Docket Number: 10046-590WO1 8300 YEH TWO-PHOTON AUTOFLORESCENCE LIFETIME ASSAY OF PHOTORECEPTORS AND RETINAL PIGMENT EPITHELIUM DURING LIGHT- DARK VISUAL CYCLES IN RETINA CROSS-REFERENCE TO RELATED APPLICATIONS [0001] This application claims priority to and the benefit of U.S. Provisional Application No. 63/627,961, titled “TWO-PHOTON AUTOFLORESCENCE LIFETIME ASSAY OF PHOTORECEPTORS AND RETINAL PIGMENT EPITHELIUM DURING LIGHT-DARK VISUAL CYCLES IN RETINA,” filed on February 1, 2024, the content of which is hereby incorporated by reference herein in its entirety. GOVERNMENT SUPPORT CLAUSE [0002] This invention was made with government support under Grant no. R21 EY033106 awarded by the National Institutes of Health. The government has certain rights in the invention. BACKGROUND [0003] Existing retinal imaging tools such as scanning laser ophthalmoscopy (SLO), Optical Coherence Tomography (OCT), and Fundus autofluorescence (FAF) are unable to provide functional retinal information. For example, OCT offers macula maps that quantify retinal anatomy, and FAF may show patterns of autofluorescence that are useful for monitoring geographic atrophy, but no imaging tool exists for measuring the biochemistry of the visual cycle. One-photon Fluorescence lifetime imaging (1P-FLIO) systems have not been widely adapted for providing functional retinal information. [0004] What are needed are methods and apparatuses for characterizing and evaluating retinal function. SUMMARY [0005] Disclosed are computer-implemented methods, systems, and apparatuses for determining structural and/or functional information for a subject’s retina. Embodiments of the present disclosure measure the fluorescence lifetime of intrinsic retinal fluorophores on a cellular scale, revealing differences in lifetime between retinal cell classes under different conditions of light and dark exposure. This information can be used to characterize normal retinal physiology and disease-related disruption of cellular metabolism in diseases such as age-related macular degeneration (AMD).
Docket Number: 10046-590WO1 8300 YEH [0006] In some implementations, a method is provided. The method can include: delivering stimulation to a subject's retina with visible light-dark cycles; obtaining an image sequence of the subject's retinal fluorophores corresponding to the light-dark cycles, and determining a fluorescence lifetime value for each of a plurality of cell classes based, at least in part, on the image sequence. [0007] In some implementations, the fluorescence lifetime value is a two-photon (2P) fluorescence lifetime value. [0008] In some implementations, the method further includes: determining structural and/or functional information for the subject's retina based, at least in part, on the determined fluorescence lifetime values. [0009] In some implementations, the stimulation includes visible light. [0010] In some implementations, the retinal fluorophores include all-trans-retinol (AT-ROL), all-trans-retinal (AT-RAL), Nicotinamide adenine dinucleotide (NADH), flavin adenine dinucleotide (FAD), and/or melanin. [0011] In some implementations, the fluorescence lifetime values are determined based, at least in part, on time-domain and/or frequency-domain analysis to correlate fluorescence lifetime responses for each cell class to light-dark visual cycles. [0012] In some implementations, the image sequence of the retinal fluorophores is obtained by: delivering the stimulation in multiple light-dark visual cycles that each include a light exposure phase and a recovery phase, and capturing a subset of the image sequence subsequent to each recovery phase. [0013] In some implementations, the fluorescence lifetime values are determined by: determining a weight of each fluorophore at each pixel in the image sequence; segmenting a plurality of cells into photoreceptors and retinal pigment epithelium (RPE); and calculating a respective fluorescence lifetime value for each segmented cell. [0014] In some implementations, the plurality of cells is segmented using a machine learning model. [0015] In some implementations, the machine learning model includes at least one of a k-nearest neighbors (KNN) algorithm, watershed algorithm, Gaussian mixture model (GMM), K-means algorithm, hierarchical ward algorithm, mean shift algorithm, or spectral clustering algorithm. [0016] In some implementations, the method further includes: determining a disease state for the subject based, at least in part, on the determined fluorescence lifetime values.
Docket Number: 10046-590WO1 8300 YEH [0017] In some implementations, the fluorescence lifetime values and/or disease state are determined using a machine learning model. [0018] In some implementations, the disease state includes age-related macular degeneration or other retinal disease. [0019] In some implementations, the method further includes: administering treatment to the subject based on the determined disease state. [0020] In some implementations, a system is provided. The system can include: at least one processor; and a memory operably coupled to the at least one processor, wherein the memory has computer executable instructions stored thereon that, when executed by the at least one processor, cause the at least one processor to: obtain an image sequence of a subject's retinal fluorophores, wherein the image sequence is captured in response to visible light stimulation delivered to a subject's retina; and determine a fluorescence lifetime value for each of a plurality of cell classes based, at least in part, on analysis of the image sequence. [0021] In some implementations, the memory has further computer executable instructions stored thereon that, when executed by the at least one processor, cause the at least one processor to: determine structural and/or functional information for the subject's retina based, at least in part, on the determined fluorescence lifetime values. [0022] In some implementations, the stimulation includes visible light. [0023] In some implementations, the retinal fluorophores include all-trans-retinol (AT-ROL), all-trans-retinal (AT-RAL), Nicotinamide adenine dinucleotide (NADH), flavin adenine dinucleotide (FAD), and/or melanin. [0024] In some implementations, the memory has further computer executable instructions stored thereon that, when executed by the at least one processor, cause the at least one processor to determine the fluorescence lifetime values by performing time-domain and/or frequency-domain analysis to correlate lifetime responses for each cell class to light- dark visual cycles. [0025] In some implementations, the image sequence is obtained by: delivering the stimulation in multiple light-dark visual cycles that each include a light exposure phase and a recovery phase, and capturing a subset of the image sequence subsequent to each recovery phase. [0026] In some implementations, the memory has further computer executable instructions stored thereon that, when executed by the at least one processor, cause the at least one processor to determine the fluorescence lifetime values by: determining a weight of each fluorophore at each pixel in the image sequence; segmenting a plurality of cells into
Docket Number: 10046-590WO1 8300 YEH photoreceptors and retinal pigment epithelium (RPE); and calculating the fluorescence lifetime value for each segmented cell. [0027] In some implementations, the memory has further computer executable instructions stored thereon that, when executed by the at least one processor, cause the at least one processor to segment the plurality of cells and/or determine the fluorescence lifetime values using a machine learning model. [0028] In some implementations, the machine learning model includes at least one of a k-nearest neighbors (KNN) algorithm, watershed algorithm, gaussian mixture model (GMM), K-means algorithm, hierarchical ward algorithm, mean shift algorithm, or spectral clustering algorithm. [0029] In some implementations, the memory has further computer executable instructions stored thereon that, when executed by the at least one processor, cause the at least one processor to further: determine a disease state for the subject based, at least in part, on the determined fluorescence lifetime values. [0030] In some implementations, the disease state includes age-related macular degeneration or other retinal diseases. [0031] In some implementations, the memory has further computer executable instructions stored thereon that, when executed by the at least one processor, cause the at least one processor to: output a summary or report of lifetime changes during light-dark visual cycles and/or the subject's disease state. [0032] In some implementations, the system is a Two-Photon Fluorescence Lifetime Microscopy/Ophthalmoscopy (2P-FLIO) system. [0033] In some implementations, the system further includes at least one of a light source, a stimulator, and a detector. [0034] In some implementations, a computer-implemented method for characterizing retinal function and/or structure is provided. The computer-implemented method can include: obtaining an image sequence of a subject's retinal fluorophores, wherein the image sequence is captured in response to visible light stimulation delivered to a subject's retina; and determining a fluorescence lifetime value for each of a plurality of cell classes, based, at least in part, on analysis of the image sequence. [0035] In accordance with another embodiment, a method is provided. The method can include: identifying, using an imaging device, a region of interest (e.g., volume of tissue) in a subject's retina; delivering stimulation with visible light-dark cycles to at least one target location (e.g., at least one cell class) within the region of interest, wherein the stimulation is
Docket Number: 10046-590WO1 8300 YEH dynamically adjusted in real-time based on the subject's eye movement; determining one or more fluorescence lifetime values (e.g., fast fluorescence lifetime (FLT) point measurements) for the at least one target location based, at least in part, on a detected response to the delivered stimulation; and determining structural and/or functional information for the subject's retina based, at least in part, on the determined one or more fluorescence lifetime values. [0036] In some implementations, determining the one or more fluorescence lifetime values includes measuring fast fluorescence lifetime (FLT) at one or more retinal voxels, including in a lateral dimension and depth in response to the delivered stimulation. [0037] In some implementations, the visible light-dark cycles include at least one of white light or wavelength-specific stimulation in a 350-700 nanometer (nm) wavelength range. [0038] In some implementations, the method further includes: determining a disease state or condition and/or corresponding therapy for the subject based, at least in part, on the determined one or more fluorescence lifetime values. [0039] In some implementations, the at least one target location includes the subject's macula, photoreceptors, or the retinal pigment epithelium. [0040] In some implementations, the stimulation is delivered via a stimulating component, and wherein the stimulation is dynamically adjusted based on measurements obtained using an eye tracking component. [0041] In some implementations, the imaging device includes an Optical Coherence Tomography (OCT) device. [0042] In some implementations, the OCT is configured to facilitate depth stabilization based on coalignment between the OCT and the eye tracking component. [0043] In some implementations, the eye tracking component is configured to facilitate lateral localization for obtaining the measurements by tracking movement of the subject's eye in the x-direction, y-direction, and z-direction. [0044] In some implementations, identifying the region of interest includes identifying one or more landmarks in the subject's retina. [0045] In some implementations, the stimulation is delivered via a 2P FLIO system. [0046] In some implementations, the stimulation is delivered based, at least in part, on a timing protocol associated with the at least one target location.
Docket Number: 10046-590WO1 8300 YEH [0047] In accordance with another embodiment, as system is provided. The system can include: an eye tracking component configured to track movement of a subject's eye to faciliate lateral localization for obtaining measurements; an Optical Coherence Tomography (OCT) device configured to identify a region of interest in a subject's retina, wherein the OCT device is configured to facilitate depth stabilization based on coalignment between the OCT device and the eye tracking component; a two-photon fluorescence lifetime imaging device configured to deliver stimulation with visible light-ark cycles to at least one target location or cell class within the identified region of interest, wherein the stimulation is dynamically adjusted in real-time based on the subject's eye movement; and a controller configured to obtain one or more fluorescence lifetime values or fast fluorescence lifetime (FLT) point measurements for the at least one target location or cell class based, at least in part, on a detected response to the delivered stimulation. [0048] In some implementations, the system is embodied as a clinical instrument. [0049] It should be understood that the above-described subject matter may also be implemented as a computer-controlled apparatus, a computer process, a computing system, or an article of manufacture, such as a computer-readable storage medium. [0050] Other systems, methods, features and/or advantages will be or may become apparent to one with skill in the art upon examination of the following drawings and detailed description. It is intended that all such additional systems, methods, features and/or advantages be included within this description and be protected by the accompanying claims. BRIEF DESCRIPTION OF THE DRAWINGS [0051] The components in the drawings are not necessarily to scale relative to each other. Like reference numerals designate corresponding parts throughout the several views. [0052] FIGURE 1A is a flowchart of an example method in accordance with certain embodiments of the present disclosure. [0053] FIGURE 1B is a flowchart of another example method in accordance with certain embodiments of the present disclosure. [0054] FIGURE 1C shows an image sequence in accordance with certain embodiments of the present disclosure. [0055] FIGURE 2A shows an experimental setup of a two-photon fluorescence lifetime imaging (FLIM) system in accordance with certain embodiments of the present disclosure.
Docket Number: 10046-590WO1 8300 YEH [0056] FIGURE 2B shows multiple retinal layers imaging and unmixing with endogenous fluorophores in accordance with certain embodiments of the present disclosure. [0057] FIGURE 2C shows photoreceptor and retinal pigment epithelium lifetime images after unmixing in accordance with certain embodiments of the present disclosure. [0058] FIGURE 2D shows lifetime phasors obtained from time-domain decay data, after conducting Fourier transform in accordance with certain embodiments of the present disclosure. [0059] FIGURE 2E shows lifetime phasor calibration using fluorescein in accordance with certain embodiments of the present disclosure. [0060] FIGURE 2F is a ^ phasor plot of a convallaria sample. [0061] FIGURE 2G shows a false-colored FLIM image of the convallaria sample. [0062] FIGURE 2H shows the 98x1 fluorescence decay curve I(t) at each pixel separately transformed into ^ phasors. [0063] FIGURE 2I shows 256×256 ^ phasor sets (gτ, sτ) used as inputs for Gaussian Mixture Models (GMM). [0064] FIGURE 2J shows photoreceptors segmentation with the watershed algorithm and retinal pigment epithelium segmentation with the k-nearest neighbors (KNN) algorithm. [0065] FIGURE 2K shows a summary of lifetime changes during light-dark visual cycles. [0066] FIGURE 2L illustrates retina heating induced by near-infrared laser and white light exposure during light-dark visual cycles. [0067] FIGURE 2M shows an exemplary system (i.e., mFLIO system) comprising a 2P-FLIO module and a spectral-domain optical coherence tomography (SD-OCT) module. [0068] FIGURE 2N shows an example two-dimensional (2D) micro- electromechanical system-based (MEMS) scanner having a Lissajous scanning pattern and the raster scanner. [0069] FIGURE 2O shows an active feedback autofocus optical coherence tomography (AFOCT) unit of the exemplary system, as a block diagram, comprising controlled modules (e.g., 2P-FLIO module, SD-OCT module) and a microcontroller. [0070] FIGURE 2P shows the axial resolution of a 2P-FLIO overlay with SD- OCT.
Docket Number: 10046-590WO1 8300 YEH [0071] FIGURE 2Q shows (i) a customized eye model used for calibration and (ii) high-quality A-Scan, B-Scan, and en-face images generated by the SD-OCT module of the exemplary system. [0072] FIGURE 2R shows example life-dark cycle measurements. [0073] FIGURE 3A shows an example white light exposure scheme. [0074] FIGURE 3B-M show results from a first Dutch-Belted rabbit - retina preparation. [0075] FIGURE 4A-M show results from a second Dutch-Belted rabbit - retina preparation. [0076] FIGURE 5A-M show results from a first New Zealand White rabbit - retina preparation. [0077] FIGURE 6A-M show results from a second New Zealand White rabbit - retina preparation. [0078] FIGURE 6N shows a typical fluorescence lifetime (FLT) measurement. [0079] FIGURE 6O shows detailed maps delineating light-dark response changes across various retinal locations. [0080] FIGURE 7 is a summary table of photoreceptor lifetime changes and retinal pigment epithelium lifetime changes during light-dark visual cycles. [0081] FIGURE 8 is a schematic diagram showing the pigment epithelium visual cycle. [0082] FIGURE 9A shows hematoxylin and eosin stain retinal histology from New Zealand White rabbit. [0083] FIGURE 9B is a schematic diagram showing a representation of the pigment epithelium visual cycle for rod and cone photoreceptors. [0084] FIGURE 10 is an example computing device. [0085] FIGURE 11 shows a comparison between the processes (e.g., image acquisition, image reconstruction) of the one-photon fluorescence lifetime imaging (1P- FLIO) and multi-modal two-photon fluorescence lifetime imaging (mFLIO). DETAILED DESCRIPTION [0086] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art. Methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present disclosure.
Docket Number: 10046-590WO1 8300 YEH [0087] The following description of the disclosure is provided as an enabling teaching of the disclosure in its best, currently known embodiment(s). To this end, those skilled in the relevant art will recognize and appreciate that many changes can be made to the various embodiments of the present disclosure, while still obtaining the beneficial results of the present disclosure. It will also be apparent that some of the desired benefits of the present disclosure can be obtained by selecting some of the features of the present disclosure without utilizing other features. Accordingly, those who work in the art will recognize that many modifications and adaptations to the present disclosure are possible and can even be desirable in certain circumstances and are a part of the present disclosure. Thus, the following description is provided as illustrative of the principles of the present disclosure and not in limitation thereof. [0088] Reference will now be made in detail to the embodiments of the present disclosure, examples of which are illustrated in the drawings and the examples. The present disclosure may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. [0089] Terminology [0090] The term “comprising” and variations thereof as used herein is used synonymously with the term “including” and variations thereof and are open, non-limiting terms. Although the terms “comprising” and “including” have been used herein to describe various embodiments, the terms “consisting essentially of” and “consisting of” can be used in place of “comprising” and “including” to provide for more specific embodiments and are also disclosed. As used in this disclosure and in the appended claims, the singular forms “a”, “an”, “the”, include plural referents unless the context clearly dictates otherwise. [0091] The following definitions are provided for the full understanding of terms used in this specification. [0092] The terms “about” and “approximately” are defined as being “close to” as understood by one of ordinary skill in the art. In one non-limiting embodiment the terms are defined to be within 10%. In another non-limiting embodiment, the terms are defined to be within 5%. In still another non-limiting embodiment, the terms are defined to be within 1%. [0093] As used herein, the terms “may,” “optionally,” and “may optionally” are used interchangeably and are meant to include cases in which the condition occurs as well as cases in which the condition does not occur. [0094] The term “subject” refers to any individual who is the target of administration or treatment. The subject can be a vertebrate, for example, a mammal. In one
Docket Number: 10046-590WO1 8300 YEH embodiment, the subject can be human, non-human primate, bovine, equine, porcine, canine, or feline. The subject can also be a guinea pig, rat, hamster, rabbit, mouse, or mole. Thus, the subject can be a human or veterinary patient. The term “patient” refers to a subject under the treatment of a clinician, for example, a physician. [0095] A “control” is an alternative subject or sample used in an experiment for comparison purposes. A control can be “positive” or “negative.” [0096] The term “detect” or “detecting” refers to an output signal released for the purpose of sensing of physical phenomenon. An event or change in environment is sensed and signal output released in the form of light, heat, or a color change (e.g., color change from red to blue, white to black, or vice versa). [0097] An “increase” can refer to any change that results in a larger amount of a symptom, disease, composition, condition, or activity. For example, an increase can be a change in the symptoms of a disorder such that the symptoms are more than previously observed. An increase can be any individual, median, or average increase in a condition, symptom, activity, composition in a statistically significant amount. Thus, the increase can be a 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, or 100% increase so long as the increase is statistically significant or observable. [0098] A “decrease” can refer to any change that results in a smaller amount of a symptom, disease, composition, condition, or activity. For example, a decrease can be a change in the symptoms of a disorder such that the symptoms are less than previously observed. A decrease can be any individual, median, or average decrease in a condition, symptom, activity, composition in a statistically significant amount. Thus, the decrease can be a 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, or 100% decrease so long as the decrease is statistically significant or observable. [0099] “Inhibit,” “inhibiting,” and “inhibition” mean to decrease an activity, response, condition, disease, or other biological parameter. This can include but is not limited to the complete ablation of the activity, response, condition, or disease. This may also include, for example, a 10% reduction in the activity, response, condition, or disease as compared to the native or control level. Thus, the reduction can be a 10, 20, 30, 40, 50, 60, 70, 80, 90, 100%, or any amount of reduction in between as compared to native or control levels. [00100] By “reduce” or other forms of the word, such as “reducing” or “reduction,” is meant lowering of an event or characteristic. It is understood that this is
Docket Number: 10046-590WO1 8300 YEH typically in relation to some standard or expected value, in other words it is relative, but that it is not always necessary for the standard or relative value to be referred to. [00101] By “prevent” or other forms of the word, such as “preventing” or “prevention,” is meant to stop a particular event or characteristic, to stabilize or delay the development or progression of a particular event or characteristic, or to minimize the chances that a particular event or characteristic will occur. Prevent does not require comparison to a control as it is typically more absolute than, for example, reduce. As used herein, something could be reduced but not prevented, but something that is reduced could also be prevented. Likewise, something could be prevented but not reduced, but something that is prevented could also be reduced. It is understood that where reduce or prevent are used, unless specifically indicated otherwise, the use of the other word is also expressly disclosed. [00102] “Effective amount” of an agent refers to a sufficient amount of an agent to provide a desired effect. The amount of agent that is “effective” will vary from subject to subject, depending on many factors such as the age and general condition of the subject, the particular agent or agents, and the like. Thus, it is not always possible to specify a quantified “effective amount.” However, an appropriate “effective amount” in any subject case may be determined by one of ordinary skill in the art using routine experimentation. Also, as used herein, and unless specifically stated otherwise, an “effective amount” of an agent can also refer to an amount covering both therapeutically effective amounts and prophylactically effective amounts. An “effective amount” of an agent necessary to achieve a therapeutic effect may vary according to factors such as the age, sex, and weight of the subject. Dosage regimens can be adjusted to provide the optimum therapeutic response. For example, several divided doses may be administered daily or the dose may be proportionally reduced as indicated by the exigencies of the therapeutic situation. [00103] A “pharmaceutically acceptable” component can refer to a component that is not biologically or otherwise undesirable, i.e., the component may be incorporated into a pharmaceutical formulation provided by the disclosure and administered to a subject as described herein without causing significant undesirable biological effects or interacting in a deleterious manner with any of the other components of the formulation in which it is contained. When used in reference to administration to a human, the term generally implies the component has met the required standards of toxicological and manufacturing testing or that it is included on the Inactive Ingredient Guide prepared by the U.S. Food and Drug Administration.
