interior components of the eye cone 200. In this example, the diffusion ring 245 and the dome 250 are both formed of white translucent plastic and attach together via a set of notches and tabs. The filter 260, which, in this example, is a Wratten #12 yellow filter, is received by the annular cavity 255 of the dome 250.

[0048] Figs. 1 D3 and 1 D4 are views of embodiments of the illumination assembly 240 as seen from the patient end of the eye cone 200. In these views, the outer portion of the annulus (illuminated in Fig. 1 D3) is the diffusion ring 245, which contains the LEDs, and the inner portion is the dome 250. Fig. 1 D5 is a view of an annular printed circuit board (PCB) 150 configured to be positioned in an interior portion of the diffusion ring 245 of the illumination assembly 240. In embodiments, the PCB 150 may be positioned in the diffusion ring 245 so that the LEDs face toward the dome 250.

[0049] Figure 2 depicts multiple instances of the system 100 connected via a network 180 to a remote computer system 185. As noted above, the memory 125 of the imaging device 110 stores code which is executed by the processor 120 to perform methods disclosed herein. In embodiments, this may include transmitting the captured images and video of the patient’s eye (or eyes) via the network 180 to the remote computer system 185 which performs at least part of the analysis and calculations to evaluate ocular conditions, e.g., to assess tear film stability. In embodiments, the remote computer system 185 may receive captured images andvideo of patients’ eyes from multiple instances of the system 100, each used by an individual patient, such as, for example, in a clinical trial.

[0050] Figures 3A and 3B are diagrams of methods for evaluating ocular conditions. In these methods, the patient’s eye is instilled with a vital dye. In embodiments, the vital dye may be, e.g., fluorescein (sodium fluorescein, NaFI), which is an organic compound and dye which is used as a diagnostic tool in ophthalmology. The fluorescein mixes with tear fluid and its fluorescence, i.e., fluorescence pattern or vital dye pattern, facilitates evaluation of the tear film.

[0051] Other types of vital dyes, such as Lissamine Green and Rose Bengal, may also be used. The patterns produced these dyes may be referred to as “vital dye patterns.” The term “vital dye pattern,” as used herein, includes fluorescence patterns produced using fluorescein and patterns produced using other vital dyes. Lissamine Green is a water-soluble dye and an ideal stain for the bulbar conjunctiva, since it provides better contrast than NaFI. It stains degenerative cells where there is a disruption or damage of mucin coating. Since the dye diffuses into cells that are already dead and degenerative, it is not as toxic to the cornea as Rose Bengal and produces less irritation upon instillation. Lissamine Green is also useful for staining of the cornea. Rose Bengal is used less in practice due to its stinging and toxic effect. Like Lissamine Green, it stains dead and devitalized cells, but also stains healthy cells and affects their viability. Both Lissamine Green and Rose Bengal can be better choices for assessing conjunctival staining than NaFI, because they provide better contrast against the palpebral and bulbar conjunctiva.

[0052] Conventional tear film assessment uses about 50 pL of fluorescein, which greatly exceeds average tear volume (about 6-7 pL). This methodology has beenimproved by using lower quantities of sodium fluorescein (e.g., about 1 pL to about 5 pL). This recalibrated technique affords more reproducible and physiologically relevant diagnostic reference values. Accordingly, in embodiments, the amount of the fluorescein instilled in the patient’s eye may be less than an average tear volume of the patient’s eye. In embodiments, the amount of the fluorescein instilled in the patient’s eye may be between about 1 pL and about 5 pL.

[0053] In the method 350 for evaluating ocular conditions of Fig. 3A, the patient’s eye is instilled with a vital dye (360). Video and images of the patient’s eye are captured using the system 100 described herein while the patient’s eye is illuminated by the LEDs (370). To facilitate the image and video capture, the illumination and support component, e.g., eye cone 200 (see Fig. 1 B1 ), is attached the imaging device 110, e.g., by inserting the base portion 138 into a receptacle 215 formed in a case 210 holding the imaging device 110. In embodiments, a patient may initiate the image and / or video capture using the controls of the imaging device 110 (e.g., a mobile device) or by using an app which is provided for this purpose. The LEDs 140 are activated, either by a controller or by the patient, and the patient presses the first end 134 of the hollow elongate member 130 to the eye to be analyzed. An ocular condition is detected based at least in part on a vital dye pattern of the captured images or video (380). This may include transmitting the captured images or video via a network to a remote system which performs, at least in part, the detecting of the ocular condition (385).

