View system and method for diagnosing and detecting glaucoma by performing adaptive map examination through head mounted display

Abstract

A system may include a head-mounted device equipped with a display screen and also incorporating an adaptive map view check algorithm configured to access or implement a standard database or model. The system can implement a VF test, the test is based on a display area of a value adaptive display screen in a standard database or model and presents the same or similar visual effect as one or more different head-mounted display devices, and the one or more different head-mounted display devices have different display screens. Different display screens have different respective shapes, formats, sizes and / or resolutions. The system may receive visual test data indicative of a user's field of view and detect one or more initial test locations. The one or more initial test locations may define one or more healthy clusters that indicate no scatoma, and one or more impaired clusters that indicate the presence of dark spots. The system may also generate spatial mappings identifying locations of one or more impaired clusters.

Description

[0001] Cross-reference to related applications This application claims the benefit of U.S. Provisional Patent Application No. 63 / 615,945, filed December 29, 2023. The entire contents of each of the preceding prior applications listed above are incorporated herein by reference. This application claims priority to U.S. Provisional Patent Application No. 63 / 673,383, filed July 19, 2024, which is incorporated herein by reference. Technical Field

[0002] This disclosure generally relates to systems and methods for glaucoma diagnosis and monitoring, and more particularly to VR systems and methods for diagnosing and monitoring glaucoma through adaptive visual field (VF) analysis on a head-mounted display, such as a virtual reality (VR) headset or other VR device. Background Technology

[0003] Glaucoma is a chronic, progressive eye disease caused by damage to the optic nerve, potentially leading to visual field (VF) loss. One of the main risk factors is intraocular pressure. Abnormalities in the eye's drainage system can cause aqueous humor to accumulate, creating excessive pressure that damages the optic nerve. Glaucoma is a leading cause of irreversible blindness and is difficult to diagnose, often resulting in late detection and irreversible vision loss. For example, 3 million Americans have glaucoma. With an aging population, this prevalence is projected to increase to 6.3 million within the next 30 years. As a disease that is asymptomatic in its early to mid-stages, glaucoma is difficult to diagnose. Early detection of visual changes is crucial for preventing vision loss; however, the subjectivity of functional testing and the limitations of current diagnostic methods complicate diagnosis and disease monitoring.

[0004] The primary purpose of visual field testing is to measure the overall performance of the visual field (VF). Before the introduction of computers that allow for the automated presentation of visual stimuli, this goal was achieved by manually moving small stimuli to find transitional positions known as isolines between the visible and invisible areas of a particular stimulus. These isolines generate a visual hill topography of the patient. This test needs to be performed by a qualified clinician. Visual field testing can identify topographical abnormalities caused by tumors compressing the optic nerve pathway, vascular injury of the central nervous system, glaucoma, etc. Manual visual field testing has several limitations. It requires highly trained testers to perform the test accurately, is costly, and limits scalability. The test is time-consuming for both testers and patients, and there is often significant variability between testers, especially over time, reducing the reliability and consistency of the data.

[0005] These drawbacks led to the replacement of manual dynamic perimetry with static automated perimetry (SAP) in glaucoma diagnosis, an examination originally intended as a supplement to dynamic perimetry. SAP offered a more uniform approach, eliminating reliance on the visual field examiner. This uniformity of testing facilitated standardized data analysis and interpretation, crucial for SAP's success in replacing dynamic perimetry. However, due to insufficient spatial sampling and high test-retest variability in estimating contrast sensitivity thresholds of damaged areas from a limited number of stimulus presentations, SAP was less reliable than dynamic perimetry in identifying spatial patterns of damage and in comparing with retinal examination results. The transition to SAP was only possible in the 1970s and 80s, when advancements in software and hardware allowed for the development of algorithmic and computationally intensive analytical tools; its limitations are a product of that era. Since the 1980s, despite significant progress in understanding the structural and anatomical changes occurring in glaucoma, visual field testing methods have not changed considerably. For example, the use of overly large grid spacing at test locations, inconsistent with results from structural testing and other imaging methods, reduces our clinical confidence in damage detection and disease progression assessment. For patients, this delay means irreversible vision loss. Furthermore, healthcare providers face challenges due to unreliable results, often requiring repeat VF tests at subsequent visits to make treatment decisions. It has been reported that 17% to 48% of VF test results are unreliable due to high rates of fixation loss, false positives, and false negatives.

[0006] Therefore, the limitations of existing diagnostic methods have led to several problems and pain points. The most commonly used visual field (VF) examination modality, SAP, can miss early symptoms and early changes in disease progression, resulting in delayed intervention and progressive vision loss. Furthermore, in the early stages, traditional SAP is prone to a high false-positive rate, meaning that overt visual field loss may be missed.

[0007] Furthermore, other issues exist because healthcare providers must rely on suboptimal / unreliable and imprecise diagnostic and monitoring tests, limiting their ability to accurately assess disease progression. The inconsistent characterization of functional deficits, often inconsistent with anatomical changes, increases uncertainty in glaucoma diagnosis and treatment management.

[0008] Furthermore, because glaucoma visual function (VF) and / or functional tests are subjective, factors affecting patient experience can directly impact a healthcare provider's ability to monitor the disease. Patients often find it difficult to complete traditional SAP tests for various reasons, including a lack of engagement. For example, traditional tests are cumbersome, requiring intense concentration and strict fixation on a central target, and may require uncomfortable postures, as patients must lean forward in a hunched position and maintain this posture throughout the test, negatively impacting the reliability and repeatability of test results.