Docket Number: 10046-590WO1 8300 YEH [00104] The terms “treat,” “treating,” “treatment,” and grammatical variations thereof as used herein, refer to the medical management of a patient with the intent to cure, ameliorate, stabilize, or prevent a disease, pathological condition, or disorder. This term includes active treatment, that is, treatment directed specifically toward the improvement of a disease, pathological condition, or disorder, and also includes causal treatment, that is, treatment directed toward removal of the cause of the associated disease, pathological condition, or disorder. In addition, this term includes palliative treatment, that is, treatment designed for the relief of symptoms rather than the curing of the disease, pathological condition, or disorder; preventative treatment, that is, treatment directed to minimizing or partially or completely inhibiting the development of the associated disease, pathological condition, or disorder; and supportive treatment, that is, treatment employed to supplement another specific therapy directed toward the improvement of the associated disease, pathological condition, or disorder. Treatments according to the present disclosure may be applied preventively, prophylactically, palliatively or remedially. Prophylactic treatments are administered to a subject prior to onset (e.g., before obvious signs of a disease state), during early onset (e.g., upon initial signs and symptoms of a disease state), or after an established development of a disease state. Prophylactic administration can occur for day(s) to years prior to the manifestation of symptoms of an infection. [00105] The term “artificial intelligence” is defined herein to include any technique that enables one or more computing devices or comping systems (i.e., a machine) to mimic human intelligence. Artificial intelligence (AI) includes, but is not limited to, knowledge bases, machine learning, representation learning, and deep learning. The term “machine learning” is defined herein to be a subset of AI that enables a machine to acquire knowledge by extracting patterns from raw data. Machine learning techniques include, but are not limited to, logistic regression, support vector machines (SVMs), decision trees, Naïve Bayes classifiers, and artificial neural networks. The term “representation learning” is defined herein to be a subset of machine learning that enables a machine to automatically discover representations needed for feature detection, prediction, or classification from raw data. Representation learning techniques include, but are not limited to, autoencoders. The term “deep learning” is defined herein to be a subset of machine learning that that enables a machine to automatically discover representations needed for feature detection, prediction, classification, etc. using layers of processing. Deep learning techniques include, but are not limited to, artificial neural network or multilayer perceptron (MLP).
Docket Number: 10046-590WO1 8300 YEH [00106] Machine learning models include supervised, semi-supervised, and unsupervised learning models. In a supervised learning model, the model learns a function that maps an input (also known as feature or features) to an output (also known as target or targets) during training with a labeled data set (or dataset). In an unsupervised learning model, the model learns patterns (e.g., structure, distribution, etc.) within an unlabeled data set. In a semi-supervised model, the model learns a function that maps an input (also known as feature or features) to an output (also known as target or target) during training with both labeled and unlabeled data. [00107] “Fluorescence lifetime” refers to a time period or duration for which a fluorescence signal persists after being in an excited state. For example, at a ground energy level or ground state (^^) (terms used interchangeably herein), a fluorophore (e.g., molecule) can absorb light energy equal to or greater than its higher energy levels (^^, ^^ ^^ ^^), and its electron(s) will become temporarily excited by the higher energy levels. The excited electron(s) undergo non-radiative vibrational relaxation and internal conversion, eventually reaching the least energetic vibrational state of the excited phase (^^) before returning to their ground energy level (^^). Fluorescence is a radiative process in which fluorophores (e.g., molecules) transition back to their ground energy level by emitting detectable photons on a timescale of nanoseconds. The emitted fluorescence photon possesses lower energy resulting in a longer emission wavelength. This emission originates from the least excited electronic level (^^), resulting in a consistent emission spectrum regardless of the excitation wavelength that is dependent on the population of vibrational levels within the ground state. The fluorescence lifetime (^) can be defined as the time it takes for the intensity to drop to ^ ^ (= 0.368) of its initial value and may indicate an average duration that a fluorophore remains in its excited state. Fluorescence intensity measurement is sensitive to artifacts from scattered light, photobleaching, instrumentation parameters, and the concentration of the fluorophores. In contrast, the fluorescence lifetime provides an absolute measurement which is less susceptible to artifacts and is sensitive to the local environment of the fluorescent molecule. Different molecular environments, such as polarity, pH, or interactions with other molecules, can affect the fluorescence lifetime. This sensitivity to environmental changes allows for more precise identification and characterization of the molecular microenvironment. [00108] Fluorescence lifetime imaging ophthalmoscopy (FLIO) is based on fluorescence lifetime imaging microscopy (FLIM) which is used in basic science for analysis of microscopic images of fixed, as well as living cells. Every single fluorophore is
Docket Number: 10046-590WO1 8300 YEH characterized by its own excitation and emission wavelength spectrum and exhibits an individual fluorescence lifetime. The fluorescence lifetimes depend on the molecular environment but are largely independent of the fluorophore's concentration. Therefore, fluorescence lifetime measurement can be applied to detect weakly fluorescing fluorophores if they differ in terms of their lifetime. Additionally, lifetimes can be used as indicator for specific metabolic conditions of changes within the molecular micro-environment. Fundus autofluorescence intensity imaging and FLIO are two very closely related imaging modalities and thereby share many common facets. Whereas conventional fundus autofluorescence measurement provides spatial resolved information on fluorescence intensities, FLIO additionally measures fluorescence lifetimes or decay times and thereby includes time as a third dimension (space and time resolved). This additional dimension enables extracting many components that have overlapping emission properties. [00109] Retinal autofluorescence intensity predominantly stems from lipofuscin which is located within the lysosomes of the retinal pigment epithelium (also referred to herein as RPE). Accumulation of lipofuscin is a hallmark of aging retinal pigment epithelium cells originating from incomplete degradation of photoreceptor outer segment disks. Major constituents of retinal pigment epithelium lipofuscin are a complex mixture of di-retinal conjugates, one of which is A2E. Extracellular fluorophores from shed outer segment debris in the subretinal space as well as extracellular proteins also contain components from di-retinal adducts. Additionally, melanin has been reported to have a peak excitation wavelength of 450 nm with a peak emission starting at 440 nm extending to the near-infrared spectra (>800 nm). Therefore, melanin is likely to contribute to fluorescence lifetime measurements with FLIO. However, given the major contribution of lipofuscin to the autofluorescence signal, it is difficult to identify weaker endogenous fluorophores using autofluorescence intensity measurement in the retina. Because fluorescence lifetimes are largely independent of the fluorophores’ concentration and intensity, the predominance of lipofuscin can be overcome with FLIO, and fluorophores other than lipofuscin can be identified by their lifetimes. [00110] Two-photon excited fluorescence (TPEF) is a powerful technique that enables the excitation of intrinsic retinal fluorophores involved in cellular metabolism and the visual cycle. Although previous intensity-based TPEF studies in non-human primates have successfully imaged several classes of retinal cells and elucidated aspects of both rod and cone photoreceptor function, fluorescence lifetime imaging (FLIM) of the retinal cells under light-dark visual cycle has yet to be fully exploited. Embodiments of the present disclosure
Docket Number: 10046-590WO1 8300 YEH provide a FLIM assay of photoreceptors and retinal pigment epithelium (RPE) that reveals key insights into retinal physiology and adaptation. It was found that photoreceptor fluorescence lifetimes increase and decrease in sync with light and dark exposure, respectively. This is likely due to changes in all-trans-retinol and all-trans- retinal levels in the outer segments, mediated by phototransduction and visual cycle activity. During light exposure, retinal pigment epithelium fluorescence lifetime was observed to increase steadily over time, suggesting all-trans-retinol accumulation during the visual cycle. In some implementations, the proposed system can measure the fluorescence lifetime of intrinsic retinal fluorophores on a cellular scale, revealing differences in lifetime between retinal cell classes under different conditions of light and dark exposure. This represents a powerful approach to shed light on both normal retinal physiology and disease-related disruption of cellular metabolism in diseases such as age-related macular degeneration (AMD). [00111] Embodiments of the present disclosure include multimodal Two- Photon Fluorescence Lifetime Microscopy (2P-FLIM) systems adapted to provide insights into visual cycle dynamics and biochemical processes that are suitable for lifetime-based evaluation of retinal health. In one embodiment described herein, a custom multimodal FLIM instrument capable of imaging the intrinsic fluorescence signatures of the retina at subcellular resolution is provided. This instrument offers a new dimension for visualizing the intricate dynamics of photoreceptors and retinal pigment epithelium during light-dark visual cycles, opening doors to advanced research possibilities. Studies were conducted delving into the dynamics of the visual cycle and its impact on fluorescence lifetimes by simultaneously monitoring the changes in fluorescence lifetimes of photoreceptors and retinal pigment epithelium cells. By employing both time domain and frequency domain analyses, a unique, single-factor approach to assess the health of photoreceptors and retinal pigment epithelium cells in relation to light-dark visual cycles is provided. This approach can provide valuable insights into retinal health, offering a new perspective on how the retina adapts and functions under varying light conditions. [00112] Photoreceptors and the retinal pigment epithelium (RPE) are essential components of the visual system, playing crucial roles in vision and the regulation of visual cycle reactions [1,2]. Photoreceptors, including rods and cones, are specialized light-sensitive cells located in the retina. They are responsible for capturing and converting light into electrical signals, which are then transmitted to the brain for visual processing. The RPE, on the other hand, is a layer of pigmented cells located between the photoreceptors and the
Docket Number: 10046-590WO1 8300 YEH underlying choroid. It provides essential support functions to maintain the integrity and functionality of the photoreceptor cells. [00113] The visual cycle is a complex process involving the continuous recycling of retinoids, which are vitamin A derivatives, within the photoreceptor and retinal pigment epithelium cells [3]. The visual cycle plays a critical role in the regeneration of photopigments required for vision. During phototransduction, retinoids undergo isomerization and conversion, allowing the photoreceptors to respond to light stimuli. The visual cycle ensures the replenishment of retinoids and enables the photoreceptors to maintain their light-sensing capabilities. Although intensity-based TPEF ophthalmoscopy in non- human primates has successfully visualized several classes of retinal structures and probed both rod and cone function [4,5], fluorescence lifetime imaging at the cellular scale has the potential to provide further insight into both basic physiology and pathology of the retina [6- 8]. Being an intrinsic property of a fluorescent molecule, fluorescence lifetime measured by a FLIM system is not biased by excitation power [9,10] or probe concentration [11]. In addition, lifetime reading is not prone to photobleaching [12] and can shed light on the microenvironment surrounding the fluorophore [9,13,14]. This is of particular importance when imaging the outer retinal layers, which require passage of both excitation and emission through several layers of vasculature and inner retina cell classes. In vivo one-photon widefield fluorescence lifetime imaging ophthalmoscopy (FLIO) has been conducted in both rodent and human fundus using a scanning laser ophthalmoscope [15,16]. Widefield FLIO shows potential as a quantitative measure of retinal health, enabling early detection of abnormalities or dysfunctions before structural changes become apparent [17]. Notably, alterations in retinal fluorescence lifetime are associated with various retinal diseases, including age-related macular degeneration [18-20], retinitis pigmentosa [21], Stargardt disease [22], and choroideremia [23], among others [10,24]. By measuring the fluorescence signals emitted by retinoids and their condensation products [25] in response to light or chemical stimuli, we can assess the efficiency of the visual cycle and detect any abnormalities or dysfunctions [4,5]. Multiphoton imaging techniques can offer improved contrast and resolution, enabling high-throughput screening of retinal tissues at the single-cell level [7,26,27]. [00114] A custom multimodal FLIM instrument was developed to image the intrinsic fluorescence signatures of the retina at subcellular resolution. A study was conducted demonstrating the application of two-photon FLIM to visualize retinal function during visual cycle. Using a rabbit model, FLIM’s ability to provide novel insights into
Docket Number: 10046-590WO1 8300 YEH photoreceptor and retinal pigment epithelium physiology during light/dark cycles is demonstrated. The dependence of fluorescence lifetime on light-dark cycles reveals the potential of FLIM to shed light on basic retinal function in health and disease. Additionally, a new method to evaluate the health of photoreceptors and retinal pigment epithelium cells corresponding to light-dark visual cycles using fluorescence lifetime data is introduced herein. Both time domain and frequency domain analysis is employed to assess the correlation between photoreceptor and retinal pigment epithelium lifetimes and the light-dark exposure function. This approach offers a single factor to quantify the response effectively, providing valuable insights into retinal health during light-dark visual cycles. [00115] Example Method [00116] FIG. 1A is a flowchart of an example computer-implemented method 100A for determining a fluorescence lifetime value for each of a plurality of cell classes. In some implementations, the method 100A can be performed by a processing circuitry (for example, but not limited to, an application-specific integrated circuit (ASIC), or a central processing unit (CPU)). In some examples, the processing circuitry may be electrically coupled to and/or in electronic communication with other circuitries of an example computing device, such as, but not limited to, the example computing device 1000 described above in connection with FIG. 10. In some examples, embodiments may take the form of a computer program product on a non-transitory computer-readable storage medium storing computer-readable program instruction (e.g., computer software). Any suitable computer- readable storage medium may be utilized, including non-transitory hard disks, CD-ROMs, flash memory, optical storage devices, or magnetic storage devices. This disclosure contemplates that the example operations can be performed using one or more computing devices (e.g., at least the basic configuration illustrated in FIG. 10 by box 1002). The example method 100A can be performed using a clinical instrument as described herein, to facilitate real-time measurements that can be used to determine a disease state, prognosis, treatment, or the like in clinical settings. [00117] The method shown in FIG. 1A includes obtaining an image sequence of a subject’s retinal fluorophores corresponding to light-dark cycles and can include determining structural and/or functional information for the subject’s retina based, at least in part, on the determined fluorescence lifetime values. Conventional technologies do not use fluorescence lifetime values as a basis for determining retinal structural and/or functional information, and as discussed above, are generally incapable of providing any functional information about the retina.
Docket Number: 10046-590WO1 8300 YEH [00118] At step 110, the method includes delivering stimulation, for example, visible light, to a subject’s retina with visible light-dark cycles (e.g., 350-650 nanometer (nm) wavelength). In some implementations, the stimulation is delivered in multiple light- dark visual cycles that each comprise a light exposure phase and a recovery phase. In other implementations, the stimulation can be delivered in a single light-dark visual cycle. In some embodiments, the stimulation is delivered via a 2P-FLIO system. The example 2P-FLIO system can include various electrical and/or mechanical components such as a light source (e.g., one or more lasers), a stimulator, and a detector. [00119] At step 120, the method includes obtaining an image sequence of the subject’s retinal fluorophores corresponding to the light-dark visual cycles. For example, the image sequence can be captured in response to visible light stimulation delivered to a subject’s retina. In some examples, the image sequence can comprise a series of individual images or a video stream. In some examples, obtaining the image sequence includes delivering the stimulation in multiple light-dark visual cycles that each comprise a light exposure phase and a recovery phase, and capturing a subset of the image sequence subsequent to each recovery phase. In some implementations, the retinal fluorophores can comprise all-trans-retinol (AT-ROL), all-trans-retinal (AT-RAL), Nicotinamide adenine dinucleotide (NADH), flavin adenine dinucleotide (FAD), and/or melanin. [00120] FIG. 1C shows an image sequence 100C demonstrating a fluorescence lifetime for retinal fluorophores. The fluorescence lifetime value (^) for the depicted retinal ^ fluorophores can be the time it takes for the intensity to drop to ^ (= 0.368) of its initial value and indicates an average duration that a fluorophore or fluorophore class remains in an excited state. The image sequence 100C begins at an initial time value (t = 1) corresponding with a ground energy level or state. The image sequence 100C progresses to a subsequent time value (t = 5) where stimulation is delivered to the subject’s retina and then terminates at a final time value (t = 97) where the excited retinal fluorophores return to their ground energy level or state. Using the systems and methods described herein, different classes of retinal fluorophores can be imaged simultaneously, and the intensity of each cell class will vary. For example, retinal pigment epithelium cells have a short fluorescence lifetime relative to photoreceptor cells, and this is observable in a given image sequence as the intensity/brightness of excited photoreceptor cells will last longer (terminating during a time period that is subsequent in time) than that of excited retinal pigment epithelium cells. Based on this lifetime difference between photoreceptors and RPE, an algorithm was developed to
Docket Number: 10046-590WO1 8300 YEH unmix the photoreceptors and retinal pigment epithelium cells and segment them into single cells for subsequent analysis. [00121] Returning to FIG. 1A, at step 130, the method includes determining a fluorescence lifetime value for each of a plurality of cell classes based, at least in part, on the image sequence. The fluorescence lifetime value can be or comprise a two-photon (2P) fluorescence lifetime value. In some examples, the fluorescence lifetime values are determined based, at least in part, on time-domain and/or frequency-domain analysis to correlate fluorescence lifetime responses for each cell class to light-dark visual cycles. In some examples, determining the fluorescence lifetime values comprises determining a weight of each fluorophore at each pixel in the image sequence, segmenting a plurality of cells into photoreceptors and RPE, and calculating a respective fluorescence lifetime value for each segmented cell. In some embodiments, the plurality of cells is segmented (e.g., partitioned into different classes such as photoreceptors and RPE) using a machine learning model. The machine learning model can be or comprise at least one of a k-nearest neighbors (KNN) algorithm, watershed algorithm, Gaussian mixture model (GMM), and/or a clustering algorithm, such as, but not limited to, K-means, hierarchical ward, mean shift, and/or spectral clustering algorithm. [00122] At step 140, the method includes determining (e.g., characterizing) structural and/or functional information for the subject’s retina based, at least in part, on the determined fluorescence lifetime values. The structural and/or functional information may be presented visually or through other mediums. [00123] Optionally, at step 150, the method includes determining a disease state and/or treatment based, at least in part, on the determined fluorescence lifetime values. Example disease states can include age-related macular degeneration or other retinal disease. Additionally, in some implementations, the disease state and/or treatment can be determined using a machine learning model. In some implementations, the method includes determining a prognosis for the subject and/or determining a response to treatment for the subject. Additionally, the method can include providing a determination of minimal or measurable residual disease for the subject or providing a treatment to the subject. [00124] Optionally, in some implementations, the method further includes generating and/or outputting a report including fluorescence lifetime changes during light- dark visual cycles and/or the subject’s disease state. Optionally, the report is integrated into the subject’s electronic health record (EHR). Alternatively or additionally, the method optionally further includes generating display data for the report. Alternatively or
Docket Number: 10046-590WO1 8300 YEH additionally, the method optionally further includes transmitting the report over a network. This disclosure contemplates that operations related to generation of the report can be performed using one or more computing devices (e.g., at least the basic configuration illustrated in FIG. 10 by box 1002). [00125] In some implementations, the method optionally further includes, in response to detecting a particular disease state in the subject (e.g., AMD), providing a diagnosis for the subject. In some embodiments, this disclosure contemplates that the determined fluorescence lifetime values described herein are the only or primary information used to make the diagnosis. Optionally, in other embodiments, the determined fluorescence lifetime values described herein are used in combination with other test results (e.g., clinical evaluation) to make the diagnosis. Additionally, the method optionally further includes, in response to detecting a disease state, providing a prognosis for the subject. Alternatively or additionally, the method optionally further includes recommending a treatment for the subject. Treatment approaches can vary depending on the specific disease state, progression, and patient factors. This disclosure contemplates that the operations related to providing diagnosis, prognosis, and/or treatment options can be performed using one or more computing devices (e.g., at least the basic configuration illustrated in FIG. 10 by box 1002). Optionally, in some implementations, the method further includes administering the recommended treatment to the subject. [00126] In some implementations, low numerical aperture (NA) optics are employed in a clinical instrument to facilitate measurement of light-dark cycles in vivo. Real- time eye tracking and depth stabilization can be used to dynamically adjust stimulation and measurement to account for eye movements. In some embodiments, an eye tracking module is used for lateral localization which allows for safely scanning a retinal volume (e.g., a 300 × 300 × 50 µm3 retinal volume) to rapidly obtain a point FLT measurement, for example, in less than four seconds. In some embodiments, using OCT tracking, the excitation depth offset is utilized to weigh the fluorescence signal for either photoreceptor outer segments or the RPE. Dual scanning optics can be employed: one for region-of-interest (ROI) selection and tracking, and one for rapid scan for the point measurement. This disclosure contemplates the use of various timing paradigms, described in more detail herein, to characterize light-dark cycles for specific retinal locations and correlate those measurements with known photoreceptor distributions. The proposed unmixing approach is directed toward separating the fluorescence signals from the RPE, PR, and background, allowing for accurate quantification of layer-specific fluorescence lifetime changes during the light/dark cycle.