[0054] In the method 300 for evaluating ocular conditions of Fig. 3B, the patient’s eye is instilled with a vital dye (310). Video and images of the patient’s eye are captured using the system 100 described herein while the patient’s eye is illuminated by theLEDs (320). To facilitate the image and video capture, the illumination and support component, e.g., eye cone 200 (see Fig. 1 B1 ), is attached the imaging device 110, e.g., by inserting the base portion 138 into a receptacle 215 formed in a case 210 holding the imaging device 110. In embodiments, a patient may initiate the image and / or video capture using the controls of the imaging device 110 (e.g., a mobile device) or by using an app which is provided for this purpose. The LEDs 140 are activated, either by a controller or by the patient, and the patient presses the first end 134 of the hollow elongate member 130 to the eye to be analyzed.

[0055] In implementations relating to tear film breakup, video, for example, may be taken by the imaging device 110 (in embodiments, images may be taken in addition to, or in lieu of, video). The patient looks into the eye cone 200 and blinks normally according to instructions which have been provided (e.g., via the app). The patient may also be asked to stare without blinking, depending on the specific test and target imaging. The video capture continues for a defined time period (e.g., a time period controlled by the app). The system 100 determines a tear film breakup time for the patient’s eye based, at least in part, on the captured video (330), using methods disclosed herein. In particular embodiments, the system 100 may transmit the captured video via the network 180 to the remote computer system 185, which performs, at least in part, the determining of the tear film breakup time, patterns, or break up area for the patient’s eye (335).

[0056] Figure 4 is a diagram of a method 400 for evaluating ocular conditions in particular embodiments by determining an ocular protection index. In implementations, this method may be performed in conjunction with the method 300 for evaluating ocular conditions of Fig. 3B. In this method 400, the system 100 maydetermine a blink rate, blink pattern, determination of incomplete or partial blinks, or an interblink interval for the patient’s eye based at least in part on the captured images or video (420). This may include transmitting the captured images or video via the network 180 to the remote computer system 185, which performs, at least in part, the determining of the interblink interval for the patient’s eye (425). The system 100 may determine an ocular protection index (OPI) for the patient’s eye based at least in part on the determined interblink interval and the determined tear film breakup time (430). This may include transmitting the captured images and video via the network 180 to the remote computer system 185, which performs, at least in part, the determining of the ocular protection index for the patient’s eye (435).

[0057] Figure 5 is an image, provided to simulate an image provided to simulate an image captured using the system, showing an example of an eye having a tear film breakup pattern with an inferior vertical streak. Figure 6 shows an example of an eye having a tear film breakup pattern with a number of inferior streaks. A tear film breakup pattern with linear streaks, also known as filamentary breakup, is characterized by the tear film breaking up into thin, elongated streaks across the corneal surface. The presence of linear streaks in the tear film breakup pattern may be indicative of severe dry eye syndrome or other ocular surface disorders where the tear film is unstable and prone to rapid evaporation. This pattern can also suggest an irregular distribution of the lipid or mucin components in the tear film, which can lead to poor tear film stability and increased ocular surface discomfort.Tear Film Breakup Time and Ocular Protection Index (OPI)

[0058] The surface of the cornea is rewet with a fresh layer of fluid, forming the precorneal tear film, by each blink of the eyelids. This periodic action is necessary owing to the deterioration of this thin fluid layer between blinks. Immediately after a new tear film is formed, it undergoes a progressive overall thinning owing to evaporation and, more importantly, begins to develop localized area that thin even more rapidly than the tear layer as a whole. It is these localized regions, usually small in area, that result in the first micelles, or “dry spots,” observed after a blink. These spots appear as dark, nonfluorescent areas when fluorescein is in the tear fluid, because where there is no fluid, there is no fluorescein and hence no fluorescence.

[0059] The time from the completion of a blink to the first appearance of these dark spots is the tear film breakup time (TFBUT). TFBUTs vary from blink to blink and person to person - an individual average TFBUT under 10 seconds has traditionally been considered indicative of dry eye syndrome. However, the innovative diagnostic combination of a microdrop of sodium fluorescein (< 5 pL), a yellow Wratten filter, and a digital video capture system with an onscreen timer, has yielded more precise TFBUT values. These measurements have indicated that 5 seconds is a more appropriate cut-off point for association of TFBUT with dry eye symptoms. Of course, persistently short blink intervals can somewhat moderate the symptomatic effects of shortened TFBUTs.

[0060] A diagnostic calculation, called the ocular protection index (OPI), quantifies the relationship between blink frequency, TFBUT, and ocular surface protection. An interblink interval ( I B I) greater than the average TFBUT value in a given eyeindicates that there is a period of time (namely, the difference between IB I and TFBUT) during which the ocular epithelium is wholly or in part unprotected by the tear film. An IBI less than the average TFBUT suggests the eye is blinking frequently enough to replenish the tear film before it breaks to expose the ocular surface. To quantify this relation in a single value, one can simply divide TFBUT by IBI - any value greater than or equal to one indicates a generally protected ocular surface, while any value less than one suggests at the very least occasional exposure of the ocular epithelium.

[0061] In disclosed embodiments, the determining of the interblink interval for the patient’s eye may include detecting blink times using a machine learning model. In such a case, the machine learning model may identify a region containing the patient’s eye in each image or each frame of the captured images or video and apply blink detection to the identified region. The machine learning model may include a convolutional neural network trained on labeled databases of facial features. The interblink interval for the patient’s eye may be based on an average or a moving average of a number of determined interblink intervals.

[0062] In disclosed embodiments, the determining of the tear film breakup time for the patient’s eye comprises detecting tear film breakup using a machine learning model. For example, the machine learning model comprises a support vector machine (SVM) trained to detect dark regions in the patient’s eye indicative of where the vital dye has receded, i.e., where the tear film has broken up. The tear film breakup time for the patient’s eye may be based on an average or a moving average of a number of determined tear film breakup times.Experimental Testing

[0063] In experimental testing, 15 patients were tested (12 female; mean age 62 years old) with a reported history of dry eye disease and a tear film breakup time less than 5 seconds. Fluorescein was instilled via micropipette in both eyes 3 minutes prior to video recording. Patients held their gaze on a crosshair positioned straight ahead of opposite eye. One-minute long segments of video, with 4k resolution at 60 frames per second, were recorded on each eye using the imaging system. Patients were asked to blink normally while the video was acquired to provide a natural interblink interval (IBI) and tear film breakup time (TFBUT). Videos were processed using a custom algorithm. OPI was calculated by dividing the TFBUT by the IBI.

[0064] Video acquisition with the imaging system was successful in all patients and reviewed by graders. In all eyes of all patients, except one, the OPI was less than 1 , indicating they had an unprotected ocular surface. The average OPI was 0.090 (standard deviation: 0.027). OPI was also measured pre- and immediately postinstallation of a tear substitute as well 15, 45 and 90 minutes afterward. There was no significant different between pre-dose and any post-dose OPI measurement.

[0065] The study demonstrated the feasibility of performing real-time, home-based assessment of tear film stability and ocular surface protection using a novel technology. This proof-of-concept study suggests the usefulness of this technology in clinical trials investigating the effect of different treatment modalities for dry eye disease outside the laboratory setting.Causes of Tear Film Breakup

[0066] Although evaporative effect progressively thin the tear film and promote eventual drying and breakup, theoretical calculations indicate that the times required to thin the tear layer to dryness should be much longer than those actually observed for measured TFBUTs. Also, long before overall drying of the anterior surface of the eye occurs, the small, localized areas of drying discussed earlier are seen. Clearly, evaporation is not the sole (or even primary) cause of tear film breakup.

[0067] The supposed non-wettability of the corneal surface is thought to be an artefact; rinsing the surface with acetylcysteine returns the surface to its wettable state. The presence of glycocalyx makes the surface wettable; however, it has been shown that newly exposed epithelial cell surfaces, revealed when overlying cells desquamate, have not yet developed or fully-expressed their glycocalyx. The surface will hence be relatively non-wettable. Theoretically, even a single non-wettable cell can initiate a dry spot; it may be that the pattern of dry spots may indicate where surface cells have recently desquamated.

[0068] It is surmised that over time (i.e. , within seconds after a blink), the mucin layer on the epithelium becomes contaminated by nonpolar components of the tear film, primarily lipid from the superficial layer. This oily layer is, of course, only ~ 5 or 6 pm above the surface of the mucin layer under the best of conditions. Microscopic flow patterns, either of thermal origin or due to the turbulence of the blink action, can bring this floating lipid into contact with the mucin on the corneal surface. Although a small amount of lipid contamination can be masked by the mucin molecules, sufficiently large areas eventually become contaminated whereby the mucin can no longer act as an effective surfactant. Then, nonwetting area develop, withspontaneous thinning of the tear layer immediately above them, with eventual rupture of the tear film. When these localized nonwetting areas resist the formation of a new, clean mucin layer during blinking, persistent tear film breakup over the same area follows within a few seconds of each blink. A series of strong, forced blinks often reestablish tear film continuity over some of these areas by covering them with mucin.Tear Film Breakup Patterns