Docket Number: 10046-590WO1 8300 YEH Although multiple fluorophores (e.g., all-trans-retinal, all-trans-retinol, 11-cis-retinal, and 11- cis-retinol) can contribute to the fluorescence signal and their concentrations may affect the lifetime measurements, the proposed approach provides a meaningful representation of changes in fluorescence lifetime and offers valuable insights into the light/dark cycle's effects on visual cycle dynamics, even without a complete unmixing of the resident fluorophores. In some embodiments, OCT guided excitation depth offset is used to preferentially excite fluorophores in RPE and PR. To account for excitation overlap, measurements can be interpreted as a weighted contribution from the RPE and PR. FLIO and OCT offer unparalleled capabilities to probe retinal dynamics non-invasively, capturing spatiotemporal changes in fluorescence lifetimes and providing high-resolution, depth-resolved structural information, respectively. [00127] FIG. 1B is a flowchart of another example method 100B that can be used to determine retinal structural and/or functional information for a subject. The method 100B can be implemented using a clinical system or instrument, such as the exemplary system depicted in FIG. 2M (i.e., mFLIO system) comprising a 2P-FLIO module and a spectral-domain optical coherence tomography (SD-OCT) module. Integrating FLIO with OCT for 2P probe-beam depth stabilization as described herein facilitates innovative methodologies to discern cellular responses within the RPE and photoreceptor layers, bridging optics, imaging, computational analysis, and ophthalmology to advance our understanding of retinal physiology. It should be understood that the method 100B can include some or all of the steps described above in connection with FIG. 1A. [00128] At step 160, the method 100B includes identifying a region-of-interest (ROI) in a subject’s retina. In some examples, identifying the ROI comprises identifying one or more landmarks in the subject’s retina that may be associated with or used to identify specific retinal locations/targets. As described in connection with FIG. 6O (panels a and b) detailed maps can be generated delineating light-dark response changes across various retinal locations, correlating these alterations with diverse cone-rod cell concentrations and retinal topography. With reference to rabbit retinas, used as a proxy for human retinas as discussed in the experimental results below, rabbits are known to have a visual streak (VS), where the rod and cone photoreceptor, ganglion cell and amacrine cell density is highest, and which is located roughly 3 mm ventral to the optic nerve head (ONH). In some embodiments, inside the visual streak (A, B, C shown in FIG. 6O) and outside the visual streak (D, E, F, G, H, I shown in FIG. 6O) are sampled and the amplitude and decay constant of the light-dark cycles with the retinal anatomy can be correlated. In some implementations, OCT guided excitation
Docket Number: 10046-590WO1 8300 YEH depth offset is used to preferentially excite fluorophores in RPE and/or PR. In some examples, to account for excitation overlap, measurements will be interpreted as a weighted contribution from the RPE and PR. In other words, the excitation depth offset can be utilized to weigh the fluorescence signal for either photoreceptor outer segments or the RPE. As noted above, dual scanning optics can be employed for ROI selection/tracking and rapid scan for point measurements. [00129] At step 165, the method 100B includes delivering stimulation with visible light-dark cycles to at least one target location (e.g., layer, portion) within the ROI (e.g., one or more cell classes), wherein the stimulation is dynamically adjusted in real-time based on the subject’s eye movement, for example, using an eye tracking component. The visible light-dark cycles can comprise at least one of white light or wavelength-specific stimulation in a 350-700 nanometer (nm) wavelength range. The stimulation can be delivered using a stimulating component or 2P-FLIO system. The eye tracking component can be configured to track movement of the subject’s eye in the x-direction, y-direction, and z- direction to enable targeted stimulation and/or point measurements. The stimulation can be delivered based on a timing protocol associated with a target retinal location. FLIO and OCT imaging modalities can be integrated to ensure precise depth co-registration. The OCT can facilitate depth stabilization based on coalignment between the OCT and an eye tracking component that serves to faciliate lateral localization for obtaining measurements. Variations in light-dark cycle amplitude and time constant can be used as a reliable proxy for determining/analyzing retinal functionality. [00130] At step 170, the method 100B includes determining (e.g., using a processor, measurement device, and/or controller that is operatively coupled to an OCT, 2P- FLIO, and/or eye tracker) one or more fluorescence lifetime values or fast fluorescence lifetime (FLT) point measurements for the at least one target location, based at least in part, on a detected response to the delivered stimulation. In some embodiments, determining the one or more fluorescence lifetime values or FFT point measurements comprises measuring FLT at one or more retinal voxels, including in a lateral dimension and depth in response to the delivered stimulation. In some implementations, the target location can be or comprise the macula, photoreceptors, or the retinal pigment epithelium. In some implementations, machine learning algorithms (e.g., operations, models, techniques) are used to determine fluorescence lifetime values, FLT point measurements, measurement differences, and/or depth offset to discriminate and extract RPE and photoreceptor lifetime responses from complex FLIO measurements/datasets. In some embodiments, step 170 includes generating and/or outputting
Docket Number: 10046-590WO1 8300 YEH one or more outputs (e.g., a detailed map in a display or report) delineating light-dark amplitude response changes across various retinal locations, correlating these alterations with diverse cone-rod cell concentrations and retinal anatomy. Subsequently, the method 100B can proceed to step 140 and/or step 150, described above in connection with FIG. 1A [00131] Experimental Results [00132] A study was conducted to evaluate the exemplary systems and methods described herein. A custom multimodal FLIO instrument to image the intrinsic fluorescence signatures of the retina at subcellular resolution was developed. To demonstrate the application of two-photon FLIM for visualizing retinal function during the visual cycle, the system was tested on dissected retina slides under high numerical aperture (NA) objective, and in the near future will be demonstrated in a live rabbit. FLIM’s ability to provide novel insights into photoreceptor and retinal pigment epithelium physiology during light/dark cycles were successfully demonstrated. [00133] The system was successfully tested on New Zealand white (NZW) rabbits and Dutch-belted (DB) rabbits. The time-correlated single-photon counting (TCSPC) method for lifetime measurements was utilized to get high-accuracy lifetime measurements from the autofluorescence signal of photoreceptors and retinal pigment epithelium in the rabbit retina. [00134] Two-photon fluorescence lifetime ophthalmoscopy (2P-FLIO) [00135] The experimental setup of 2P-FLIO with raw data presentation is depicted in FIG. 2A. As shown, the setup includes a first laser 218, a second laser 219, a first photomultiplier tube (PMT1) 220, a second photomultiplier tube (PMT2) 221, a 1P-FLIO 222, a 2P FLIO 223, a TCSPC module 225, a SLO 226, lens 227, polarizing beam splitter (PBS) 228, emission filter (EM) 229, mirror (M) 231, a first dichroic mirror (DM1) for the 2P FLIO 223, and a second dichroic mirror (DM2) for the 1P-FLIO 222, and a galvo mirror (GM) 234. [00136] A fiber-based femtosecond laser (CFL-04RFF, Calmar Laser) was utilized, providing 90- femtosecond (fs) pulses at the central wavelength of 780 nm with pulse repetition frequency (PRF) at 80 MHz, with close to transform-limited temporal shape were delivered to the retinal plane. After dispersion pre-compensation, the beam entered a 2P-FLIO with x-y galvanometer scanners (GM) and a telescope relaying the GM’s plane to the pupil plane of the eye or the front focal plane of the 60x, NA 1.2 silicone immersion objective (UPLSAPO60XS2, Olympus). This disclosure discusses results with the high NA objective and freshly dissected retinal flat mounts, although the same experimental set-up can
Docket Number: 10046-590WO1 8300 YEH be used with low NA lenses to image the living eye. The 2P-FLIO allows for simultaneous frame registration in 2 channels, de-scanned fluorescence and reflectance. Reflectance images were obtained with the same light as for 2P-FLIO images and served to adjust eye position before imaging as guidance for subsequent alignment of fluorescence frames and to correct motion artifacts within the frame. The two-photon excited fluorescence (TPEF) emission was collected with a 500-720 nm emission filter by a cooled low dark count photomultiplier tube (PMT) (H74229-40, Hamamatsu Corp.) and amplified with 2 GHz cutoff bandwidth preamplifiers (HFAC-26, Becker and Hickl GmbH). The amplified signal was then measured and correlated to the reference clock of the femtosecond laser with a time-correlated single photon counting (TCSPC) module (PicoHarp 300, PicoQuant). Collected TPEF photons were assigned to one of 98 time bins for each pixel, depending on arrival time with respect to the synchronization pulse of the femtosecond laser. For comparison purposes, a 470 nm picosecond pulse laser (LDH-D-C-470, PicoQuant) was installed for one-photon fluorescence lifetime ophthalmoscopy (1P-FLIO). The 470 nm laser beam followed the same excitation path of the 2P-FLIO. The one-photon excited fluorescence was collected through a 150 μm pinhole (“P” in FIG. 2A) and focused on the same PMT. In the case of 1P-FLIO, the amplified signal is correlated to the reference clock of the picosecond pulse laser. [00137] In the 2P-FLIO, lifetime information is sampled in 98 bins (from 0 to 12.5 ns), which creates an sFLIM dataset with the size of 256(x: pixels) × 256(y: pixels) × 98(t: time bins). This dataset, denoted as xyt, is a multidimensional array where x and y represent the spatial dimensions (0.3 m wide per pixel), and t represents the fluorescence in a time bin (128 ps long each bin). At each pixel, there is a 98×1 sub-dataset termed fluorescence decay curve. This fluorescence decay curve can be transformed into a point (gτ, sτ) called a lifetime phasor through the digital Fourier transform (DFT) as shown in FIG. 2B which depicts multiple retinal layers imaging and unmixing with endogenous fluorophores. From there, a phasor plot can be built, which is denoted as ^ phasors (or lifetime phasors), and the plot contains the lifetime information of the embedded fluorophores. [00138] FIG. 2C shows photoreceptor and retinal pigment epithelium lifetime images after unmixing. Customized 2P-FLIO unmixing software (termed UT-2P-FLIO in GitHub) takes τ phasors as inputs, and outputs the weight of each fluorophore at each pixel, which can be used to segment photoreceptors and retinal pigment epithelium and calculate lifetime for each segmented cell. The 2P-FLIO imaging method was tested on fixed and live cells labeled with diverse fluorescent dyes, before applying it to rabbit retina samples.
Docket Number: 10046-590WO1 8300 YEH [00139] FIG. 2D shows lifetime phasors obtained from time-domain decay data, after conducting Fourier transform. For a single species, lifetime decreases clockwise along the universal semicircle. FIG. 2E shows lifetime phasor calibration using fluorescein (4 ns, excited by 780 nm fs laser) for 2P-FLIO. FIG. 2F is a ^ phasor plot of a convallaria sample. FIG. 2G shows a false-colored FLIM image of the convallaria (scale bars are 5 μm). [00140] FIGS. 2H-2J show data analysis procedure with Gaussian Mixture Models (GMM). FIG. 2H shows the 98x1 fluorescence decay curve I(t) at each pixel separately transformed into ^ phasors. FIG. 2I shows 256×256 ^ phasor sets (gτ, sτ) used as inputs for the GMM. After GMM unmixing, the corresponding clusters for photoreceptor and retinal pigment epithelium cells are selected. FIG. 2J shows photoreceptors segmentation with the watershed algorithm and retinal pigment epithelium segmentation with the k-nearest neighbors (KNN) algorithm. FIG. 2K shows a summary of lifetime changes during light-dark visual cycles (scale bar is 50 μm). [00141] Lifetime phasor analysis [00142] For lifetime phasors (^ phasors), the multi-exponential fluorescence decay at each pixel in the xyt dataset is transformed to a point (gτ, sτ) called a lifetime phasor through the digital Fourier transform (DFT), given by Eq. (1) [28]:
) ∙ 2,- Eq. (1) [00144] where Ix,y (t) is the number of the photon counts recorded in the time bin t, at the pixel location (x, y), ω is the angular frequency, f is the repetition frequency of the pulsed excitation light, and n is the harmonic number. In this disclosure, only the first harmonic frequency (n = 1) is used for generating phasors. The acquired lifetime phasor information at each pixel is then used for FLIM image generation. [00145] The .x,y and /x,y are the modulation ratio . and the phase delay / measured at pixel location (x, y) are calculated by Eq. (2) [28]. The phasor of a single- exponential decay should be positioned on the universal semicircle (FIG. 2D): [00146]
arctan
and ω=)∙2,- Eq. (2)
Docket Number: 10046-590WO1 8300 YEH [00147] The ^^>,? and %^>,? also can be expressed in the function of .x,y and /x,y as shown in Eq. (3). [00148] ^>,?(ω) = .>,?(ω)·cos[/>,?(ω)] , %^>,?(ω) = .>,?(ω)·sin[/>,?(ω)] Eq. (3) [00149] To eliminate the artifacts given by the instrument response function and the delays of the electronics, phase / and modulation . of the phasor cloud [29] are first calibrated using well-characterized dyes, such as fluorescein (lifetime of 4 ns [30]). FIG. 2E shows ^ phasor calibration procedure based on fluorescein, where . can be corrected by multiplying a constant factor α and / can be corrected by adding a constant offset Δ/ to the initial value /i, given by the following formulas: [00150] .@ = .A ∗ C Eq. (4) [00151] /@ = /A + Δ/ Eq. (5) [00152] Since the phasor cloud of a two-component mixture lies on a straight line joining the phasors of two individual components, the phasor cloud can be used to uncover the fractions of individual components at each pixel as shown in FIG. 2K and FIG. 2J. [00153] The lifetime ^ can be determined by the s^ and g^ phasors. [00154] ^^(ω) = ^ ^D;E:E, ^^(ω) = ;:
^D;E:E, ^ =G .<:(;) Eq. (6) [00155] In retinal imaging in the study, the detected fluorescence signal can come from multiple fluorophores. The fit of the decay in the time- and the frequency- domains to recover the lifetime of an unknown number of fluorophores by trial-and-error is time-consuming [35, 36]. Furthermore, when the photon counts were very low as the study imaged the endogenous retinal fluorophores, the estimation of the statistics of the time histogram was also difficult. [00156] Instead, with the phasor plot method, given the phase φ and modulation m at a given angular frequency ω, it is possible to rapidly calculate the phase lifetime τϕ and the modulation lifetime τm by simple formulas [37, 38]. [00157]
tan/ Eq. (7)
Docket Number: 10046-590WO1 8300 YEH
Eq. (8) [00159] Light - dark Visual Cycles Experimental Setup [00160] The procedure of our light-dark visual cycles experiment is illustrated in FIG. 2L which illustrates retina heating induced by near-infrared laser and white light exposure during light-dark visual cycles. The graph shows the retina’s temperature changes during light-dark visual cycles with the red circle on the graph (e.g., circle 204) denoting white light exposure. Subjects initially were dark adapted for 30 to 35 minutes. Retina autofluorescence was monitored with the 2P-FLIO while the retina adapted to stimulation from the white light exposure and from the dark cycle. Two-photon autofluorescence lifetime images were collected over 5 minutes at the end of each of the light and dark cycle intervals. For the experiment, the white light power was kept at 0.5 milliwatt (mW) and the power of the fiber-based femtosecond laser was set at 3mW. The experimental protocol is shown below: [00161] (1) The selected field of view (FOV) was first exposed to the white light for 5 minutes (red lines 201 depicted in FIG. 2L). [00162] (2) After each exposure, the retinal autofluorescence was recorded by our 2P-FLIO for 5 minutes (blue box 202, FIG. 2L). [00163] (3) Following the imaging interval, the retina prep was allowed to recover in the dark for 15 minutes (dark boxes 203, FIG. 2L). [00164] (4) After each recovery period, the retinal autofluorescence was recorded to investigate dark cycles (blue box 202, FIG. 2L (t1, t2….t6)). [00165] (5) Repeat the above-mentioned steps to image multiple light-dark visual cycles. [00166] Data Analysis Procedure with Gaussian Mixture Models [00167] An approach that leverages the 2P-FLIO excitation/detection scheme and the Gaussian Mixture Models to unmix fluorescence signals from multiple fluorophores in photoreceptors and retinal pigment epithelium cells based on their distinguished fluorescence lifetimes is detailed below. The six-step workflow is described in conjunction with FIGS. 2H-2K. [00168] (1) The 98x1 fluorescence decay curve I(t) at each pixel is separately transformed into ^ phasors by Eq. (1) (FIG. 2H). An intensity threshold and 3x3 median filter are applied to the ^ phasors to reduce the noise of the phasor location (FIG. 2I) [28,32].
Docket Number: 10046-590WO1 8300 YEH When the image size is 256x256 pixels, this creates 256×256 sets of ^ phasors (gτ, sτ). In other words, temporally resolved fluorescence detection at each pixel eventually leads to one ^ phasor sets at that pixel. [00169] (2) The 256×256 ^ phasor sets (gτ, sτ) are used as inputs for the GMM (FIG. 2I), with initial guesses on the number of species and their associated mean phasors. [00170] (3) After retrieving the unmixed ^ phasors with the GMM, we then select the corresponding cluster for photoreceptor and retinal pigment epithelium cells (FIG. 2I). [00171] (4) Photoreceptors are segmented using the watershed algorithm and mean lifetimes are calculated for each photoreceptor (FIG. 2J). [00172] (5) retinal pigment epithelium cells are segmented using the k-nearest neighbors (KNN) algorithm and mean lifetime are calculated for each retinal pigment epithelium cell (FIG. 2K). [00173] (6) Repeat step (1) to step (5) for each timepoint in the light-dark visual cycles experiments (t1 to t6) and then, summarize lifetime changes during light-dark visual cycles with box plot and histogram (FIG. 2K). [00174] Experimentally manipulating the focal planes during 2P excitation enabled selective imaging of RPE and photoreceptor layers to decipher their distinct fluorescence signatures. The study leveraged fluorescence lifetime differences and depth offset by utilizing advanced machine learning algorithms to discriminate and extract RPE and photoreceptor lifetime responses from complex FLIO datasets. [00175] The study demonstrated the 2P-FLIO function of the exemplary system on NZW and Dutch-belted (DB) rabbit’s retina preps and monitored the retinal dynamics through photoreceptor and RPE lifetime changes during light-dark cycles [4’]. The results were collected using a high NA objective. The study found that photoreceptor and retinal pigment epithelium lifetimes decreased and increased in sync with light-dark cycles in DB rabbits (FIG. 2I). [00176] The change in fluorescence lifetime with light-dark cycles was likely due to changes in the concentrations of all-trans-retinol (AT-ROL) and all-trans-retinal (AT- RAL). The temperature changes induced by the white light exposure and the excitation laser were very small (FIG. 2L), and the fluorescence lifetime changes due to light-dark cycles decreased as the biochemical processes wound down in the retina prep. This interpretation
Docket Number: 10046-590WO1 8300 YEH emphasized the central role of biochemical dynamics in modulating fluorescence lifetime changes with light-dark cycles. [00177] One possible explanation for the changes that the study measured may be as follows. When exposed to white light: (i) AT-ROL concentration inside the photoreceptor increased, while AT-RAL and 11-cis-RAL decreased [68’]. Consequently, the mean lifetime of photoreceptors increased. (ii) Simultaneously, a portion of the newly formed AT-ROL transferred back to the RPE, increasing the mean lifetime. When left in the dark: (i) AT-ROL continued to transfer from photoreceptors to the RPE, decreasing the photoreceptor mean lifetime. (ii) Additionally, the AT-ROL, which transferred back to the RPE, continued to convert into AT-retinyl ester, causing a corresponding decrease in the mean lifetime of RPE. [00178] Gaussian Mixture Models [00179] GMM is a probabilistic model widely used for clustering tasks [33]. It posits that observed data points originate from a combination of several Gaussian distributions, each representing a distinct cluster within the data. The GMM learns parameters such as mean, covariance, and weight for each Gaussian distribution, enabling the characterization of the underlying data distribution. [00180] The GMM was adapted to unmix fluorescence signals in live-cell imaging based on fluorescence lifetime of retinal fluorophores. The GMM assumes that the observed fluorescence lifetime phasor plot can be represented as linear combinations of Gaussian components. GMM aims to estimate the parameters (means, covariances, and weights) that best describe the observed fluorescence lifetime distribution. The estimation is done using the Expectation-Maximization (EM) algorithm [34], which iteratively maximizes the likelihood of the observed data. Each Gaussian component within the GMM corresponds to a specific fluorophore in the sample. The mean of a Gaussian component signifies the fluorescence lifetime and the emission spectrum of the corresponding fluorophore, while the weight represents the proportion or abundance of that fluorophore in the mixture. To build up the input dataset for our GMM, the fluorescence I(t) for each pixel is transformed into ^ phasors to form a vector of 2 features: (gτ, sτ) - ^ phasor. [00181] Estimating the number of distinct fluorophores in a sample is critically challenging. To address this issue, we establish an automated approach that determines the optimal number of clusters using the Bayesian Information Criterion (BIC) [35,36]. By fitting the GMM with varying numbers of clusters, we compute BIC values and select the model that best balances goodness of fit and model complexity. This automatic estimation of cluster
Docket Number: 10046-590WO1 8300 YEH numbers gives more flexibility to our approach. For rabbit retina preps, BIC algorithm suggested that the model with 3 clusters offer the best balances goodness of fit and model complexity, as corresponding to autofluorescence from photoreceptors, retinal pigment epithelium cells, and background (FIG. 2H and FIG. 2I). [00182] The GMM was fitted with an expectation–maximization (EM) algorithm [37,38], which iteratively estimates the parameters of the Gaussian components and assigns phasor points to clusters based on the estimated probabilities. This process continues until convergence, and the resulting model outputs the unmixing results at each pixel. Each pixel has a weight for each of the clusters/fluorophores, where the largest weight determines the assigned cluster of that pixel. [00183] The unmixed clusters of data points within those clusters are visualized with ^ phasor plots, unmixed intensity image of each cluster (photoreceptors and retinal pigment epithelium cells) (FIG. 2J and FIG. 2K). This comprehensive visualization allows us to intuitively assess the quality of unmixing and the accuracy of photoreceptors and retinal pigment epithelium cells identification. [00184] Two-dimensional (2D) micro-electromechanical system-based (MEMS) scanner [00185] FIG. 2M shows the exemplary system (i.e., mFLIO system) comprising a 2P-FLIO module 250 and a spectral-domain optical coherence tomography (SD-OCT) module 252. As shown, the 2P-FLIO module was configured with a 2D MEMS scanner 256, reducing the acquisition time required to determine fluorescence lifetimes, tracking the lateral movement of the living eye, and stabilizing the 2P excitation in depth. The MEMS scanner 256 was part of a high-resolution (0.04 arcmin), ultrafast (I kHz), wide- field (300°/s across 8° view angle) retinal eye-tracking system 254 that provided multimodal 2P-FLIO (mFLIO) measurements in anesthetized rabbits. [00186] Eye motion may deteriorate the mFLIO measurements. Even during anaesthetization and stable fixation, eye movements can still fluctuate in magnitudes and frequencies [71’], [73’]. Current state-of-the-art methods to track eye movements include suction caps [74’], scleral search coils [75’], and the Purkinje images [76’], but they may experience a poor signal-to-noise ratio (SNR) or involve invasive components (e.g., contact lenses) [72], [73], [75]. Other state-of-the-art non-invasive methods (e.g., video-based tracking) may experience low image sampling rate, distortion by motion artifacts in reference frame [77], [78], or limited detectable range and scanning velocities [79-81].
Docket Number: 10046-590WO1 8300 YEH [00187] The exemplary system, to minimize fluctuation in magnitudes and frequencies of the measurements, estimated eye motion only based on a small 1 mm × 1 mm region of a full frame (denoted as a subframe), thus lowering acquisition times and speeding up the computation of eye displacements. As for the reference frame issues, the exemplary system quantified eye displacements using the shifts of a subset of frames in a sequence spanning the full acquisition cycle, bypassing the need for a single reference frame and providing the precise measurement of eye movements exceeding the spatial extent of single acquired frames. [00188] As shown in FIG. 2M, the 2P-FLIO module 250 used a fiber-based femtosecond laser 258 (e.g., CFL-04RFF, Calmar Laser), providing 90-fs pulses at the central wavelength of 780 nm with pulse repetition frequency (PRF) at 80 MHz, which was suitable for two-photon excitation of endogenous retinal fluorophores. The excitation beam was then guided through a prism pair compressor 260 to pre-compensate for the chromatic dispersion by subsequent optical elements and the eye itself. As a result, 100-fs pulses with close to transform-limited temporal shape were delivered to the retinal plane. After dispersion pre- compensation, the beam entered a mFLIO with x-y galvanometer scanners 262 and 264 (shown as GM1 and GM2) and a telescope 266 relaying the GM1’s plane to the pupil plane 267 of the eye or the front focal plane of the 60x, NA 1.2 silicone immersion objective (e.g., UPLSAPO60XS2, Olympus). [00189] The 2P-FLIO module 250 provided simultaneous frame registration in 2 channels, non-descanned fluorescence, and reflectance. Reflectance images were obtained from a 785-nm laser diode 268 and served to adjust eye position before imaging as guidance for subsequent alignment of fluorescence frames and to correct motion artifacts within the frame. The two-photon excited fluorescence (TPEF) emission was collected with a 500-720 nm emission filter by a cooled low dark count photomultiplier tube (PMT) 269 (e.g., H74229-40, Hamamatsu) and amplified with 2 GHz cutoff bandwidth preamplifiers (e.g., HFAC-26, Becker and Hickl GmbH). The amplified signal was then measured and correlated to the reference clock of the femtosecond laser with a time-correlated single photon counting (TCSPC) module 270 (e.g., PicoHarp 300, PicoQuant). Collected TPEF photons were assigned to one of the 98-time bins for each pixel, depending on an arrival time with respect to the synchronization pulse of the femtosecond laser. [00190] For a given average power, the TPE fluorescence rate may be inversely proportional to the pulse repetition frequency (PRF) [82’ – 84’], so the study developed a
Docket Number: 10046-590WO1 8300 YEH TPE imaging system with adjustable PRF to optimize the TPE fluorescence rate with minimal laser power. Reducing PRF at an average excitation power translated to higher peak power, resulting in a higher fluorescence yield [30’], [85’]. The study employed a pulse picker system 272 (e.g., Pulse selection system, model 305, Conoptics) to choose PRF within the 1 to 10 MHz range. The 10 MHz PRF used resulted in an over 10-fold increase in fluorescence signal with respect to the original 80 MHz PRF of the fiber-based femtosecond laser 258. With the increased efficiency of TPE fluorescence generation from intrinsic retinal fluorophores, the study scanned a 300 µm × 300 µm × 50 µm volume of the retina to obtain a point measurement of fluorescence lifetime (FLT) in less than 4 seconds (Fig. 2I). [00191] 2D MEMS scanner with two-photon FLIO (2P-FLIO) imaging. FLIO is a tool for investigating the human retina in both normal and diseased eyes. The motion of the human eye, characterized by constant, involuntary, microscopic movements during fixations, is an issue for high-resolution fluorescence lifetime imaging [86’], [87’]. These eye movements may cause the scanned field of the FLIO to traverse the retina, mirroring the eye motion pattern continuously. While fixational eye movements in a normal eye are of small amplitude, individuals with retinal diseases or impaired vision may experience amplified movements, leading to significant distortions in FLIO frames [88’]. FLIO imaging is hindered and, in some cases, rendered impossible due to these movements. There is a need, especially in clinical imaging, to minimize or eliminate this motion. [00192] The two-photon fluorescence signal was too weak to enable accurate estimation of the motion from frame to frame, so the study simultaneously collected a high signal-to-noise ratio reflectance video of vascular structures in the inner retina (e.g., Reflectance SLO 254 in FIG. 2M). The contrast due to the vessels provides the post- processing registration signal from which eye motion can be corrected. The reflectance SLO 254 (i.e., eye-tracking system 254) used a 785-nm laser diode 268 (e.g., LP785-SAV50, Thorlabs) to minimize the laser exposure to the retina. The pellicle beam splitter (BS) 274 reflected the beam and directed it onto a 2D scanning mirror 256 with a 1-mm microelectromechanical system (MEMS) based active aperture (e.g., VC3141/5/48.4, VarioS 2D microscanner, Fraunhofer IPMS). After reflecting off the MEMS scanning mirror 256, the 785 nm beam passed through a 4f telescope system composed of 2 achromatic doublets. The telescope conjugated the MEMS scanner’s aperture with the galvanometer scanners 262 and 264 (e.g., GM2), which steered the position of the scanning pattern to the selected region of interest in the retina.
Docket Number: 10046-590WO1 8300 YEH [00193] The conjugate plane of the MEMS scanning mirror 256 was then imaged onto the eye pupil plane 267 by the same light path as 2P-FLIO. The beam reflected off of the retina reverted to the same path, was de-scanned by the MEMS scanner 256, passed through the pellicle beam splitter 274, and was collected by an avalanche photodiode 278 (e.g., MPD-SPAD, PicoQuant) with a 100 µm confocal pinhole 276 (e.g., PH 276, Reflectance SLO 254 in FIG. 2M). [00194] FIG. 2N shows the 2D MEMS scanner having a Lissajous scanning pattern and the raster scanner. The Lissajous pattern, formed by the intersection of two perpendicular sine waves with different frequencies, can cover the imaging area with fewer data points compared to raster scanning [73’], [89’]. In subpanel (a), by collecting reflectance images at 300 frames per second using the MEMS scanner with Lissajous pattern, the 2P- FLIO module tracked the eye moment efficiently and compensated for the motion in real- time. The reflectance images were processed in real-time using a combination of optical flow and feature-based tracking algorithms. The optical flow algorithm estimated pixel-wise motion between consecutive frames, providing a dense motion field. Simultaneously, feature- based tracking detected distinct landmarks [79’] (e.g., vessel bifurcations, microaneurysms) using techniques like Scale-Invariant Feature Transform (SIFT) or Speeded-Up Robust Features (SURF). These features were tracked across frames to estimate eye movement. The estimated motion was used to adjust the field of view in real-time. The Galvo scanner (i.e., GM2) was controlled to counteract the estimated eye movement. By modulating the scan position, the FLIO imaging system can maintain a stable scanning field, fixating on the targeted retinal region despite ongoing eye movements due to ventilation, the cardiac cycle, and retinal drift [90’]. [00195] Validation of 2D MEMS scanner. The study achieved a 10-fold increase in the signal-to-noise ratio (SNR) of fluorescence signals at an optimized pulse repetition frequency of 10 MHz compared to the original 80 MHz. Additionally, the study completed a 300 µm × 300 µm × 50 µm volume scan in under 4 seconds. Furthermore, the exemplary system demonstrated a reduction in motion artifacts in FLIO frames compared to pre-correction conditions. The MEMS scanner provided an improvement in image stability as measured by reduced pixel displacement between consecutive frames, resulting in clear and distortion-free retinal reflectance and FLIO images. [00196] Potential problems and alternative solutions. The animal experiments in the study demonstrated that the anesthetized rabbits had slow eye movements over approximately six degrees of visual angle. These movements should be compensated. Eye
Docket Number: 10046-590WO1 8300 YEH movement may also be problematic in clinical measurements. A clinical instrument can use a fixation target to limit large-angle movements but micro saccades remain. To overcome this issue, a feedback-control circuit can be employed to monitor the actual position of the galvanometric mirrors (e.g., GM1 and GM2) and compare them with the eye-tracking measurements. The study imitated slow movements and micro saccades using an artificial eye and sequences of horizontal and vertical back-and-forth movements. [00197] Spectral-Domain Optical Coherence Tomography (SD-OCT) Module [00198] Integration of FLIO and OCT modules. The study developed and validated multiplexing methods synchronizing FLIO and OCT imaging in a common optical path, providing precise depth co-registration. [00199] FIG. 2O shows an active feedback autofocus optical coherence tomography (AFOCT) unit, as a block diagram, comprising controlled modules (e.g., 2P- FLIO module 250, SD-OCT module 252 in FIG. 2M) and a microcontroller 280. The AFOCT unit utilized a feedback mechanism to maintain optimal focus during imaging by adjusting the optical power of OCT and mFLIO beams using an electrically tunable lens 282 (ETL) (shown as 282 in FIG. 2M). The ETL 282 (e.g., EL-16-40-TC, OptoTune AG, Switzerland) provided variable power (−6 to +10 dpt) over an 11.6 mm diameter clear aperture with a 3 ms response time. The OCT autofocus feature was essential in the exemplary system to compensate for many variations, including the thickness of retinal layers and motion caused by factors such as ventilation, the cardiac cycle, and retinal drift. [00200] In FIG. 2O, at step 302, the microcontroller analyzes OCT A-scans for depths and amplitudes of the vitreous-RNFL and retinal pigment epithelium (RPE). [00201] At step 304, the microcontroller determines the optimal power of the electronically tunable lens (ETL) to ensure the 2P beam is correctly focused on targeted layers of the retina. In other words, the microcontroller dynamically adjusts the focal plane of the 2P beam based on each OCT A-scan. [00202] At steps 306 and 308, when the targeted layers of the retina are not in the focal plane of the 2P beam (e.g., focal error), the microcontroller computes the change in optical power and adjusts the control current to the ETL, changing the position of the ETL. [00203] In FIG. 2M, the spectral-domain OCT module 252 used a broadband light source 284 (BLS) (e.g., EXS210022-03, EXALOS) with a center wavelength of 840 nm and a bandwidth of 50 nm. This wavelength was close to the 2P excitation wavelength to minimize optical dispersion. Output from the source was coupled, via a fiber coupler 286, to an interferometer setup, in which light was split to the reference arm 288 and sample arm 290
Docket Number: 10046-590WO1 8300 YEH by a fiber splitter 286 (i.e., fiber coupler (FC)) (e.g., TW850R5A2, Thorlabs) in the ratio of 50/50, respectively. In the sample arm 290, light delivered via the fiber was collimated by collimator 292 (denoted as C), reflected on DM3294, and then combined with a two-photon excitation beam. Sharing the same galvanometer GM2264, the two-photon laser and OCT laser were delivered and focused on the same area 267 of the rabbit retina through a telescope configuration. Reflected light from the retina was collected back and traced in the reverse direction of the OCT illumination beam to the sample arm fiber 290. Light reflected from the reference and sample arms travels back to the beam splitter 286 and recombines to generate an interference pattern captured by a customized high-speed spectrometer 296. In the spectrometer 296, a transmission diffraction grating 298 (DG) (e.g., 1800 lines/mm, Wasatch Photonics) dispersed the interfered light, and a high-speed complementary metal-oxide- semiconductor (CMOS) line scan camera 300 (LSC) (e.g., raL2048-80km, Basler, Germany) captured raw fringe signals. The theoretical axial resolution of the SD-OCT module was 6.2 µm in air. OCT A-Scan rate was 80 kHz. OCT images were displayed in real-time with standard SD-OCT signal processing, and the images also served as input for the AFOCT unit. [00204] FIG. 2P shows the axial resolution of a 2P-FLIO overlay with SD- OCT. As shown, the exemplary system, via the AFOCT unit, detected the location of maximum intensity on the A-Scan image, which corresponded to the RPE layer location, and then adjusted the focal plane accordingly. As a result, the focal plane of both OCT and mFLIO can be kept in a specific retina layer without any mechanical movement, at rates up to hundreds of Hertz. The SD-OCT module required only reflective optics and can be implemented at a fraction of the cost required for a comparable piezo-based actuator. In the study, the OCT and mFLIO maintained the best possible focus at a specific retinal layer throughout the imaging process. [00205] FIG. 2Q shows (i) a customized eye model used for calibration and (ii) high-quality A-Scan, B-Scan, and en-face images generated by the SD-OCT module of the exemplary system. The exemplary system may be useful in applications where precise imaging is crucial, such as in ophthalmology for retinal imaging or in other medical fields for imaging various tissues. The active feedback autofocus enhanced the efficiency and reliability of OCT systems, contributing to improved image quality and diagnostic accuracy. [00206] Validation of multimodule imaging. The combined system (i.e., exemplary system) was tested first by imaging fluorescent microsphere samples, which were made by immobilizing yellow-green fluorescent microspheres in 3D with 2% agarose gel. These samples were prepared by mixing two different-sized microspheres (4 μm and 10 μm
Docket Number: 10046-590WO1 8300 YEH in diameter) and embedded into a customized eye model (subpanel a). These samples were imaged, and results were shown as an overlay of 2P-FLIO and SD-OCT images, which were color-coded as blue and red colors, respectively. This overlay image were used to evaluate the co-localization of the 2P-FLIO and SD-OCT modules. Individual microspheres were co- registered in both images overall, though some spheres appeared only in an OCT image. These microsphere images were used as calibration data to co-register 2P-FLO and SD-OCT images. The study validated the effectiveness and utility of the integrated multimodal system (FLIO and OCT) in visualizing and characterizing photoreceptors and retinal pigment epithelium layers in vivo in rabbit eyes. Furthermore, comparative analyses between OCT and FLIO images was conducted to evaluate the complementary advantages and limitations of each modality for precise assessment and understanding of light-dark cycles in the retina. [00207] In the study, the ETL of the exemplary system dynamically adjusted the focal plane in response to changes in the retinal environment, ensuring continuous and optimal focus throughout the imaging session. The study set an optimal focus in at least 95% of imaging frames as a benchmark to validate the performance of the active feedback autofocus mechanism, which was crucial for the integration of FLIO and OCT modules. The ability to maintain consistent focus in the presence of physiological variations was pivotal for the accuracy and precision of the multimodal imaging system, enhancing its potential for applications in ophthalmology and other medical fields. [00208] Potential problems and alternative solutions. Variability in retinal thickness and motion may lead to occasional frames being out of focus, compromising the overall image quality and co-registration between FLIO and OCT modules. The study implemented a post-processing algorithm to identify and correct frames that fell out of focus during the imaging session. This algorithm analyzed each frame, detected deviations from the optimal focus, and applied corrective measures to enhance focus retrospectively. Real-time retinal tracking and fast point measurement should minimize lateral motion artifacts. [00209] Scan Pattern and Timing Optimization [00210] Safety and scan pattern. Retinal damage from laser exposure can be categorized as photothermal, photoacoustic, and photochemical. [00211] Non-linear laser exposure can produce retinal damage by any of the three mechanisms. Preliminary data on multiple imaging sessions with the same animal showed no ophthalmoscopic visible retinal lesions by direct ophthalmoscopy. Preliminary data on thermal measurements on retinal mounts shows no thermal response at 5 mW (FIG. 2L). Photochemical damage was unlikely at 780 nm excitation beam wavelength.