[0069] Examination of TFBUT video images has not only demonstrated the dynamic nature of tear film breakup, but has shown that the tear film may breakup in any of a number of unique patterns. The tear film breakup patterns (TFBUP) may be identified using descriptors and / or categories, such as, for example: spotting, amorphous blob, linear, fractured, and wispy. Each pattern is generally reproducible by patient, by eye. It is evident that certain patterns are considerably more prevalent in certain dry eye populations, leading to the conclusion that the pattern present upon breakup is at least partially indicative of the advancement or nature of the syndrome. TFBUPs may be modified by alteration of tear film composition, e.g., by treatment with a tear substitute, meibomian gland expression, or stimulation of reflex tearing. Thus, tear film breakup patterns show that there is more to the dispersion of the tear film than how quickly it occurs, and these patterns can be a useful visual diagnostic.

[0070] Thus, tear film breakup patterns provide additional valuable information about the stability, composition, and overall health of the tear film. The tear film consists of three layers: the inner mucin layer, the middle aqueous layer, and the outer lipidlayer. These layers work together to provide hydration, lubrication, and protection to the ocular surface. Tear film breakup refers to the process where the tear film starts to evaporate or break apart, leaving dry spots on the ocular surface. The pattern and speed at which this occurs can be used to diagnose the underlying cause of various ocular surface disorders, such as dry eye syndrome.

[0071] As an alternative to the pattern categories discussed above, there are many other ways in which the various types of tear film breakup patterns can be described, including:

[0072] (i) random (or non-specific) breakup: This pattern is characterized by the tear film breaking up in multiple areas across the corneal surface. It is often seen in patients with general dry eye syndrome and can be associated with a decrease in tear production or an increase in tear film evaporation;

[0073] (ii) dimple breakup: This pattern involves small, round dry spots scattered across the cornea. It is typically associated with localized areas of poor tear film stability, possibly due to an uneven corneal surface or irregularities in the lipid or mucin layers of the tear film;

[0074] (iii) filamentary breakup: In this pattern, the tear film breaks up into thin, linear streaks. It is often seen in patients with severe dry eye, where the tear film is unstable and prone to rapid evaporation; and

[0075] (iv) area breakup: This pattern involves large, confluent areas of tear film breakup. It is commonly associated with meibomian gland dysfunction (MGD), a condition where the lipid-producing glands in the eyelids are not functioning correctly, leading to poor tear film stability and increased evaporation.Analyzing Tear Film Breakup Time and Patterns ,and Blink Rate and Patterns, Interblink interval, and interpalpebral fissure height (distance between lid margins) Using Neural Networks Trained with Labeled Data

[0076] In implementations, machine learning techniques, including various neural network architectures, may be used to analyze Tear Film Breakup Time and patterns, and blink rate and patterns, interblink interval, and interpalpebral fissure height (distance between lid margins) from eye images (e.g., images and / or video of the patient’s eye captured using the system 100 described herein). As one example there may be a data collection stage which includes gathering a large dataset of eye images that show different tear film breakup patterns. These images may be labeled with the appropriate pattern (e.g., random, dimple, filamentary, area, or spotting, amorphous blob, linear, fractured, and wispy) to serve as ground truth during the training process.

[0077] In a data preprocessing stage, the dataset may be preprocessed by, for example, resizing the images to a consistent size, normalizing the pixel values, and augmenting the dataset with image transformations such as rotations and translations. This helps improve the neural network’s ability to generalize to new images.

[0078] There may be a model selection stage in which a particular neural network architecture is chosen which is suitable for the eye image classification tasks. For example, a Convolutional Neural Network (CNN) is particularly effective for this purpose, because it can very efficiently learn to recognize patterns and features in the images.

[0079] In a model training stage, the dataset may be split into a training set and a validation set. The training set is used to train the neural network by adjusting the model’s weights to minimize the classification error. The performance of the model on the validation set is monitored during training to avoid overfitting and to fine-tune hyperparameters, such as learning rate, batch size, and the number of training epochs.