Docket Number: 10046-590WO1 8300 YEH [00212] There may be photochemical or photoacoustic retinal damage, so the safety thresholds were determined in the study. The key was widefield fundus autofluorescence measured longitudinally over repeat imaging sessions. The study correlated subsequent wide-field fundus autofluorescence images with the initial virgin image to measure retinal damage due to the excitation laser. OCT images was used to build a 3D representation of the retina and RPE and analyzed for structural abnormalities. Additional histology was obtained if OCT was abnormal. The planned visual cycle measurements were a verification of the integrity of the photoreceptor-RPE function and, as such, were the most sensitive threshold measure of any laser-induced damage. If the probe laser were to damage the photochemistry of the visual cycle, it may be impossible to measure repeat light-dark cycles. In the experiments, the study focused on the measurement of the amplitude of the light-dark cycles at a specific location of the retina over a minimum of 10 cycles. A decrease in the amplitude may indicate that the probe beam was interfering with the biochemistry of the visual cycle. Validation of laser safety in mFLIO may be an important contribution to the future application of this technique in clinics. [00213] Laser safety and data acquisition protocol were interrelated. A minimum photon count at each pixel was required to calculate the FLT, but the continuous exposure of the excitation laser at any spot cannot exceed the permissible threshold. The study met the exposure threshold by scanning the excitation laser over a region of interest (ROI) that satisfied the retinal exposure safety limits and allowed the fastest point calculation of FLT. [00214] The raster scan setup for single-point FLT measurement in the study was set as 10x10 pixels with pixel dwell time set at 4 ms and combined 10 frames to get a final measurement (total acquisition time of 4 s), which was a conservative approach to perform safe 2-photon imaging of the human retina with a conservative approach. The aberrations of the eye's optical system led to variations of the spot size on the retina, which may be from 5 to 30 µm [92’]. To calculate the most conservative laser safety requirement for the exemplary system, the study calculated the laser safety requirement based on 5 µm spot size. The study referred to an article by François C. Delori et al. [93’] and ANSI 2000 for maximum permissible exposures for ocular safety. The study calculated the safety limit for both the single pulse limit (Rule 1) and the average power limit (Rule 2). [00215] Rule 1. The MPφav was derived from the single pulse MPφ, using cell = 4a of Table 3 [93’]: [00216]
0.8 W = 667 mW
Docket Number: 10046-590WO1 8300 YEH [00217] Using δ = Ft1=8×10-6, CT = 100.002(780 - 700) = 1.445, CJ = 1 [Table 2 (lower part)], CE = 1 [Table 2 (upper part)] [93’]. [00218] Rule 2. The thermal “average-power” limit is the MPφ of a continuous exposure of T = 4 ms duration (cell 2, Table 5) and was given in cell 4a of Table 3: MPφav,2 = MPφ[T] = {6.93×10-4 CT CE T-0.25} = 4×10-3 W = 4 mW. [00219] The power safety limit for most extreme cases of the 2P-FLIO was 4 mW. Based on the preliminary data from live rabbit experiments with the laser power set at the limit, the study got 10000 photon counts per second (cps) by collecting multiple frames of 10x10 pixels for 4s and 400 photon counts per pixel for FLT fitting (laser repetition rate set at 80 MHz). If the SNR was good, the study got the lifetime for each pixel and calculated the mean and variance for that ROI. [00220] Because of the nonlinear dependence on pulse peak power [33’], [83’], by reducing the laser repetition rate to 8 MHz, the study set the laser average power at 0.4 mW, one order lower than the established laser safety limit [93’], and got the same photon counts per second. The 758 nm diode laser for eye tracking is set at 100 μW, which was below the safety exposure limits [73’], [93’]. [00221] Thermometry in the retina prep during light-dark visual cycles. the study used thermocouple probes (e.g., IT24P, Physitemp) to measure temperature changes induced by two-photon microscopy and white light exposure in the rabbit retina prep4 (subpanel a, FIG. 2L). Flexible thermocouple probes were mounted in rigid capillary glass, leaving 2 mm of the probe exposed at the tip. The study characterized heating as a function of two-photon laser power following each step of light-dark visual cycles (subpanel b, FIG. 2L). The thermocouple was inserted 100 µm below the retina surface using a micromanipulator. The retina prep temperature was monitored for 1.5 hours during light-dark visual cycle experiments. The study performed the thermometry measurement for three laser power, 100 mW, 20 mW, and 5 mW, with the white light power set at 0.5 mW. In subpanel b, Fig. 2L, with the lower laser power settings (20 mW and 5 mW), there was no temperature change during the light-dark visual cycles experiment. When the femtosecond laser power was set at 100mW, the maximum temperature difference was 0.9°C. The minimum temperature difference was 0.1°C when the femtosecond laser power was set at 5 mW. Accordingly, the study set the 2P excitation laser power at 4 mW to minimize the temperature change during visual cycles. [00222] Timing of light-dark cycles. The exemplary system safely measured a point FLT in the retina in less than 4 seconds. The previous studies required 5 minutes.
Docket Number: 10046-590WO1 8300 YEH Because the FLT measurement was no longer the limiting factor in data acquisition, the study investigated and optimized the timing of light-dark cycles. The limiting factor was the current physiological process being measured and not the sampling time limitation of the exemplary system. [00223] FIG. 2R shows example life-dark cycle measurements. The study investigated the responses in different retinal locations to identify the light stimulus that gave the highest SNR. It may be possible to optimize the stimulus for specific retinal locations depending on the density and type of photoreceptors. The FLT response to light-dark cycles had both an amplitude and decay time constant. FLT response recorded during light-dark cycles was fitted with a single exponential decay function: [00224] ?(^) = ?^ + N O>P (−^/^) Eq. (9) [00225] where ?(^) is the FLT amplitude at a given time, ?^ is the initial amplitude right after white light exposure, N is the delta between the plateau amplitude and the initial amplitude, O>P is the exponential symbol, ^ is time after white light exposure, and tau (^) is the time constant of the decay function. Both measures (^ and N) may be biomarkers for photopigment regeneration, and the optimum stimulus paradigm may be specific to the retinal location. The study cannot determine the optimum stimulus without the exemplary system and making measurements in vivo retina. There can be an infinite combination of stimulus light amplitude, duration, and wavelengths. [00226] Evaluating the correlation between photoreceptor and retinal pigment epithelium lifetimes with light-dark visual cycles in time domain and frequency domain [00227] In time domain, normalized cross-correlation methods are employed to evaluate the similarities and correlation between photoreceptor and retinal pigment epithelium lifetime response functions (the red dashed line in FIG. 4A, FIG. 4B, FIG. 4C and FIG. 5A, FIG. 5B, and FIG. 5C) and the light-dark exposure function (the red solid line in FIG. 4A, FIG. 4B, FIG. 4C and FIG. 5A, FIG. 5B, and FIG. 5C). The lifetime response functions are calculated from interpolating the line equation between the mean lifetime of photoreceptors and the mean lifetime of retinal pigment epithelium cells at two consecutive time points. The lifetime response functions and the light-dark exposure function were first normalized, and then cross-correlation was applied to identify the lag at which the correlation is maximized [39, 40].
Docket Number: 10046-590WO1 8300 YEH [00228] The cross-correlation between the lifetime response functions and light-dark exposure functions is defined as: [00229] RSTSU(V) = W-^(^)-X(^ + V) Eq. (10) [00230] where E[] is the estimation operator, -^ is the light-dark exposure function, -X is the lifetime response function of the photoreceptor or retinal pigment epithelium, and V is the displacement in time or lag. Assuming ergodicity, for single time- limited realizations of each random process, this is determined using the integral: [00231] RSTSU(V) = YZ [Z -∗ ^ (^)-X(^ + V)V^ Eq. (11) [00232] where -^ ∗ denotes the complex conjugate of -^(t). Cross-correlation functions are unbounded measures and are typically normalized by the values of the autocorrelations at zero lag to bound the estimate between -1 and 1. The autocorrelation functions are the time domain equivalent of the auto power spectra and their value at zero lag represents the total energy in the signal. The normalized and bounded measure is known as the cross-correlation coefficient, \-^-X(V), which provides a measure of the linear association between the two signals at a given time lag and is given by: [00233] \-^-X(V) = ]STSU(^) Eq. _]STST(^)]SUSU(`) (12) [00234] The result is displayed along with the cross-correlation coefficient plot (FIG. 4D, FIG. 4E, FIG. 4F, and FIG. 4H - Time Domain). The higher the cross-correlation coefficient peak (Xcorr. peak) shows stronger photoreceptor and retinal pigment lifetime response to light-dark visual cycles. [00235] The correlation relationship between lifetime response functions and the light-dark exposure function was also evaluated in the frequency domain. The functions are first normalized and transformed to frequency domain representations using the Fast Fourier Transform (FFT) algorithm. The magnitude spectra of these transformed signals are then computed by taking the absolute values of their respective FFT results (FIG. 4D, FIG. 4E, FIG. 4G, and FIG. 4I - Frequency Domain). [00236] ^^ = |bcc(-^(^))| Eq. (13)
Docket Number: 10046-590WO1 8300 YEH [00237] ^X = |bcc(-X(^))| Eq. (14) [00238] To quantify the similarity between the magnitude spectra, the Pearson correlation coefficient is calculated [41,42].
(15) [00240] Where \^^^X is the Pearson correlation coefficient, ^^ is magnitude spectra of the light-dark exposure function in frequency domain, ^X is magnitude spectra of the lifetime response function in frequency domain of photoreceptor or retinal pigment epithelium, ^eee^ is mean of ^^, and
is mean of ^X. The coefficient ranges from -1 (perfect negative correlation) to 1 (perfect positive correlation). The higher the correlation coefficient (Corr. Coeff), the stronger photoreceptor and retinal pigment lifetime response to light-dark visual cycles. [00241] Rabbit surgical procedure [00242] The experimental procedures adhere to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were conducted under the IACUC protocol AUP-2021-00191. New Zealand White (NZW) and Dutch-Belted (DB) rabbits weighing 4 kilogram (Kg) and 2.2 Kg respectively were pre-medicated with a mixture of 5mg/kg Ketamine and 20mg/kg Xylazine and euthanized using 4ml of diluted Euthasol IV. The rabbits were immediately enucleated. Each eye was then surgically prepared for imaging. A stab incision was made 3 mm posterior to the limbus using a BP #11 blade, and a 360- degree peritomy was performed using the curved corneal-scleral scissors. The anterior segment, consisting of the ciliary body, lens, and cornea, was then removed, leaving an intact eye cup. The eye cup was then divided into four quadrants with the optic streak at the apex using Wescott scissors. The vitreous was surgically dissected from the anterior surface of the retina. Careful attention was taken to preserve the retinal attachment to the retinal pigment epithelium. Each quadrant consisting of sclera, choroid, RPE, and retina was placed on a microscope slide and a cover slip was placed on top. The full-thickness retina prep was then imaged. FIG. 9A shows hematoxylin and eosin stain retinal histology from New Zealand White rabbit. The H&E stain histology results shown in FIG. 9A were used to demonstrate the integrity of our retina prep, as the photoreceptors and retinal pigment epithelium stay intact with normal cell structure. [00243] Results
Docket Number: 10046-590WO1 8300 YEH [00244] Retina Heating Induced by Near-Infrared Lasers and White Light Exposure During Light-Dark Visual Cycles [00245] Flexible thermocouple probes (IT24P; Physitemp) were mounted in rigid capillary glass, leaving 2 mm of the probe exposed at the tip (FIG. 2L). The maximum diameter of the probe entering the retina prep was 220 μm. The retina prep and environment temperature were recorded simultaneously with a two-channel thermometer/calibrator (CL3515R; Omega). [00246] Thermometry in the retina prep during light-dark visual cycles: In the current study, thermocouple probes were used to measure temperature changes induced by two-photon microscopy and white light exposure in the rabbit retina prep. Heating was characterized as a function of two-photon laser power following each step of light-dark visual cycles (FIG. 2L). The thermocouple was inserted 100 μm below the retina surface using a micromanipulator. The retina prep temperature was monitored for 1.5 hours during light-dark visual cycles experiments. The thermometry measurement was performed for three laser power, 100 mW, 20 mW and 5 mW, with the white light power set at 0.5 mW. [00247] As shown in FIG. 2L, with the laser power set at 100 mW, the temperature increases when exposed to two-photon excitation during imaging time and decreases to the base line temperature when exposed to white light or resting in the dark. With the lower laser power settings (20 mW and 5 mW), no significant temperature change was seen during the light-dark visual cycles experiment. For the 100mW laser, the average temperature change was 0.5°C; for the 20 mW laser, the average temperature change was 0.2°C; and for the 5 mW laser, the average temperature change was 0.1°C. Following this experiment, the laser power was set at 5 mW to minimize the temperature change during light-dark visual cycles. [00248] Photoreceptor and Retinal Pigment Epithelium Lifetimes Decrease and Increase in Sync With Light-Dark Cycles in Dutch-Belted Rabbits [00249] We collected the photoreceptors and retinal pigment epithelium cells lifetime changes during light-dark visual cycles on multiple retina preps from 2 DB rabbits and 2 NZW rabbits. We follow the light-dark visual cycles experimental setup described above to collect 8 data sets to quantify differences between the 2 species. [00250] FIG. 3A-M show results from a first Dutch-Belted rabbit – retina prep. FIG. 3A shows an example white light exposure scheme. FIG. 3B and FIG. 3C show photoreceptor lifetime changes. FIG. 3D and FIG. 3E show retinal pigment epithelium lifetime changes during light-dark visual cycles. FIG. 3F, FIG. 3G, FIG. 3H, and FIG. 3I
Docket Number: 10046-590WO1 8300 YEH quantify photoreceptor changes during light-dark visual cycles with normalized cross- correlation method in the time domain and correlation coefficient in the frequency domain. FIG. 3J, FIG. 3K, FIG. 3L, and FIG. 3M quantity retinal pigment epithelium lifetime changes during light-dark visual cycles with normalized cross-correlation method in the time domain and correlation coefficient in the frequency domain. [00251] FIG. 4A-M show results from a second Dutch-Belted rabbit – retina prep. FIG. 4A shows an example white light exposure scheme. FIG. 4B and FIG. 4C show photoreceptor lifetime changes. FIG. 4D and FIG. 4E show retinal pigment epithelium lifetime changes during light-dark visual cycles. FIG. 4F, FIG. 4G, FIG. 4H, and FIG. 4I quantify photoreceptor changes during light-dark visual cycles with normalized cross- correlation method in the time domain and correlation coefficient in the frequency domain. FIG. 4J, FIG. 4K, FIG. 4L, and FIG. 4M quantity retinal pigment epithelium lifetime changes during light-dark visual cycles with normalized cross-correlation method in the time domain and correlation coefficient in the frequency domain. [00252] FIG. 5A-M show results from a first NZW rabbit – retina prep. FIG. 5A shows an example white light exposure scheme. FIG. 5B and FIG. 5C show photoreceptor lifetime changes. FIG. 5D and FIG. 5E show retinal pigment epithelium lifetime changes during light-dark visual cycles. FIG. 5F, FIG. 5G, FIG. 5H, and FIG. 5I quantify photoreceptor changes during light-dark visual cycles with normalized cross- correlation method in the time domain and correlation coefficient in the frequency domain. FIG. 5J, FIG. 5K, FIG. 5L, and FIG. 5M quantity retinal pigment epithelium lifetime changes during light-dark visual cycles with normalized cross-correlation method in the time domain and correlation coefficient in the frequency domain. [00253] FIG. 6A-M show results from a second NZW rabbit – retina prep. FIG. 6A shows an example white light exposure scheme. FIG. 6B and FIG. 6C show photoreceptor lifetime changes. FIG. 6D and FIG. 6E show retinal pigment epithelium lifetime changes during light-dark visual cycles. FIG. 6F, FIG. 6G, FIG. 6H, and FIG. 6I quantify photoreceptor changes during light-dark visual cycles with normalized cross- correlation method in the time domain and correlation coefficient in the frequency domain. FIG. 6J, FIG. 6K, FIG. 6L, and FIG. 6M quantity retinal pigment epithelium lifetime changes during light-dark visual cycles with normalized cross-correlation method in the time domain and correlation coefficient in the frequency domain. [00254] It was observed that the photoreceptors’ lifetime in both DB rabbits and NZ rabbits range from 0.5 to 3 ns. The retinal pigment epithelium cells have a much
Docket Number: 10046-590WO1 8300 YEH lower lifetime, which range from 0 to 1.5 ns. As shown in FIG. 3A-3M and FIG. 4A-4M, the photoreceptor and retinal pigment epithelium lifetimes decrease and increase in sync with light-dark cycles in Dutch-Belted rabbits. This relationship was clearly shown in the shape of the lifetime response function of the photoreceptor (FIG. 3A-M and FIG. 4A-M) and the lifetime response function of retinal pigment epithelium cells (FIG. 3A-M and FIG. 4A-M). The relationships are shown in both time domain cross-correlation analysis and frequency domain correlation coefficient analysis as the correlation coefficients were higher than 0.5 for both photoreceptors and retinal pigment epithelium cells’ lifetime (FIG. 3A-M and FIG. 4A- M). [00255] Photoreceptor and Retinal Pigment Epithelium Lifetimes Gradually Increase with Light-Dark Cycles in New Zealand White Rabbits [00256] As shown in FIG. 5A-M and FIG. 6A-M, the photoreceptor and retinal pigment epithelium lifetimes gradually increase with light-dark cycles in New Zealand White rabbits. This relationship was clearly shown in the shape of the lifetime response function of the photoreceptor (FIG. 5A-M and FIG. 6A-M) and the lifetime response function of retinal pigment epithelium cells (FIG. 5A-M and FIG. 6A-M). The relationships render a much lower time domain cross-correlation peak (0.1-0.3). However, the frequency domain correlation coefficients are still in the high range (0.4-0.7) (FIG. 5A-M and FIG. 6A-M). However, the retinal pigment epithelium cells have lower correlation coefficients than photoreceptors, meaning lower fluorescence response to light-dark visual cycles. [00257] The correlation between photoreceptor and retinal pigment epithelium lifetimes with light-dark visual cycles degenerate as the enucleated samples die in both Dutch-Belted rabbits and New Zealand White rabbits. [00258] As shown in FIG. 3B, FIG. 3D, FIG. 3F-G, and FIG. 3J-K, for example, the correlation between photoreceptor and retinal pigment epithelium lifetimes with light-dark visual cycles degenerate as the enucleated samples die in both Dutch-Belted rabbits and New Zealand White rabbits. This relationship was clearly shown in the shape of the lifetime response function of photoreceptor (b) and the lifetime response function of retinal pigment epithelium cells (c). The relationships render a significant decrease in both time domain cross-correlation peaks and the frequency domain correlation coefficients (d, e) FIG. 7 is a summary table of photoreceptor lifetime changes and retinal pigment epithelium lifetime changes during light-dark visual cycles. [00259] Additional in vivo rabbit surgical procedure and results
Docket Number: 10046-590WO1 8300 YEH [00260] The study further imaged nine rabbit eyes in vivo. All the experimental procedures were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and approval of the IRB of The University of Texas at Austin. Butch Belted Rabbits weighing about 2.5 kg were anesthetized with a mixture of ketamine (40 mg/kg) and xylazine (5 mg/kg) via intramuscular (IM) injection. Following this, anesthesia was maintained with 2% isoflurane and oxygen using a V-Gel® (e.g., D10004, Jorgensen Laboratories, Loveland, CO). Before the experiment, a subcutaneous injection with one dose of meloxicam (0.2 mg/kg) was administered to the rabbit for pain (NSAID). The pupil of the eye to be imaged was dilated with topical application of phenylephrine hydrochloride 2.5% and tropicamide 1% eye drops. Topical tetracaine drops were applied for topical anesthesia prior to the initiation of the experiments. To evaluate anesthesia levels and temperature, continuous monitoring of the heart rate and respiratory rate was performed. Rectal temperature was measured every 15 min and used to adjust a water-circulating heating pad (e.g., TP-700, Stryker Corp) to keep the body temperature stable. The rabbit was placed on its side with the imaged eye facing upwards towards the exemplary mFLIO system. The head was secured in a three-degree-of-freedom mount for initial positioning. [00261] A pediatric wire speculum was used to open the imaged eye. Methylcellulose was applied to the cornea and a custom, plano, 44.00 diopter base curve, hard contact lens was placed on the cornea. The position was adjusted for pupil-centric scanning with the retina at the image plane of the exemplary mFLIO system and SLO. En- face images of the ovoid optic streak and peripapillary retina (6mm×6mm) were obtained with co-aligned SLO to select the field of view and track lateral movement of the eye. FIG. 6N shows a typical fluorescence lifetime (FLT) measurement. [00262] When doing the automatic unmixing by GMM, there were 3-lifetime distributions, corresponding to a high lifetime range (0.35-0.5 ns, cluster 602 in subpanel a), a low lifetime range (0.1-0.3 ns, cluster 604 in subpanel a), and background (cluster 606 in subpanel a). The RPE lifetime was in the range of 0.4-1 ns, and the photoreceptor lifetime was around 1.5-3 ns. It was impossible to discriminate the contributions of the photoreceptors and RPE due to uncertainty in excitation depth. The translational movement was also problematic. The in vivo experiments motivated the need for precise depth excitation using OCT guidance and real-time tracking to stabilize the ROI for fast and safe FLT calculation. [00263] The exemplary mFLIO system observed and tracked these phenomena in real-time within the living eye. This advancement holds substantial promise for early detection and continuous monitoring of functional changes in the visual cycle, providing a
Docket Number: 10046-590WO1 8300 YEH valuable tool for timely interventions and the development of personalized treatment strategies. Ultimately, such proactive measures can enhance the management and prognosis of patients grappling with AMD. [00264] Moreover, the exemplary mFLIO system can evaluate the efficacy of potential interventions that preserve photoreceptor function and prevent vision loss. The ability to assess these aspects in vivo facilitates a more comprehensive understanding of disease and accelerates the translation of research findings into practical, patient-centered solutions. [00265] FIG. 6O shows the detailed maps delineating light-dark response changes across various retinal locations, correlating these alterations with diverse cone-rod cell concentrations and retinal topography. The study used the rabbit retina as an anatomical proxy for the human retina. Rabbits are known to have a visual streak (VS), where the rod and cone photoreceptor, ganglion cell and amacrine cell density is highest, and which is located roughly 3 mm ventral to the optic nerve head (ONH) [1’, 2’]. The study sampled inside the visual streak (A, B, C) and outside the visual streak (D, E, F, G, H, I) and correlated the amplitude and decay constant of the light-dark cycles with the retinal anatomy. Each imaging session took 90 minutes to obtain a map of the amplitude variation of the light- dark cycle over approximately 10mm×10mm retina area centered on the visual streak. Using OCT-guided excitation depth offset; the study separated the light-dark cycle response of RPE and photoreceptors. The light-dark response had both an amplitude and a decay time constant (τ), and both measures may be biomarkers of photoreceptor-RPE physiology. [00266] Discussion [00267] Discussion #1. The change in fluorescence lifetime with light-dark cycles was most likely due to changes in the concentrations of all-trans-retinol (AT-ROL) and all-trans-retinal (AT-RAL). The temperature changes induced by the white light exposure and the 5 mW laser were very small and the fluorescence lifetime changes due to light-dark cycles decreased significantly as the biochemical processes wound down in the retina prep. This interpretation emphasizes the central role of biochemical dynamics in modulating fluorescence lifetime changes with light-dark cycles. [00268] FIG. 8 is a schematic diagram showing the pigment epithelium visual cycle for rod and cone photoreceptors and Müller cells showing the rod outer segment (ROS) 802, rod inner segment (RIS) 804, cone outer segment (COS), cone inner segment (CIS) 808, and retinal pigment epithelium (RPE) 810. After the absorption of photons by the photopigment in photoreceptors, the chromophore 11-cis-retinal (11-cis-RAL in FIG. 8)
Docket Number: 10046-590WO1 8300 YEH undergoes isomerization into all-trans-retinal (AT-RAL in FIG. 8). This all-trans-retinal must be transformed back into 11-cis-retinal in order to reattach to the opsin through the visual cycle [43] as depicted in FIG. 8. This regeneration process starts with the conversion of all- trans-retinal into all-trans-retinol (AT-ROL in FIG. 8) within the photoreceptor outer segments. Subsequently, all-trans-retinol is transported to retinal pigment epithelium (RPE) cells 810, where it is converted into all-trans-retinyl esters (AT-RE in FIG. 8) and stored in retinosomes until needed. These all-trans-retinyl esters are then transformed into 11-cis- retinol (11-cis-ROL in FIG. 8) and eventually into 11-cis-retinal (11-cis-RE in FIG. 8). The 11-cis-retinal is then transported to the photoreceptor outer segment to rebind with the opsin, thereby completing the visual cycle (FIG. 8, rod 801 and cone 803). the fluorescence lifetimes of all-trans-retinal (AT-RAL) and all-trans-retinol (AT-ROL) differ, with AT-RAL having a shorter lifetime (56 ps for RAL Schiff base in ethanol [44])) than AT-ROL (3 ns in ethanol [45], 3.4 ns [46]). Additionally, the emission spectrum of AT-ROL is slightly shifted towards the blue compared to AT-RAL [46,47]. [00269] FIG. 9B is a schematic diagram showing a representation of the pigment epithelium visual cycle for rod and cone photoreceptors. Cones, in addition to following the conventional visual cycle for chromophore regeneration, have an alternative pathway involving Müller cells [48] for photopigment regeneration as shown in FIG. 9B. In this alternative cycle specific to cones, all-trans-retinol (AT-ROL) is transported to Müller cells 902, where it undergoes conversion into 11-cis-retinol (11-cis-RE in FIG. 9B), after which it is carried back to the cones 903. A unique feature of cones 903, as opposed to rods 901, is their ability to directly transform 11-cis-retinol (11-cis-RE) into 11-cis-retina (11-cis- RE in FIG. 9B). This alternative mechanism stands out for its remarkable efficiency in producing the requisite 11-cis-retinal promptly, facilitating the binding with opsin to create cone photopigment [48]. [00270] The lifetime of a single fluorophore is not affected by its concentration. However, our system simultaneously detected the fluorescence signal from multiple fluorophores in photoreceptors and retinal pigment epithelium such as AT-ROL, AT-RAL, 11-cis-RAL, and 11-cis-ROL. The fluorophores are all excited by a 780 nm fs fiber laser, and the fluorescence emission is collected with a 500-720 nm emission. The lifetime detected in each pixel will be the mean lifetime of the mixture, therefore, the concentration of the fluorophores affects our lifetime measurement.