[0080] In a model evaluation stage, after training of the neural network, its performance may be evaluated on a separate test set of eye images with known tear film breakup patterns. Relevant performance metrics may be determined, such as accuracy, precision, recall, and F1 -score, to assess the effectiveness of the model in classifying the different patterns. In a model deployment stage, the trained and validated neural network is used in the system 100 described herein to automatically analyze tear film breakup patterns. The accuracy of such a model depends on factors such as the quality and size of the dataset, the choice of the neural network architecture, and the fine-tuning of the model’s hyperparameters. It may be necessary to iteratively refine the model and dataset to achieve satisfactory performance in analyzing tear film breakup patterns from eye imaging.Analyzing Tear Film Breakup Patterns Using Neural Networks Trained with Unlabeled Data

[0081] In implementations using unlabeled training data, unsupervised learning techniques, such as clustering or autoencoders, may be used to analyze tear film breakup patterns from eye imaging. For example, one of the most effective approaches in this context is clustering-based techniques combined with deeplearning features, such as using a pre-trained CNN as a feature extractor. In such a case, there may be a data collection stage in which a large dataset of eye images is gathered which show different tear film breakup patterns. Because the data is unlabeled, the system will not have ground truth labels for the breakup patterns. In embodiments, model training may primarily involve training on unannotated data for categorization. In embodiments, a transfer learning method may be used, in which a model is trained with unrelated data and then used to train a new model to allow for a smaller dataset.

[0082] In a data preprocessing stage, the dataset may be preprocessed by resizing the images to a consistent size, normalizing the pixel values, and augmenting the dataset with image transformations such as rotations and translations. In a feature extraction stage, a pre-trained neural network, e.g., a CNN, may be used to extract high-level features from the images, such as the locations of dark (i.e. , dry) spots.

[0083] In implementations, the last fully connected layers may be removed from the pre-trained model and the remaining layers may be used as a feature extractor. Preprocessed images and / or video may be passed through the modified network to obtain a feature vector for each image. Dimensionality reduction techniques, such as Principal Component Analysis (PCA) or t-Distributed Stochastic Neighbor Embedding (t-SNE), may be used to reduce the complexity of the feature vectors and improve clustering performance.

[0084] In a model training and evaluation stage, the newly labeled data obtained by the labeling of the clusters may be used to train a supervised model like a CNN, as discussed above, for tear film breakup pattern classification. The performance of the resulting model may be evaluated using appropriate metrics such as accuracy,precision, recall, and F1 -score, as discussed above. The accuracy of this approach depends on factors such as the quality of the clustering algorithm, the features extracted, and the cluster labeling process.Evaluating Corneal Staining to Assess Keratitis, corneal and conjunctival staining, abrasions, wound area size, and Other Ocular Conditions

[0085] Figure 7 is an image, provided to simulate an image captured using the system, showing an example of an eye having inferior Superficial Punctate Keratitis (SPK) corneal staining. Figure 8 shows another example of an eye having corneal SPK. SPK is a corneal condition characterized by small, scattered, and superficial disruptions or damage to the corneal epithelium. In disclosed embodiments, the system may evaluate the staining pattern, e.g., by analyzing the distribution, size, and shape of the stained areas. SPK typically presents as multiple, discrete, and small punctate stains across the corneal surface. The pattern can help determine the underlying cause of the SPK, such as dry eye syndrome, contact lens wear, or ocular surface inflammation.

[0086] In disclosed embodiments, the evaluation of the staining pattern may include identifying the affected areas, i.e., determining which parts of the cornea are affected by the staining. SPK can be localized to specific areas or involve larger portions of the cornea. This information can help characterize the SPK, which may make it possible to narrow down the potential causes of the condition. The evaluation of the staining pattern may further include evaluating the severity and extent of the SPK, e.g., by determining the number of punctate lesions (which may be referred to as “dots”) and the degree of corneal involvement. Staining of the cornea and / orconjunctiva may be assessed, as well as corneal abrasions, size and area of abrasions and / or wounds, and staining of the lid margins.Assessment of SPK Using Algorithmic Techniques

[0087] In disclosed embodiments, the seventy of the SPK may be assessed using algorithmic techniques. In such a case, the assessment may include a selection of a region of Interest. Figure 9 shows an example image in which an area of the inferior cornea is stained with dots which are characteristic of SPK. In this example, fluorescein was used, but other vital dyes, such as Lissamine Green or Rose Bengal may also be used. Figure 10 shows that the stained area may be geometrically delineated as a region of interest. The assessment algorithm may include extracting the delineated region of interest from the remainder of the image. Due to variations in corneal diameter, eyelid shape, and other geometric parameters among subjects, the delineation of the region of interest may be done using machine learning tools, such as, for example, a support vector machine (SVM) trained to detect and delineate regions which include dots characteristic of SPK.