Docket Number: 10046-590WO1 8300 YEH [00271] Our findings demonstrate dynamic changes in photoreceptor lifetimes and retinal pigment epithelium lifetimes during the light-dark visual cycles. One possible explanation for the changes that we measured is as follows: [00272] When exposed to white light: The concentration of AT-ROL inside the photoreceptor increases, while AT-RAL and 11-cis-RAL decrease [49]. Consequently, the mean lifetime of photoreceptors increases. Simultaneously, a portion of the newly formed AT-ROL transfers back to the RPE, leading to an increase in retinal pigment epithelium mean lifetime. [00273] When left in the dark: AT-ROL continues to transfer from photoreceptors to the RPE, resulting in a decrease in photoreceptor mean lifetime. Additionally, the AT-ROL, which transferred back to the retinal pigment epithelium continued converted into AT-retinyl ester causing a corresponding decrease in retinal pigment epithelium mean lifetime. [00274] It is worth noting that the retinal pigment epithelium lifetime response differs between NZW rabbits and DB rabbits. In DB rabbits, the retinal pigment epithelium lifetime synchronizes with the light-dark visual cycles, while in NZW rabbits, the retinal pigment epithelium lifetime continues to increase throughout these cycles. [00275] Photoreceptors and the retinal pigment epithelium play critical roles in the visual cycle and vision. By monitoring the functional readout signature of photoreceptors and the retina, we can gain a better understanding of the mechanisms underlying age-related macular diseases. Early detection and monitoring of functional changes in the visual cycle can facilitate timely interventions and personalized treatment strategies, ultimately improving the management and prognosis of patients with AMD. Furthermore, these techniques hold promise for identifying novel therapeutic targets and evaluating the efficacy of potential interventions aimed at preserving photoreceptor function and preventing vision loss. [00276] The next challenge will be to measure the light-dark cycle in the living eye. We have not been able to measure the light-dark cycles in live rabbit eyes using 1P excitation and low NA optics, and 2P excitation in live eyes poses significant challenges. Motion is a significant challenge and the low NA optics of the live eye make it impossible to resolve the photoreceptors without adaptive optics. Motion artifacts can be minimized by decreasing acquisition time. The 2P excitation has improved depth resolution compared with 1P excitation but the eye must be stable to realize the improved depth resolution. [00277] In this disclosure, a FLIM that measures a change in the fluorescence lifetime of endogenous fluorophores in the photoreceptors and retinal pigment epithelium of
Docket Number: 10046-590WO1 8300 YEH freshly dissected rabbit retina in response to light-dark visual cycles is described. The biochemistry of the visual cycle is complex but changes in the distribution or binding of AT- ROL, 11-cis RAL and AT-RAL are likely fluorophore candidates. [00278] Discussion #2. Age-related macular degeneration (AMD) is an eye disease that affects the central part of the retina (macula) and leads to impairments in central vision and sharp vision. According to the report from the Centers for Disease Control and Prevention (CDC), AMD affects 0.8% of people in the US between 50 and 60, 1.5% between 60 and 70, 4.8% between 70 and 80, and 12% over 80 years old [9’]. Whereas early diagnosis of AMD relies on retina imaging, the current imaging methods (e.g., OCT) only provide structural information about the retina. The retina and its surrounding tissues are rich with endogenous fluorophores that absorb/emit light in the range of 250-700 nm. Many of these fluorophores are redox-active chromophores that regulate cell metabolism, which can be markers of many age-related retinal disorders [10’]. Although fundus autofluorescence microscopy (FAFM) is a technique that probes endogenous fluorophores in the retinal pigment epithelium for diagnosing various retinal diseases [11’] (FIG. 11) (e.g., degenerative [12’], dystrophic [13’], inflammatory [14’], neoplastic [14’] and toxic etiologies [15’]), FAFM experiences intensity- and spectrum-based biases [10’]. Dim endogenous fluorophores are difficult to image and interpret due to the spectral overlap among them. Unlike FAFM, fluorescence lifetime imaging ophthalmoscopy (FLIO) is not biased by fluorophore concentration or excitation power [16’], [17’]. Moreover, many endogenous fluorophores can be differentiated by their distinct lifetimes. The use of FLIO in retinal studies has shown reproducible lifetime image patterns around the posterior pole of the healthy retina [18’], [19’], making the lifetimes of retinal endogenous fluorophores a useful biomarker for early retinal disease diagnostics [20’], [21’]. [00279] Whereas more demonstrations of one-photon fluorescence lifetime imaging ophthalmoscopy (1P-FLIO) in retina diagnosis at different stages of AMD have been seen in recent years, 1P FLIO has not reached widespread use in clinics due to fundamental limitations. Limitations include (ii) one-photon excitation for most fluorophores found in the retina lies primarily in the ultraviolet range, and their fluorescence cannot be excited non- invasively through the pupil of the eye because of the ocular transmission window [22’-26’], (ii) 1P FLIO has low penetration depth with high background fluorescence due to excitation of fluorophores outside the focal plane, and (iii) the generation of fluorescence lifetime image is tedious and time-consuming, which can take hours [27’].
Docket Number: 10046-590WO1 8300 YEH [00280] As ocular tissues, such as the sclera, cornea, lens, and neurosensory retina, are transparent to infrared (IR) light, the advance of two-photon excitation techniques can overcome the limitation of ocular transmission windows [28’-33’] and the retina's fluorophores can be excited with red or near-infrared light [25’], [33’-35’]. The advantages of two-photon (2P) as compared to one-photon (1P) excitation imaging arise from two physical properties, namely, excitation of molecules at twice the wavelength and localization (optical confinement), which enables spatial reduction of the fluorescence volume [36’], [37’]. When using near-infrared radiation, the excitation radiation can be delivered deep into the tissue with decreased scattering and minimal absorption along the light path [38’]. Additionally, out-of-focus signals are minimized, resulting in low-noise images and avoiding excessive bleaching of the dyes outside the tight focus of the light beam, thus avoiding phototoxicity [39’] (FIG. 11, mFLIO). [00281] The characterization of autofluorescence lifetime changes of endogenous fluorophores in the retina, with a specific focus on retinal and retinol inside photoreceptors and retinal pigment epithelium (RPE) cells, is a crucial aspect of understanding retinal health [40’]. Retinal and retinol play a pivotal role in the visual cycle within photoreceptors and RPE cells [41’]. By examining the autofluorescence lifetimes of retinal and retinol, researchers can gain insights into the metabolic processes and integrity of these cells. Additionally, focusing on the RPE cells provides a comprehensive view of retinal health. Changes in autofluorescence lifetimes in these specific cell types can serve as early indicators of cellular stress, degeneration, or dysfunction, offering a valuable diagnostic tool for conditions like AMD and diabetic retinopathy [42’], [43’]. This targeted approach enhances the ability to monitor and understand the intricate dynamics of retinal components, contributing to advancements in both research and clinical applications for ocular health. [00282] Discussion #3. The 2P-FLIO module can provide further insight into the retina's physiology and pathology. Although intensity-based TPEF ophthalmoscopy in non-human primates has visualized several classes of retinal structures and probed both rod and cone function [28’], [30’], fluorescence lifetime imaging at the cellular scale has the potential to provide further insight into both physiology and pathology of the retina [42’-44’]. As a fluorescent molecule's intrinsic property, fluorescence lifetime measured by a FLIM system is not biased by excitation power [38’], [45’], or probe concentration [46’]. In addition, lifetime reading is not prone to photobleaching [47’] and can shed light on the microenvironment surrounding the fluorophore [45’], [48’], [49’]. This is important when imagining the outer retinal layers, which require passage of excitation and emission through
Docket Number: 10046-590WO1 8300 YEH several layers of vasculature and inner retina cells. In vivo, one-photon widefield fluorescence lifetime imaging ophthalmoscopy (FLIO) has been conducted in both rodent and human fundus using a scanning laser ophthalmoscope [18’], [50’]. Widefield FLIO shows potential as a quantitative measure of retinal health, enabling early detection of abnormalities or dysfunctions before structural changes become apparent [51’]. Alterations in retinal fluorescence lifetime are associated with various retinal diseases, including age-related macular degeneration [52’-54’], retinitis pigmentosa [55’], Stargardt disease [56’], and choroideremia [57’], among others [38’], [58’]. By measuring the fluorescence signals emitted by retinoids and their condensation products [59’] in response to light or chemical stimuli, the efficiency of the visual cycle can be assessed, and abnormalities or dysfunctions can be detected [28’], [30’]. Multiphoton imaging techniques can offer improved contrast and resolution, enabling high-throughput screening of retinal tissues at the single-cell level [32’], [33’], [43’]. [00283] The study developed a 2P-FLIO module to image the intrinsic fluorescence signatures of the retina at cellular resolution. The study demonstrated the application of two-photon FLIM to visualize retinal function during the visual cycle. Using a rabbit model, the study showed FLIM’s ability to provide insights into photoreceptor and RPE physiology during light/dark cycles. The dependence of fluorescence lifetime on light- dark cycles revealed the potential of FLIM to shed light on basic retinal function in health and disease (FIG. 8). [00284] Integration of 2P-FLIO and OCT allows the targeting of specific retinal layers for structural and functional imaging. The integration of FLIO and OCT modules in the exemplary system addresses a need for precise, stable, and consistent 2P retinal imaging. The synchronization of FLIO and OCT imaging allows precise and stable depth-of-focus co-registration. Three OCT A-scans can be recorded at different optical powers provided by the electronically tunable lens (ETL) at each retinal location. The three A-scans may be analyzed by the microcontroller for depths and amplitudes of the vitreous- RNFL and retinal pigment epithelium (RPE). Using the amplitude and depths of the vitreous- RNFL interface and RPE, the optimal power of the electronically tunable lens is determined to ensure the 2P beam is correctly focused. The active feedback autofocus mechanism employed in the SD-OCT module of the exemplary system utilizing the ETL, ensures optimal focus is maintained at each retinal location during imaging. This approach dynamically adjusts the focal plane based on each OCT A-scan. The PI and collaborators have previously employed similar approaches using OCT feedback for focus control in laser brain cancer
Docket Number: 10046-590WO1 8300 YEH surgery systems [60’-62’]. The exemplary method eliminates the need for mechanical movement, allowing the focal plane of both OCT and mFLIO beams to remain steady at a specific retinal layer, even in the presence of physiological variations. The continuous feedback loop can operate at rates up to hundreds of Hertz and enhances the efficiency and reliability of OCT systems, contributing to improved image quality and diagnostic accuracy. The integration of FLIO and OCT provides structural insights into retinal layers and functional information for ophthalmology applications and related medical fields. Comparative analyses of OCT and FLIO images allow for elucidation of the complementary advantages and limitations of each module, promoting a comprehensive assessment of light- dark cycles in the retina. [00285] Artificial Intelligence and Machine Learning [00286] The term “artificial intelligence” is defined herein to include any technique that enables one or more computing devices or comping systems (i.e., a machine) to mimic human intelligence. Artificial intelligence (AI) includes, but is not limited to, knowledge bases, machine learning, representation learning, and deep learning. The term “machine learning” is defined herein to be a subset of AI that enables a machine to acquire knowledge by extracting patterns from raw data. Machine learning techniques include, but are not limited to, logistic regression, support vector machines (SVMs), decision trees, Naïve Bayes classifiers, and artificial neural networks. The term “representation learning” is defined herein to be a subset of machine learning that enables a machine to automatically discover representations needed for feature detection, prediction, or classification from raw data. Representation learning techniques include, but are not limited to, autoencoders. The term “deep learning” is defined herein to be a subset of machine learning that enables a machine to automatically discover representations needed for feature detection, prediction, classification, etc. using layers of processing. Deep learning techniques include, but are not limited to, artificial neural network or multilayer perceptron (MLP). [00287] Machine learning models include supervised, semi-supervised, and unsupervised learning models. In a supervised learning model, the model learns a function that maps an input (also known as feature or features) to an output (also known as target or targets) during training with a labeled data set (or dataset). In an unsupervised learning model, the model learns patterns (e.g., structure, distribution, etc.) within an unlabeled data set. In a semi-supervised model, the model learns a function that maps an input (also known as feature or features) to an output (also known as target or target) during training with both labeled and unlabeled data.
Docket Number: 10046-590WO1 8300 YEH [00288] Artificial Neural Networks: An artificial neural network (ANN) is a computing system including a plurality of interconnected neurons (e.g., also referred to as “nodes”). This disclosure contemplates that the nodes can be implemented using a computing device (e.g., a processing unit and memory as described herein). The nodes can be arranged in a plurality of layers such as input layer, output layer, and optionally one or more hidden layers. An ANN having hidden layers can be referred to as deep neural network or multilayer perceptron (MLP). Each node is connected to one or more other nodes in the ANN. For example, each layer is made of a plurality of nodes, where each node is connected to all nodes in the previous layer. The nodes in a given layer are not interconnected with one another, i.e., the nodes in a given layer function independently of one another. As used herein, nodes in the input layer receive data from outside of the ANN, nodes in the hidden layer(s) modify the data between the input and output layers, and nodes in the output layer provide the results. Each node is configured to receive an input, implement an activation function (e.g., binary step, linear, sigmoid, tanH, or rectified linear unit (ReLU) function), and provide an output in accordance with the activation function. Additionally, each node is associated with a respective weight. ANNs are trained with a dataset to maximize or minimize an objective function. In some implementations, the objective function is a cost function, which is a measure of the ANN’s performance (e.g., error such as L1 or L2 loss) during training, and the training algorithm tunes the node weights and/or bias to minimize the cost function. This disclosure contemplates that any algorithm that finds the maximum or minimum of the objective function can be used for training the ANN. Training algorithms for ANNs include, but are not limited to, backpropagation. It should be understood that an artificial neural network is provided only as an example machine learning model. This disclosure contemplates that the machine learning model can be any supervised learning model, semi-supervised learning model, or unsupervised learning model. Optionally, the machine learning model is a deep learning model. Machine learning models are known in the art and are therefore not described in further detail herein. [00289] A convolutional neural network (CNN) is a type of deep neural network that has been applied, for example, to image analysis applications. Unlike a traditional neural networks, each layer in a CNN has a plurality of nodes arranged in three dimensions (width, height, depth). CNNs can include different types of layers, e.g., convolutional, pooling, and fully-connected (also referred to herein as “dense”) layers. A convolutional layer includes a set of filters and performs the bulk of the computations. A pooling layer is optionally inserted between convolutional layers to reduce the computational
Docket Number: 10046-590WO1 8300 YEH power and/or control overfitting (e.g., by downsampling). A fully-connected layer includes neurons, where each neuron is connected to all of the neurons in the previous layer. The layers are stacked similar to traditional neural networks. [00290] Logistic Regression: A logistic regression (LR) classifier is a supervised classification model that uses the logistic function to predict the probability of a target, which can be used for classification. LR classifiers are trained with a data set (also referred to herein as a “dataset”) to maximize or minimize an objective function, for example a measure of the LR classifier’s performance (e.g., error such as L1 or L2 loss), during training. This disclosure contemplates that any algorithm that finds the maximum or minimum of the objective function can be used. LR classifiers are known in the art and are therefore not described in further detail herein. [00291] Naïve Bayes: An Naïve Bayes’ (NB) classifier is a supervised classification model that is based on Bayes’ Theorem, which assumes independence among features (i.e., presence of one feature in a class is unrelated to presence of any other features). NB classifiers are trained with a data set by computing the conditional probability distribution of each feature given label and applying Bayes’ Theorem to compute conditional probability distribution of a label given an observation. NB classifiers are known in the art and are therefore not described in further detail herein. [00292] KNN: A k-NN classifier is a supervised classification model that classifies new data points based on similarity measures (e.g., distance functions). k-NN classifier is a non-parametric algorithm, i.e., it does not make strong assumptions about the function mapping input to output and therefore has flexibility to find a function that best fits the data. k-NN classifiers are trained with a data set (also referred to herein as a “dataset”) by learning associations between all samples and classification labels in the training dataset. k- NN classifiers are known in the art and are therefore not described in further detail herein. [00293] Watershed Algorithm: A watershed algorithm is an image processing technique for analyzing objects in image data (e.g., image segmentation) in order to identify particular features or objects. [00294] Gaussian Mixture Model (GMM): A GMM is a probabilistic model used in statistical modeling and machine learning that represents a mixture of multiple Gaussian (normal) distributions. Each component Gaussian distribution in the mixture model represents a cluster or subpopulation within the overall dataset. [00295] Clustering model: A clustering model is a type of machine learning model that is used for unsupervised learning tasks. Clustering is the process of grouping
Docket Number: 10046-590WO1 8300 YEH similar data points together based on certain features or characteristics, without any predefined labels. The goal is to identify inherent patterns or structures in the data. In a clustering model, the algorithm aims to partition a dataset into groups or clusters, where data points within the same cluster are more similar to each other than they are to points in other clusters. Clustering is often used for tasks such as segmentation and anomaly detection. Examples of clustering algorithms include the K-means algorithm, hierarchical clustering, Density-Based Spatial Clustering of Applications with Noise (DBSCAN), Mean Shift, Spectral Clustering, and Agglomerative Clustering. [00296] K-means algorithm: The K-means algorithm is an unsupervised machine learning model that is used for iterative clustering. The K-means algorithm minimizes the sum of squared distances between data points and their respective cluster centroids. It can be used to partition a dataset into k non-overlapping, distinct subsets or clusters, where each data point belongs to a cluster with the nearest mean or centroid. [00297] Computing Devices and Methods of Use [00298] It should be appreciated that the logical operations described herein with respect to the various figures may be implemented (1) as a sequence of computer- implemented acts or program modules (i.e., software) running on a computing device (e.g., the computing device described in FIG. 4), (2) as interconnected machine logic circuits or circuit modules (i.e., hardware) within the computing device and/or (3) a combination of software and hardware of the computing device. Thus, the logical operations discussed herein are not limited to any specific combination of hardware and software. The implementation is a matter of choice dependent on the performance and other requirements of the computing device. Accordingly, the logical operations described herein are referred to variously as operations, structural devices, acts, or modules. These operations, structural devices, acts and modules may be implemented in software, in firmware, in special-purpose digital logic, and any combination thereof. It should also be appreciated that more or fewer operations may be performed than shown in the figures and described herein. These operations may also be performed in a different order than those described herein. [00299] Referring to FIG. 10, an example computing device 1000 upon which embodiments of the present disclosure may be implemented is illustrated. It should be understood that the example computing device 1000 is only one example of a suitable computing environment upon which embodiments of the present disclosure may be implemented. Optionally, the computing device 1000 can be a well-known computing system including, but not limited to, personal computers, servers, handheld or laptop devices,
Docket Number: 10046-590WO1 8300 YEH multiprocessor systems, microprocessor-based systems, personal network computers (PCs), minicomputers, mainframe computers, embedded systems, and/or distributed computing environments including a plurality of any of the above systems or devices. Distributed computing environments enable remote computing devices, which are connected to a communication network or other data transmission medium, to perform various tasks. In the distributed computing environment, the program modules, applications, and other data may be stored on local and/or remote computer storage media. [00300] In its most basic configuration, the computing device 1000 typically includes at least one processing unit 1006 and system memory 1004. Depending on the exact configuration and type of computing device, system memory 1004 may be volatile (such as random-access memory (RAM)), non-volatile (such as read-only memory (ROM), flash memory, etc.), or some combination of the two. This most basic configuration is illustrated in FIG. 10 by the dashed line 1002. The processing unit 1006 may be a standard programmable processor that performs arithmetic and logic operations necessary for the operation of the computing device 1000. The computing device 1000 may also include a bus or other communication mechanism for communicating information among various components of the computing device 1000. [00301] Computing device 1000 may have additional features/functionality. For example, the computing device 1000 may include additional storage such as removable storage 1008 and non-removable storage 1010 including, but not limited to magnetic or optical disks or tapes. Computing device 1000 may also contain network connection(s) 1016 that allow the device to communicate with other devices. Computing device 1000 may also have input device(s) 1014 such as a keyboard, mouse, touch screen, etc. Output device(s) 1012, such as a display, speakers, printer, etc., may also be included. The additional devices may be connected to the bus in order to facilitate communication of data among the components of the computing device 1000. All these devices are well-known in the art and need not be discussed at length here. [00302] The processing unit 1006 may be configured to execute program code encoded in tangible, computer-readable media. Tangible, computer-readable media refers to any media that is capable of providing data that causes the computing device 1000 (i.e., a machine) to operate in a particular fashion. Various computer-readable media may be utilized to provide instructions to the processing unit 1006 for execution. Example of tangible, computer-readable media may include but is not limited to, volatile media, non-volatile media, removable media and non-removable media implemented in any method or
Docket Number: 10046-590WO1 8300 YEH technology for storage of information such as computer-readable instructions, data structures, program modules or other data. System memory 1004, removable storage 1008, and non- removable storage 1010 are all examples of tangible computer storage media. Examples of tangible, computer-readable recording media include but are not limited to, an integrated circuit (e.g., field-programmable gate array or application-specific IC), a hard disk, an optical disk, a magneto-optical disk, a floppy disk, a magnetic tape, a holographic storage medium, a solid-state device, RAM, ROM, electrically erasable program read-only memory (EEPROM), flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices. [00303] In an example implementation, the processing unit 1006 may execute program code stored in the system memory 1004. For example, the bus may carry data to the system memory 1004, from which the processing unit 1006 receives and executes instructions. The data received by the system memory 1004 may optionally be stored on the removable storage 1008 or the non-removable storage 1010 before or after execution by the processing unit 1006. [00304] It should be understood that the various techniques described herein may be implemented in connection with hardware or software or, where appropriate, with a combination thereof. Thus, the methods and apparatuses of the presently disclosed subject matter, or certain aspects or portions thereof, may take the form of program code (i.e., instructions) embodied in tangible media, such as floppy diskettes, CD-ROMs, hard drives, or any other machine-readable storage medium wherein, when the program code is loaded into and executed by a machine, such as a computing device, the machine becomes an apparatus for practicing the presently disclosed subject matter. In the case of program code execution on programmable computers, the computing device generally includes a processor, a storage medium readable by the processor (including volatile and non-volatile memory and/or storage elements), at least one input device, and at least one output device. One or more programs may implement or utilize the processes described in connection with the presently disclosed subject matter, for example, through the use of an application programming interface (API), reusable controls, or the like. Such programs may be implemented in a high- level procedural or object-oriented programming language to communicate with a computer system. However, the program(s) can be implemented in assembly or machine language if desired. In any case, the language may be a compiled or interpreted language, and it may be combined with hardware implementations.
Docket Number: 10046-590WO1 8300 YEH [00305] In one embodiment, disclosed herein is a non-transitory computer- readable storage medium comprising instructions that, when executed, cause at least one processor to perform the method of any preceding embodiments. [00306] Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims. [00307] The following patents, applications and publications as listed below and throughout this document are hereby incorporated by reference in their entirety herein. Reference list #1 [1] 1 Rando, R. R. The Biochemistry of the Visual Cycle. Chemical Reviews 101, 1881-1896 (2001). https://doi.org/10.1021/cr960141c [2] Strauss, O. The Retinal Pigment Epithelium in Visual Function. Physiological Reviews 85, 845-881 (2005). https://doi.org/10.1152/physrev.00021.2004 [3] Rattner, A., Smallwood, P. M. & Nathans, J. Identification and Characterization of All- trans-retinol Dehydrogenase from Photoreceptor Outer Segments, the Visual Cycle Enzyme That Reduces All-trans-retinal to All-trans-retinol. Journal of Biological Chemistry 275, 11034-11043 (2000). https://doi.org/10.1074/jbc.275.15.11034 [4] Rossi, E. A. et al. Imaging individual neurons in the retinal ganglion cell layer of the living eye. Proceedings of the National Academy of Sciences 114, 586-591 (2017). https://doi.org/10.1073/pnas.1613445114 [5] Sharma, R. et al. In Vivo Two-Photon Fluorescence Kinetics of Primate Rods and Cones. Investigative Opthalmology & Visual Science 57, 647 (2016). https://doi.org/10.1167/iovs.15- 17946 [6] Samimi, K. et al. In situ autofluorescence lifetime assay of a photoreceptor stimulus response in mouse retina and human retinal organoids. Biomedical Optics Express 13, 3476 (2022). https://doi.org/10.1364/BOE.455783 [7] Walters, S., Feeks, J. A., Huynh, K. T. & Hunter, J. J. Adaptive optics two-photon excited fluorescence lifetime imaging ophthalmoscopy of photoreceptors and retinal pigment epithelium in the living non-human primate eye. Biomedical Optics Express 13, 389 (2022). https://doi.org/10.1364/BOE.444550
Docket Number: 10046-590WO1 8300 YEH [8] Huynh, K. T., Walters, S., Foley, E. K. & Hunter, J. J. Separate lifetime signatures of macaque S cones, M/L cones, and rods observed with adaptive optics fluorescence lifetime ophthalmoscopy. Scientific Reports 13, 2456 (2023). https://doi.org/10.1038/s41598-023- 28877-6 [9] Becker, W. Fluorescence lifetime imaging – techniques and applications. Journal of Microscopy 247, 119-136 (2012). https://doi.org/https://doi.org/10.1111/j.1365- 2818.2012.03618.x [10] Sauer, L., Vitale, A. S., Modersitzki, N. K. & Bernstein, P. S. Fluorescence lifetime imaging ophthalmoscopy: autofluorescence imaging and beyond. Eye 35, 93-109 (2021). https://doi.org/10.1038/s41433-020-01287-y [11] Suhling, K. et al. Fluorescence lifetime imaging (FLIM): Basic concepts and some recent developments. Medical Photonics 27, 3-40 (2015). https://doi.org/https://doi.org/10.1016/j.medpho.2014.12.001 [12] Berezin, M. Y. & Achilefu, S. Fluorescence Lifetime Measurements and Biological Imaging. Chemical Reviews 110, 2641-2684 (2010). https://doi.org/10.1021/cr900343z [13] Wallrabe, H. & Periasamy, A. Imaging protein molecules using FRET and FLIM microscopy. Current Opinion in Biotechnology 16, 19-27 (2005). https://doi.org/https://doi.org/10.1016/j.copbio.2004.12.002 [14] Schrimpf, W. et al. Chemical diversity in a metal–organic framework revealed by fluorescence lifetime imaging. Nature Communications 9, 1647 (2018). https://doi.org/10.1038/s41467-018-04050-w [15] Schweitzer, D. et al. In vivo measurement of time-resolved autofluorescence at the human fundus. Journal of Biomedical Optics 9, 1214 (2004). https://doi.org/10.1117/1.1806833 [16] Dysli, C. et al. Quantitative Analysis of Fluorescence Lifetime Measurements of the Macula Using the Fluorescence Lifetime Imaging Ophthalmoscope in Healthy Subjects. [17] Skala, M. C. et al. In vivo multiphoton microscopy of NADH and FAD redox states, fluorescence lifetimes, and cellular morphology in precancerous epithelia. Proceedings of the National Academy of Sciences 104, 19494-19499 (2007). https://doi.org/10.1073/pnas.0708425104 [18] Dysli, C., Wolf, S. & Zinkernagel, M. S. Autofluorescence Lifetimes in Geographic Atrophy in Patients With Age-Related Macular Degeneration. Investigative Opthalmology & Visual Science 57, 2479 (2016). https://doi.org/10.1167/iovs.15-18381
Docket Number: 10046-590WO1 8300 YEH [19] Dysli, C., Fink, R., Wolf, S. & Zinkernagel, M. S. Fluorescence Lifetimes of Drusen in Age-Related Macular Degeneration. Investigative Opthalmology & Visual Science 58, 4856 (2017). https://doi.org/10.1167/iovs.17-22184 [20] Sauer, L., Komanski, C. B., Vitale, A. S., Hansen, E. D. & Bernstein, P. S. Fluorescence Lifetime Imaging Ophthalmoscopy (FLIO) in Eyes With Pigment Epithelial Detachments Due to Age-Related Macular Degeneration. Investigative Opthalmology & Visual Science 60, 3054 (2019). https://doi.org/10.1167/iovs.19-26835 [21] Andersen, K. M., Sauer, L., Gensure, R. H., Hammer, M. & Bernstein, P. S. Characterization of Retinitis Pigmentosa Using Fluorescence Lifetime Imaging Ophthalmoscopy (FLIO). Translational Vision Science & Technology 7, 20 (2018). https://doi.org/10.1167/tvst.7.3.20 [22] Dysli, C., Wolf, S., Hatz, K. & Zinkernagel, M. S. Fluorescence Lifetime Imaging in Stargardt Disease: Potential Marker for Disease Progression. Investigative Opthalmology & Visual Science 57, 832 (2016). https://doi.org/10.1167/iovs.15-18033 [23] Dysli, C., Wolf, S., Tran, H. V. & Zinkernagel, M. S. Autofluorescence Lifetimes in Patients With Choroideremia Identify Photoreceptors in Areas With Retinal Pigment Epithelium Atrophy. [24] Dysli, C., Berger, L., Wolf, S. & Zinkernagel, M. S. Fundus Autofluorescence Lifetimes and Central Serous Chorioretinopathy. Retina 37, 2151-2161 (2017). https://doi.org/10.1097/IAE.0000000000001452 [25] Sparrow, J. R. et al. The bisretinoids of retinal pigment epithelium. Progress in Retinal and Eye Research 31, 121-135 (2012). https://doi.org/10.1016/j.preteyeres.2011.12.001 [26] Boguslawski, J. et al. In vivo imaging of the human eye using a 2-photon-excited fluorescence scanning laser ophthalmoscope. The Journal of Clinical Investigation 132 (2022). https://doi.org/10.1172/JCI154218 [27] Palczewska, G. et al. Noninvasive two-photon optical biopsy of retinal fluorophores. Proceedings of the National Academy of Sciences 117, 22532-22543 (2020). https://doi.org/10.1073/pnas.2007527117 [28] Digman, M. A., Caiolfa, V. R., Zamai, M. & Gratton, E. The Phasor Approach to Fluorescence Lifetime Imaging Analysis. Biophysical Journal 94, L14-L16 (2008). https://doi.org/10.1529/biophysj.107.120154 [29] Ranjit, S., Malacrida, L., Jameson, D. M. & Gratton, E. Fit-free analysis of fluorescence lifetime imaging data using the phasor approach. Nature Protocols 13, 1979-2004 (2018). https://doi.org/10.1038/s41596-018-0026-5
Docket Number: 10046-590WO1 8300 YEH [30] Chen, Y.-I. et al. Generative adversarial network enables rapid and robust fluorescence lifetime image analysis in live cells. Communications Biology 5, 18-18 (2022). https://doi.org/10.1038/s42003-021-02938-w [31] Bishop, C. M. Pattern recognition and machine learning. Springer (2006). [32] Chiang, H. J. et al. HyU: Hybrid Unmixing for longitudinal in vivo imaging of low signal-to-noise fluorescence. Nature Methods 20, 248-258 (2023). https://doi.org/10.1038/s41592-022-01751-5 [33] Vallmitjana, A., Torrado, B. & Gratton, E. Phasor-based image segmentation: machine learning clustering techniques. Biomedical Optics Express 12, 3410-3422 (2021). https://doi.org/10.1364/BOE.422766 [34] Huang, T., Peng, H. & Zhang, K. Model Selection for Gaussian Mixture Models. Statistica Sinica (2017). https://doi.org/10.5705/ss.2014.105 [35] Bianchetti, G., Spirito, M. D. & Maulucci, G. Unsupervised clustering of multiparametric fluorescent images extends the spectrum of detectable cell membrane phases with sub-micrometric resolution. Biomedical Optics Express 11, 5728-5744 (2020). https://doi.org/10.1364/BOE.399655 [36] Gideon, S. Estimating the Dimension of a Model. The Annals of Statistics 6, 461-464 (1978). https://doi.org/10.1214/aos/1176344136 [37] Dempster, A. P., Laird, N. M. & Rubin, D. B. Maximum Likelihood from Incomplete Data via the EM Algorithm. Journal of the Royal Statistical Society. Series B (Methodological) 39, 1-38 (1977). [38] Ng, S. K., Krishnan, T. & McLachlan, G. J. Handbook of Computational Statistics: Concepts and Methods, Springer (2012). [39] Levine, M. W. An analysis of the cross-correlation between ganglion cells in the retina of goldfish. Visual Neuroscience 14, 731-739 (1997). https://doi.org/10.1017/S0952523800012682 [40] DeVries, S. H. Correlated Firing in Rabbit Retinal Ganglion Cells. Journal of Neurophysiology 81, 908-920 (1999). https://doi.org/10.1152/jn.1999.81.2.908 [41] Myers, L. J., Erim, Z. & Lowery, M. M. Time and frequency domain methods for quantifying common modulation of motor unit firing patterns. Journal of NeuroEngineering and Rehabilitation 1, 2 (2004). https://doi.org/10.1186/1743-0003-1-2 [42] Litvack, D. A., Oberlander, T. F., Carney, L. H. & Saul, J. P. Time and frequency domain methods for heart rate variability analysis: A methodological comparison. Psychophysiology 32, 492-504 (1995). https://doi.org/10.1111/j.1469-8986.1995.tb02101.x
Docket Number: 10046-590WO1 8300 YEH [43] Kiser, P. D., Golczak, M., Maeda, A. & Palczewski, K. Key enzymes of the retinoid (visual) cycle in vertebrate retina. Biochimica et Biophysica Acta (BBA) - Molecular and Cell Biology of Lipids 1821, 137-151 (2012). https://doi.org/10.1016/j.bbalip.2011.03.005 [44] Bachilo, S. M. & Gillbro, T. Fluorescence of Retinal Schiff Base in Alcohols. The Journal of Physical Chemistry A 103, 2481-2488 (1999). https://doi.org/10.1021/jp983646a [45] Chihara, K. et al. Spectroscopy and Photophysical Dynamics of Retinol and Retinyl Ether. Photochemistry and Photobiology 29, 1001-1008 (1979). https://doi.org/10.1111/j.1751-1097.1979.tb07804.x [46] Chmykh, Y. G. & Nadeau, J. L. Characterization of Retinol Stabilized in Phosphatidylcholine Vesicles with and without Antioxidants. ACS Omega 5, 18367-18375 (2020). https://doi.org/10.1021/acsomega.0c02102 [47] Georghiou, S. & Churchich, J. E. Nanosecond spectroscopy of retinol. Journal of Biological Chemistry 250, 1149-1151 (1975). https://doi.org/10.1016/S0021-9258(19)41902- 9 [48] Wang, J.-S. & Kefalov, V. J. The Cone-specific visual cycle. Progress in Retinal and Eye Research 30, 115-128 (2011). https://doi.org/10.1016/j.preteyeres.2010.11.001 [49] Saari, J. C. Vitamin A Metabolism in Rod and Cone Visual Cycles. Annual Review of Nutrition 32, 125-145 (2012). https://doi.org/10.1146/annurev-nutr-071811-150748 Reference list #2 [1’] Lavaud A, Soukup P, Martin L, Hartnack S, Pot S. Spectral Domain Optical Coherence Tomography in Awake Rabbits Allows Identification of the Visual Streak, a Comparison with Histology. Transl Vis Sci Technol. 2020;9(5):13. [2’] Sonntag SR, Seifert E, Hamann M, Lewke B, Theisen-Kunde D, Grisanti S, Brinkmann R, Miura Y. Fluorescence Lifetime Changes Induced by Laser Irradiation: A Preclinical Study towards the Evaluation of Retinal Metabolic States. Life (Basel). 2021;11(6). [3’] Nguyen TD, Chen Y-I, Nguyen A-T, Chen LH, Yonas S, Litvinov M, He Y, Kuo Y-A, Hong S, Rylander HG, Yeh H-C. Multiplexed imaging in live cells using pulsed interleaved excitation spectral FLIM. Optics Express. 2024;32(3):3290-307. [4’] Nguyen TD, Chen YI, Nguyen AT, Yonas S, Sripati MP, Kuo YA, Hong S, Litvinov M, He Y, Yeh HC, Grady Rylander H, 3rd. Two-photon autofluorescence lifetime assay of rabbit photoreceptors and retinal pigment epithelium during light-dark visual cycles in rabbit retina. Biomed Opt Express. 2024;15(5):3094-111.