[0088] After isolating the region of interest, various types of image processing can be done to improve the clarity of the dots. For example, the blue and red channels of the RGB image may be discarded and analysis of the green channel may be done using a process of image normalization with background smoothing. After extraction of the green RGB image channel, image brightness may be normalized to mean pixel intensity of, e.g., 100. The image background may be smoothed to remove the effects of slowly varying image intensities due to the spherical geometry of thecorneal surface. Punctate dots may be isolated by thresholding the image by pixel intensity level of, e.g., approximately 20% above the image mean intensity.

[0089] To reduce noise in the image, the image may be converted to binary format and a blob detection algorithm may be applied to filter connected regions of white pixels below the typical size of a corneal epithelial cell (assumed to be about 36 pm). The minimum blob size in pixels may be determined by the apparent diameter of the cornea (i.e., in pixels) in the image, relative to an assumed diameter of 10 mm. The blob filtering may also used to remove flash artifacts from the image by filtering blobs above, e.g., 500 pixels. Blobs which do not fall within the geometric area of interest (e.g., the inferior cornea) may also be discarded. The remaining blobs may be counted by the algorithm to obtain the punctate dot number, Ndots, which may be used as a measure of the seventy of the SPK.Assessment of SPK and Other Ocular Conditions Using Machine Learning

[0090] In disclosed embodiments, the seventy of the SPK may be assessed using machine learning technigues, with supervised or unsupervised learning, in a manner akin to the use of machine learning in tear film breakup pattern analysis, discussed below. This may involve training a model to analyze eye images and provide a guantitative measure of the SPK severity.

[0091] For supervised learning approaches, there may be a data collection stage in which a large dataset of eye images with varying degrees of SPK severity is gathered. The images may be labeled, e.g., with seventy scores, typically on a scale ranging from mild to severe, based on expert evaluation. Alternatively, the images may be labeled guantitatively based on a determined guantity of punctuate dots. Thelabeled data serves as ground truth during the training process. In embodiments, eye images may be segmented based on labelled data using a u-net shaped deep residual convolutional neural network.

[0092] In a model training stage, the dataset may be divided into a training set and a validation set. The model may be trained by adjusting the model's weights to minimize the error between the predicted severity scores and the ground truth labels. The performance of the model on the validation set may be monitored during training to avoid overfitting and to allow for fine tuning of hyperparameters, such as learning rate, batch size, and the number of training epochs.

[0093] In a model evaluation stage, after training the machine learning model, its performance may be evaluated on a separate test set of eye images with known SPK seventy scores. Performance metrics, such as mean squared error (MSE) or mean absolute error (MAE), may be used to assess the effectiveness of the model in predicting SPK severity.

[0094] In disclosed embodiments, unsupervised learning techniques, such as clustering or autoencoders, may be used to analyze corneal staining from eye imaging in a manner similar to the analysis of unlabeled tear film breakup patterns. For example, clustering-based techniques combined with deep learning features, such as using a pre-trained CNN as a feature extractor, may be used. In such a case, there may be a data collection stage in which a large dataset of eye images is gathered which show different staining patterns associated with SPK and other ocular conditions.

[0095] Figures 11 and 12 are examples of corneal staining patterns for ocular conditions other than SPK, which may be assessed in accordance with disclosedembodiments. Fig. 11 is an image, provided to simulate an image captured using the system, showing an example of an eye having an inferior abrasion. Fig. 12 is an image, provided to simulate an image captured using the system, showing an example of an eye having inferior confluent staining. “Confluent staining” is a term used to describe a pattern of corneal staining in which multiple stained areas on the cornea merge or come together to form larger, continuous areas of staining. Confluent staining typically indicates more extensive damage or disruption to the corneal epithelium, the outermost layer of the cornea, compared to isolated or scattered staining patterns. Confluent staining can be associated with various ocular conditions, including:

[0096] Severe dry eye syndrome: A significant lack of tear production or poor tear film quality can cause widespread damage to the corneal epithelium, resulting in confluent staining.

[0097] Contact lens-related issues: Prolonged or improper contact lens wear, poor lens hygiene, or ill-fitting lenses can cause extensive mechanical irritation, hypoxia, or infection, leading to confluent staining on the corneal surface. The edge of the lens itself can also be tracked with the imaging described herein, referred to as “lens drift”.

[0098] Ocular surface infections: Bacterial, viral, or fungal infections can cause widespread damage to the corneal epithelium, resulting in confluent staining patterns.