Docket Number: 10046-590WO1 8300 YEH [5’] Rein DB, Wittenborn JS, Burke-Conte Z, Gulia R, Robalik T, Ehrlich JR, Lundeen EA, Flaxman AD. Prevalence of Age-Related Macular Degeneration in the US in 2019. JAMA Ophthalmology. 2022;140(12):1202-8. [6’] Fleckenstein M, Schmitz-Valckenberg S, Chakravarthy U. Age-Related Macular Degeneration: A Review. JAMA. 2024;331(2):147-57. [7’] Fleckenstein M, Keenan TDL, Guymer RH, Chakravarthy U, Schmitz-Valckenberg S, Klaver CC, Wong WT, Chew EY. Age-related macular degeneration. Nature Reviews Disease Primers. 2021;7(1):31. [8’] Fritsche LG, Fariss RN, Stambolian D, Abecasis GR, Curcio CA, Swaroop A. Age- Related Macular Degeneration: Genetics and Biology Coming Together. Annual Review of Genomics and Human Genetics. 2014;15(1):151-71. [9’] NEI N. Facts About Age-Related Macular Degeneration. Obtido em. 2015;10. [10’] Dysli C, Wolf S, Berezin MY, Sauer L, Hammer M, Zinkernagel MS. Fluorescence lifetime imaging ophthalmoscopy. Progress in Retinal and Eye Research. 2017;60:120-43. [11’] Yung M, Klufas MA, Sarraf D. Clinical applications of fundus autofluorescence in retinal disease. International Journal of Retina and Vitreous. 2016;2(1):12. [12’] Wing GL, Blanchard GC, Weiter JJ. The topography and age relationship of lipofuscin concentration in the retinal pigment epithelium. Investigative Ophthalmology & Visual Science. 1978;17(7):601-7. [13’] Wallrabe H, Svindrych Z, Alam SR, Siller KH, Wang T, Kashatus D, Hu S, Periasamy A. Segmented cell analyses to measure redox states of autofluorescent NAD (P) H, FAD & Trp in cancer cells by FLIM. Scientific Reports. 2018;8(1):1-11. [14’] Vianna RN, Muralha A, Muralha L. Indocyanine-green angiography in acute idiopathic exudative polymorphous vitelliform maculopathy. Retina. 2003;23(4):538-41. [15’] Yonekawa Y, Miller JW, Kim IK. Age-related macular degeneration: advances in management and diagnosis. Journal of Clinical Medicine. 2015;4(2):343-59. [16’] Kantelhardt SR, Kalasauskas D, König K, Kim E, Weinigel M, Uchugonova A, Giese A. In vivo multiphoton tomography and fluorescence lifetime imaging of human brain tumor tissue. Journal of neuro-oncology. 2016;127(3):473-82. [17’] Alfonso‐Garcia A, Bec J, Sridharan Weaver S, Hartl B, Unger J, Bobinski M, Lechpammer M, Girgis F, Boggan J, Marcu L. Real‐time augmented reality for delineation of surgical margins during neurosurgery using autofluorescence lifetime contrast. Journal of Biophotonics. 2020;13(1):e201900108.
Docket Number: 10046-590WO1 8300 YEH [18’] Dysli C, Quellec G, Abegg M, Menke MN, Wolf-Schnurrbusch U, Kowal J, Blatz J, La Schiazza O, Leichtle AB, Wolf S. Quantitative analysis of fluorescence lifetime measurements of the macula using the fluorescence lifetime imaging ophthalmoscope in healthy subjects. Investigative Ophthalmology & Visual Science. 2014;55(4):2106-13. [19’] Klemm M, Dietzel A, Haueisen J, Nagel E, Hammer M, Schweitzer D. Repeatability of autofluorescence lifetime imaging at the human fundus in healthy volunteers. Current Eye Research. 2013;38(7):793-801. [20’] Sauer L, Andersen KM, Li B, Gensure RH, Hammer M, Bernstein PS. Fluorescence lifetime imaging ophthalmoscopy (FLIO) of macular pigment. Investigative Ophthalmology & Visual Science. 2018;59(7):3094-103. [21’] Sauer L, Andersen KM, Dysli C, Zinkernagel MS, Bernstein PS, Hammer M. Review of clinical approaches in fluorescence lifetime imaging ophthalmoscopy. Journal of Biomedical Optics. 2018;23(9):091415. [22’] Croce AC, Bottiroli G. Autofluorescence spectroscopy and imaging: a tool for biomedical research and diagnosis. European journal of histochemistry : EJH. 2014;58(4):2461-. [23’] Garwin GG, Saari JC. High-performance liquid chromatography analysis of visual cycle retinoids. Methods in enzymology. 2000;316:313-24. [24’] Schweitzer D, Schenke S, Hammer M, Schweitzer F, Jentsch S, Birckner E, Becker W, Bergmann A. Towards metabolic mapping of the human retina. Microscopy research and technique. 2007;70(5):410-9. [25’] Zipfel WR, Williams RM, Christie RH, Nikitin AY, Hyman BT, Webb WW. Live tissue intrinsic emission microscopy using multiphoton-excited native fluorescence and second harmonic generation. Proceedings of the National Academy of Sciences of the United States of America. 2003;100(12):7075-80. [26’] Zipfel WR, Williams RM, Webb WW. Nonlinear magic: multiphoton microscopy in the biosciences. Nature biotechnology. 2003;21(11):1369-77. [27’] Datta R, Heaster TM, Sharick JT, Gillette AA, Skala MC. Fluorescence lifetime imaging microscopy: fundamentals and advances in instrumentation, analysis, and applications. Journal of Biomedical Optics. 2020;25(7):071203. [28’] Rossi EA, Granger CE, Sharma R, Yang Q, Saito K, Schwarz C, Walters S, Nozato K, Zhang J, Kawakami T, Fischer W, Latchney LR, Hunter JJ, Chung MM, Williams DR. Imaging individual neurons in the retinal ganglion cell layer of the living eye. Proceedings of the National Academy of Sciences. 2017;114(3):586-91.
Docket Number: 10046-590WO1 8300 YEH [29’] Sharma R, Williams DR, Palczewska G, Palczewski K, Hunter JJ. Two-Photon Autofluorescence Imaging Reveals Cellular Structures Throughout the Retina of the Living Primate Eye. Investigative Ophthalmology & Visual Science. 2016;57(2):632-46. [30’] Sharma R, Schwarz C, Williams DR, Palczewska G, Palczewski K, Hunter JJ. In Vivo Two-Photon Fluorescence Kinetics of Primate Rods and Cones. Investigative Opthalmology & Visual Science. 2016;57(2):647. [31’] Sharma R, Yin L, Geng Y, Merigan WH, Palczewska G, Palczewski K, Williams DR, Hunter JJ. In vivo two-photon imaging of the mouse retina. Biomedical optics express. 2013;4(8):1285-93. [32’] Boguslawski J, Palczewska G, Tomczewski S, Milkiewicz J, Kasprzycki P, Stachowiak D, Komar K, Marzejon MJ, Sikorski BL, Hudzikowski A, Głuszek A, Łaszczych Z, Karnowski K, Soboń G, Palczewski K, Wojtkowski M. In vivo imaging of the human eye using a 2-photon-excited fluorescence scanning laser ophthalmoscope. The Journal of Clinical Investigation. 2022;132(2). [33’] Palczewska G, Boguslawski J, Stremplewski P, Kornaszewski L, Zhang J, Dong Z, Liang X-X, Gratton E, Vogel A, Wojtkowski M, Palczewski K. Noninvasive two-photon optical biopsy of retinal fluorophores. Proceedings of the National Academy of Sciences. 2020;117(36):22532-43. [34’] Bachilo SM, Gillbro T. Fluorescence of retinal Schiff base in alcohols. The Journal of Physical Chemistry A. 1999;103(15):2481-8. [35’] Lakowicz JR, Szmacinski H, Nowaczyk K, Johnson ML. Fluorescence lifetime imaging of free and protein-bound NADH. Proceedings of the National Academy of Sciences. 1992;89(4):1271-5. [36’] Lakowicz JR. Principles of Fluorescence Spectroscopy: Springer; 2007. [37’] Wang B-G, König K, Halbhuber KJ. Two-photon microscopy of deep intravital tissues and its merits in clinical research. Journal of microscopy. 2010;238(1):1-20. [38’] Sauer L, Vitale AS, Modersitzki NK, Bernstein PS. Fluorescence lifetime imaging ophthalmoscopy: autofluorescence imaging and beyond. Eye. 2021;35(1):93-109. [39’] Palczewska G, Wojtkowski M, Palczewski K. From mouse to human: Accessing the biochemistry of vision in vivo by two-photon excitation. Progress in Retinal and Eye Research. 2023;93:101170. [40’] Kiser PD, Golczak M, Maeda A, Palczewski K. Key enzymes of the retinoid (visual) cycle in vertebrate retina. Biochimica et Biophysica Acta (BBA) - Molecular and Cell Biology of Lipids. 2012;1821(1):137-51.
Docket Number: 10046-590WO1 8300 YEH [41’] Wang J-S, Kefalov VJ. The Cone-specific visual cycle. Progress in Retinal and Eye Research. 2011;30(2):115-28. [42’] Huynh KT, Walters S, Foley EK, Hunter JJ. Separate lifetime signatures of macaque S cones, M/L cones, and rods observed with adaptive optics fluorescence lifetime ophthalmoscopy. Scientific Reports. 2023;13(1):2456. [43’] Walters S, Feeks JA, Huynh KT, Hunter JJ. Adaptive optics two-photon excited fluorescence lifetime imaging ophthalmoscopy of photoreceptors and retinal pigment epithelium in the living non-human primate eye. Biomedical optics express. 2022;13(1):389. [44’] Samimi K, Pattnaik BR, Capowski EE, Saha K, Gamm DM, Skala MC. In situ autofluorescence lifetime assay of a photoreceptor stimulus response in mouse retina and human retinal organoids. Biomedical optics express. 2022;13(6):3476. [45’] Becker W. Fluorescence lifetime imaging – techniques and applications. Journal of microscopy. 2012;247(2):119-36. [46’] Suhling K, Hirvonen LM, Levitt JA, Chung P-H, Tregidgo C, Le Marois A, Rusakov DA, Zheng K, Ameer-Beg S, Poland S, Coelho S, Henderson R, Krstajic N. Fluorescence lifetime imaging (FLIM): Basic concepts and some recent developments. Medical Photonics. 2015;27:3-40. [47’] Berezin MY, Achilefu S. Fluorescence Lifetime Measurements and Biological Imaging. Chemical Reviews. 2010;110(5):2641-84. [48’] Wallrabe H, Periasamy A. Imaging protein molecules using FRET and FLIM microscopy. Current Opinion in Biotechnology. 2005;16(1):19-27. [49’] Schrimpf W, Jiang J, Ji Z, Hirschle P, Lamb DC, Yaghi OM, Wuttke S. Chemical diversity in a metal–organic framework revealed by fluorescence lifetime imaging. Nature Communications. 2018;9(1):1647. [50’] Schweitzer D, Hammer M, Schweitzer F, Anders R, Doebbecke T, Schenke S, Gaillard ER. In vivo measurement of time-resolved autofluorescence at the human fundus. Journal of Biomedical Optics. 2004;9(6):1214. [51’] Skala MC, Riching KM, Gendron-Fitzpatrick A, Eickhoff J, Eliceiri KW, White JG, Ramanujam N. In vivo multiphoton microscopy of NADH and FAD redox states, fluorescence lifetimes, and cellular morphology in precancerous epithelia. Proceedings of the National Academy of Sciences. 2007;104(49):19494-9. [52’] Dysli C, Wolf S, Zinkernagel MS. Autofluorescence Lifetimes in Geographic Atrophy in Patients With Age-Related Macular Degeneration. Investigative Opthalmology & Visual Science. 2016;57(6):2479.
Docket Number: 10046-590WO1 8300 YEH [53’] Dysli C, Fink R, Wolf S, Zinkernagel MS. Fluorescence Lifetimes of Drusen in Age- Related Macular Degeneration. Investigative Opthalmology & Visual Science. 2017;58(11):4856. [54’] Sauer L, Komanski CB, Vitale AS, Hansen ED, Bernstein PS. Fluorescence Lifetime Imaging Ophthalmoscopy (FLIO) in Eyes With Pigment Epithelial Detachments Due to Age- Related Macular Degeneration. Investigative Opthalmology & Visual Science. 2019;60(8):3054. [55’] Andersen KM, Sauer L, Gensure RH, Hammer M, Bernstein PS. Characterization of Retinitis Pigmentosa Using Fluorescence Lifetime Imaging Ophthalmoscopy (FLIO). Translational Vision Science & Technology. 2018;7(3):20. [56’] Dysli C, Wolf S, Hatz K, Zinkernagel MS. Fluorescence Lifetime Imaging in Stargardt Disease: Potential Marker for Disease Progression. Investigative Opthalmology & Visual Science. 2016;57(3):832. [57’] Dysli C, Wolf S, Tran HV, Zinkernagel MS. Autofluorescence Lifetimes in Patients With Choroideremia Identify Photoreceptors in Areas With Retinal Pigment Epithelium Atrophy. Investigative Ophthalmology & Visual Science. 2016;57(15):6714-21. [58’] Dysli C, Berger L, Wolf S, Zinkernagel MS. Fundus Autofluorescence Lifetimes and Central Serous Chorioretinopathy. Retina. 2017;37(11):2151-61. [59’] Sparrow JR, Gregory-Roberts E, Yamamoto K, Blonska A, Ghosh SK, Ueda K, Zhou J. The bisretinoids of retinal pigment epithelium. Progress in Retinal and Eye Research. 2012;31(2):121-35. [60’] Katta N, McElroy AB, Estrada AD, Milner TE. Optical coherence tomography image- guided smart laser knife for surgery. Lasers in Surgery and Medicine. 2018;50(3):202-12. [61’] Katta N, Estrada AD, McElroy AB, Gruslova A, Oglesby M, Cabe AG, Feldman MD, Fleming RYD, Brenner AJ, Milner TE. Laser brain cancer surgery in a xenograft model guided by optical coherence tomography. Theranostics. 2019;9(12):3555-64. [62’] Katta N, Estrada AD, McElroy AB, Milner TE. Er:YAG laser brain surgery with vascular specific coagulation. Lasers in Surgery and Medicine. 2022;54(8):1107-15. [63’] Wang H, Vilela M, Winkler A, Tarnawski M, Schlichting I, Yumerefendi H, Kuhlman B, Liu RH, Danuser G, Hahn KM. LOVTRAP: an optogenetic system for photoinduced protein dissociation. Nature Methods. 2016;13(9):755-8. [64’] Sun Y, Nguyen TD, Chen Y-I, Coskun UC, Liao S-C, Yeh H-C, editors. Characterization of a time-resolved spectral detector for spectral fluorescence lifetime
Docket Number: 10046-590WO1 8300 YEH imaging with the parallel 16-channel FastFLIM and the phasor analysis. Proc SPIE - Multiphoton Microscopy in the Biomedical Sciences XXIII; 2023. [65’] Bishop CM. Pattern recognition and machine learning: Springer; 2006. [66’] Digman MA, Caiolfa VR, Zamai M, Gratton E. The Phasor Approach to Fluorescence Lifetime Imaging Analysis. Biophysical Journal. 2008;94(2):L14-L6. [67’] Chiang HJ, Koo DES, Kitano M, Burkitt S, Unruh JR, Zavaleta C, Trinh LA, Fraser SE, Cutrale F. HyU: Hybrid Unmixing for longitudinal in vivo imaging of low signal-to-noise fluorescence. Nature Methods. 2023;20(2):248-58. [68’] Saari JC. Vitamin A Metabolism in Rod and Cone Visual Cycles. Annual Review of Nutrition. 2012;32(1):125-45. [69’] Levine MW. An analysis of the cross-correlation between ganglion cells in the retina of goldfish. Visual Neuroscience. 1997;14(4):731-9. [70’] DeVries SH. Correlated Firing in Rabbit Retinal Ganglion Cells. Journal of Neurophysiology. 1999;81(2):908-20. [71’] Martinez-Conde S, Macknik SL, Hubel DH. The role of fixational eye movements in visual perception. Nature Reviews Neuroscience. 2004;5(3):229-40. [72’] Martinez-Conde S, Otero-Millan J, Macknik SL. The impact of microsaccades on vision: towards a unified theory of saccadic function. Nature Reviews Neuroscience. 2013;14(2):83-96. [73’] Bartuzel MM, Wróbel K, Tamborski S, Meina M, Nowakowski M, Dalasiński K, Szkulmowska A, Szkulmowski M. High-resolution, ultrafast, wide-field retinal eye-tracking for enhanced quantification of fixational and saccadic motion. Biomedical optics express. 2020;11(6):3164-80. [74’] Yarbus AL. Eye movements and vision: Springer; 2013. [75’] Robinson DA. Movement using a scleral search in a magnetic field. IEEE Transactions on Bio-Medical Electronics. 1963;10(4):137-45. [76’] Cornsweet TN, Crane HD. Accurate two-dimensional eye tracker using first and fourth Purkinje images. Journal of the Optical Society of America. 1973;63(8):921-8. [77’] Salmon AE, Cooper RF, Langlo CS, Baghaie A, Dubra A, Carroll J. An automated reference frame selection (ARFS) algorithm for cone imaging with adaptive optics scanning light ophthalmoscopy. Translational Vision Science & Technology. 2017;6(2):9-. [78’] Azimipour M, Zawadzki RJ, Gorczynska I, Migacz J, Werner JS, Jonnal RS. Intraframe motion correction for raster-scanned adaptive optics images using strip-based cross- correlation lag biases. PLoS One. 2018;13(10):e0206052.
Docket Number: 10046-590WO1 8300 YEH [79’] Sheehy CK, Yang Q, Arathorn DW, Tiruveedhula P, de Boer JF, Roorda A. High- speed, image-based eye tracking with a scanning laser ophthalmoscope. Biomedical optics express. 2012;3(10):2611-22. [80’] Yang Q, Zhang J, Nozato K, Saito K, Williams DR, Roorda A, Rossi EA. Closed-loop optical stabilization and digital image registration in adaptive optics scanning light ophthalmoscopy. Biomedical optics express. 2014;5(9):3174-91. [81’] Sheehy CK, Bensinger ES, Romeo A, Rani L, Stepien-Bernabe N, Shi B, Helft Z, Putnam N, Cordano C, Gelfand JM. Fixational microsaccades: A quantitative and objective measure of disability in multiple sclerosis. Multiple Sclerosis Journal. 2020;26(3):343-53. [82’] Denk W, Strickler JH, Webb WW. Two-Photon Laser Scanning Fluorescence Microscopy. Science (New York, NY). 1990;248(4951):73-6. [83’] Palczewska G, Stremplewski P, Suh S, Alexander NS, Salom D, Dong Z, Ruminski D, Choi EH, Sears AE, Kern TS, Wojtkowski M, Palczewski K. Two-photon imaging of the mammalian retina with ultrafast pulsing laser. JCI insight. 2018;3(17). [84’] Helmchen F, Denk W. Deep tissue two-photon microscopy. Nature Methods. 2005;2(12):932-40. [85’] Hunter JJ, Merigan WH, Schallek JB. Imaging Retinal Activity in the Living Eye. Annual Review of Vision Science. 2019;5(1):15-45. [86’] Martinez-Conde S, Macknik SL, Hubel DH. The role of fixational eye movements in visual perception. Nature reviews Neuroscience. 2004;5(3):229-40. [87’] Rolfs M. Microsaccades: Small steps on a long way. Vision research. 2009;49(20):2415-41. [88’] Rossi EA, Achtman RL, Guidon A, Williams DR, Roorda A, Bavelier D, Carroll J. Visual Function and Cortical Organization in Carriers of Blue Cone Monochromacy. PLoS One. 2013;8(2):e57956. [89’] Hwang K, Seo Y-H, Ahn J, Kim P, Jeong K-H. Frequency selection rule for high definition and high frame rate Lissajous scanning. Scientific Reports. 2017;7(1):14075. [90’] Dubra A, Harvey Z, editors. Registration of 2D Images from Fast Scanning Ophthalmic Instruments. Biomedical Image Registration; 2010: Springer. [91’] Kraus MF, Potsaid B, Mayer MA, Bock R, Baumann B, Liu JJ, Hornegger J, Fujimoto JG. Motion correction in optical coherence tomography volumes on a per A-scan basis using orthogonal scan patterns. Biomedical optics express. 2012;3(6):1182-99. [92’] Donnelly WJ, 3rd, Roorda A. Optimal pupil size in the human eye for axial resolution. J Opt Soc Am A Opt Image Sci Vis. 2003;20(11):2010-5.
Docket Number: 10046-590WO1 8300 YEH [93’] Delori FC, Webb RH, Sliney DH. Maximum permissible exposures for ocular safety (ANSI 2000), with emphasis on ophthalmic devices. Journal of the Optical Society of America A. 2007;24(5):1250-65.