[0099] Chemical or physical trauma: Exposure to toxic substances, chemical bums, or physical injuries to the eye can cause extensive damage to the corneal epithelium, leading to confluent staining.Conclusion

[0100] Aspects of the present invention may be embodied in the form of a system, a computer program product, or a method. Similarly, aspects of the present invention may be embodied as hardware, software, or a combination of both. Aspects of the present invention may be embodied as a computer program product saved on one or more computer-readable media in the form of computer-readable program code embodied thereon.

[0101] The computer-readable medium may be a computer-readable storage medium. A computer-readable storage medium may be, for example, an electronic, optical, magnetic, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof.

[0102] Computer program code in embodiments of the present invention may be written in any suitable programming language. The program code may execute on a single computer, or on a plurality of computers. The computer may include a processing unit in communication with a computer-usable medium, where the computer-usable medium contains a set of instructions, and where the processing unit is designed to carry out the set of instructions.

[0103] The above discussion is meant to be illustrative of the principles and various embodiments of the present invention. Numerous variations and modifications will become apparent to those skilled in the art once the above disclosure is fully appreciated. It is intended that the following claims be interpreted to embrace all such variations and modifications.

Claims

CLAIMS1 . A system for evaluating ocular conditions, comprising: an imaging device comprising at least one lens, a processor, and memory; a hollow elongate member having a first end adapted to be positioned around a patient’s eye and a second end having a base portion adapted to attach to the imaging device so that the second end is positioned around said at least one lens of the imaging device; and one or more light sources positioned to illuminate the patient’s eye, wherein the memory stores code which when executed by the processor causes the processor to perform: capturing video or at least one image of a patient’s eye, which has been instilled with a vital dye, while said one or more light sources illuminate the patient’s eye; and detecting an ocular condition based at least in part on vital dye pattern of the captured video or image.

2. The system of claim 1 , wherein the memory further stores code which when executed by the processor causes the processor to perform: transmitting the captured video or image via a network to a remote computer system which performs, at least in part, said detecting an ocular condition.

3. The system of claim 1 , wherein the memory further stores code which when executed by the processor causes the processor to perform determining a tear film breakup time based at least in part on the vital dye pattern.

4. The system of claim 3, wherein the memory further stores code which when executed by the processor causes the processor to perform: determining an interblink interval for the patient’s eye based at least in part on the vital dye pattern.

5. The system of claim 4, wherein the memory further stores code which when executed by the processor causes the processor to perform:determining an ocular protection index for the patient’s eye based at least in part on the determined interblink interval and the determined tear film breakup time.

6. The system of claim 1 , wherein the memory further stores code which when executed by the processor causes the processor to perform determining a tear film breakup pattern category, of a defined set of categories, based at least in part on the vital dye pattern.

7. The system of claim 6, wherein the memory further stores code which when executed by the processor causes the processor to perform said determining the tear film breakup pattern category using machine learning techniques.

8. The system of claim 1 , wherein the memory further stores code which when executed by the processor causes the processor to perform determining a severity and / or a category of corneal staining, of a defined set of categories, based at least in part on the vital dye pattern.

9. The system of claim 8, wherein the defined set of categories of corneal staining comprises one or more of: epithelial staining, Superficial Punctate Keratitis, wounds, abrasion, and confluent staining.

10. The system of claim 8, wherein the memory further stores code which when executed by the processor causes the processor to perform said determining the seventy and / or category of corneal staining using machine learning techniques.11 . The system of claim 8, wherein the memory further stores code which when executed by the processor causes the processor to perform: extracting a region from an image of the eye; performing image processing on the extracted region to improve visibility of areas of corneal staining; and quantifying the areas of corneal staining in the region to provide a measure of seventy.

12. The system of claim 1 , wherein the hollow elongate member has a larger circumference at the first end than at the second end.

13. The system of claim 1 , wherein the hollow elongate member has a length of between about 4 cm and about 16 cm.

14. The system of claim 1 , wherein the hollow elongate member is formed of plastic.

15. The system of claim 1 , wherein the hollow elongate member comprises a rigid portion extending from the second end along a length of the hollow elongate member and an elastic portion or component at the first end, the elastic portion or component adapted to be pressed against an area surrounding the patient’s eye.

16. The system of claim 1 , wherein the hollow elongate member is substantially uniform in rigidity.

17. The system of claim 1 , wherein the hollow elongate member has a length that is determined based on a defined focal distance between the imaging device and the patient’s eye.

18. The system of claim 1 , wherein said one or more light sources are positioned in an interior portion of the hollow elongate member.

19. The system of claim 1 , wherein said one or more light sources form part of the imaging device.

20. The system of claim 1 , wherein the light sources comprise light emitting diodes (LEDs).21 . The system of claim 20, wherein said one or more LEDs have an emission wavelength of about 450 nm to about 500 nm.

22. The system of claim 20, wherein said one or more LEDs are formed on a printed circuit board (PCB).

23. The system of claim 22, wherein the PCB is positioned in the hollow elongate member to form an arrangement of said one or more LEDs around an inner circumference of the hollow elongate member.

24. The system of claim 20, wherein said one or more LEDs are positioned in an illumination assembly positioned within the hollow elongate member.

25. The system of claim 24, wherein the illumination assembly comprises a diffusion ring and said one or more LEDs are formed on an annular printed circuit board and are positioned in an interior portion of the diffusion ring.

26. The system of claim 25, wherein the illumination assembly further comprises a dome having a concave surface facing the diffusion ring to reflect light produced by the LEDs in a direction of the patient’s eye, the dome having an opening formed in a central portion thereof to allow light to pass through to the imaging device.

27. The system of claim 26, wherein the illumination assembly further comprises an annular cavity formed on a side of the dome opposite the diffusion ring, the annular cavity configured to receive a filter to filter the light passing through to the imaging device.

28. The system of claim 20, further comprising a battery connected to said one or more LEDs.

29. A method for evaluating ocular conditions, comprising: instilling a patient’s eye with a vital dye; capturing video or at least one image of the patient’s eye using a system comprising an imaging device having at least one lens, a processor, and memory, the system further comprising a hollow elongate member having a first end adapted to be positioned around the patient’s eye and a second end having a base portion adapted to attach to the imaging device so that the second end is positioned around said at least one lens of the imaging device, and the system further comprising one or more light sources provided in an interior portion of the hollow elongate member and positioned to illuminate the patient’s eye, wherein said capturing of the video or image is performed while said one or more light sources illuminate the patient’s eye; and detecting an ocular condition based at least in part on a vital dye pattern of the captured video or image.

30. The method of claim 29, further comprising transmitting the captured video or image via a network to a remote computer system which performs, at least in part, said detecting an ocular condition.31 . The method of claim 29, determining a tear film breakup time based at least in part on the vital dye pattern.

32. The method of claim 31 , further comprising determining an interblink interval for the patient’s eye based at least in part on the vital dye pattern or tracking of the lid margins or cornea.

33. The method of claim 32, further comprising determining an ocular protection index for the patient’s eye based at least in part on the determined interblink interval and the determined tear film breakup time or area.

34. The method of claim 29, further comprising determining a tear film breakup pattern category, of a defined set of categories, based at least in part on the vital dye pattern.

35. The method of claim 34, wherein said determining the tear film breakup pattern category is performed using machine learning techniques.

36. The method of claim 29, further comprising determining a severity and / or a category of corneal staining, of a defined set of categories, based at least in part on the vital dye pattern.

37. The method of claim 36, wherein the defined set of categories of corneal or conjunctival staining comprises one or more of: epithelial staining, Superficial Punctate Keratitis, abrasion, wounds and confluent staining.

38. The method of claim 36, wherein the memory further stores code which when executed by the processor causes the processor to perform said determining the seventy and / or category of corneal staining using machine learning techniques.

39. The method of claim 36, wherein the memory further stores code which when executed by the processor causes the processor to perform:extracting a region from an image of the eye; performing image processing on the extracted region to improve visibility of areas of corneal staining; and quantifying the areas of corneal staining in the region to provide a measure of seventy.

40. The method of claim 29, wherein said determining of the tear film breakup time for the patient’s eye comprises detecting tear film breakup using a machine learning model.41 . The method of claim 40, wherein the machine learning model is trained to detect dark regions in the patient’s eye indicative of where the vital dye has receded.

42. The method of claim 40, wherein said determining of the tear film breakup time for the patient’s eye further comprises determining an average or a moving average of a plurality of determined tear film breakup times.

43. The method of claim 29, wherein said determining of the interblink interval for the patient’s eye comprises detecting blink times using a machine learning model.

44. The method of claim 43, wherein the machine learning model: identifies a region containing the patient’s eye in frames of the captured video or image; and applies blink detection to the identified region.

45. The method of claim 43, wherein said determining of the interblink interval for the patient’s eye further comprises determining an average or a moving average of a plurality of determined interblink intervals.

Images