[0009] Therefore, to address these issues, there is a need for systems and methods for diagnosing and monitoring glaucoma using adaptive VF analysis on head-mounted displays such as VR headsets. Summary of the Invention

[0010] This invention discloses a revolutionary system and method for diagnosing and monitoring glaucoma, which implements groundbreaking adaptive visual field (VF) methods and analytical tools through head-mounted displays such as virtual reality (VR) headsets. The innovations described herein allow for improved testing that captures functional and structural changes in the patient's eye, thus enabling applications in eye care, health, and monitoring, and allowing for user-specific fidelity monitoring.

[0011] This disclosure discloses head-mounted display-based systems, methods, and analytical tools that offer a revolutionary breakthrough in glaucoma diagnosis by providing visual field examinations for the diagnosis and monitoring of glaucoma. The head-mounted display-based systems and methods include autonomous systems and methods capable of detecting regions of interest (e.g., known or suspected damaged retinal areas) and performing spatial analysis to iteratively select new test locations. Such spatial analysis generates spatial mapping data of scotoma defects. This spatial mapping provides high-fidelity analysis and output with a resolution far exceeding that of traditional 24-2 test location patterns for the user's eye. The spatial mapping of scotoma is adaptive and unique to each user (i.e., user-specific), and its results enable the detection or identification of structural defects.

[0012] As described herein, a key feature of the head-mounted display-based systems and methods is that, unlike conventional static automated perimetry (SAP), where the test position has a visual interval of 6° in the vertical and horizontal directions and 8.5° diagonally, the head-mounted display-based systems and methods running on a VR headset implement an adaptive map-based perimetry algorithm, enabling test intervals as small as 0.5°. For a typical Goldman III-sized stimulus (radius 0.43°), this resolution is 12 to 17 times that of SAP. Depending on the limitations of the head-mounted display, for even smaller stimuli (e.g., a Goldman I-sized stimulus with a radius of approximately 0.1°), the resolution can be as small as 0.1°. This high fidelity improves the sensitivity for detecting spatial progression of scotomas. Furthermore, compared to conventional SAP, the head-mounted systems and methods of this disclosure offer one or more improvements, a drawback of which is the presentation of stimuli at a fixed location regardless of retinal structure. In contrast, the head-mounted display-based systems and methods introduce a novel user-specific approach, enabling real-time adaptive adjustments as the patient undergoes testing. The difference between head-mounted display-based systems and methods and traditional methods lies in their use of technology to implement adaptive map-based field of view checks, making such tests feasible and readily available, and significantly improving the granularity and accuracy of glaucoma progression monitoring.

[0013] Based on the foregoing and the description of this disclosure, this disclosure improves computer functionality or other technologies at least as follows: for example, by implementing an adaptive map field of view (APLA) algorithm on a head-mounted display (HMD), it improves the field of glaucoma diagnosis and monitoring, increasing image granularity and accuracy in glaucoma progression monitoring. Furthermore, the HMD itself is also improved by installing or implementing standard databases and / or models configured to adapt to different HMD configurations, which may have different hardware configurations and corresponding limitations. In other words, standard models and / or their associated standard databases allow the ALA algorithm to be implemented on a variety of HMDs, regardless of their differences in hardware and limitations, including displays, field of view, processors, etc. This is at least an improvement over existing technologies because traditional glaucoma monitoring methods lack the adaptability and accuracy provided by the ALA algorithm implemented on a HMD.

[0014] Furthermore, this disclosure includes the application of certain elements using or employing specific machines, such as VR headsets or devices or other head-mounted displays.

[0015] This disclosure includes specific features beyond those known, conventional, and traditional in the art, and / or adds non-traditional steps that limit this disclosure to specific useful applications, such as systems and methods for glaucoma diagnosis and monitoring via adaptive VF analysis on head-mounted displays such as VR headsets.

[0016] The advantages of the present invention will become more apparent to those skilled in the art from the following description of preferred embodiments, which are shown and described by way of example. It should be understood that other different embodiments may be possible, and details can be modified in various aspects. Therefore, the drawings and descriptions should be considered illustrative rather than restrictive. Attached Figure Description

[0017] The accompanying drawings described below illustrate various aspects of the disclosed systems and methods. It should be understood that each drawing depicts one embodiment of a particular aspect of the disclosed systems and methods, and each drawing is intended to be consistent with possible embodiments. Furthermore, wherever possible, the following description refers to the reference numerals included in the drawings, wherein features depicted in the plurality of drawings are identified using consistent reference numerals.

[0018] The accompanying drawings illustrate the arrangement currently under discussion; however, it should be understood that this embodiment is not limited to the precise arrangement and equipment shown, wherein: Figure 1 Examples of methods for diagnosing and monitoring glaucoma via adaptive visual field (VF) analysis on a head-mounted display such as a virtual reality (VR) headset are shown according to various embodiments of the present disclosure.

[0019] Figure 2A This paper presents an example spatial clustering analysis of defect locations from a 24-2 conventional static automated perimetry (SAP) inspection to identify dark spots (locations marked "2" in the example) and their edges (solid black circles and connecting lines), which can be... Figure 1 The example methods or adaptive map view checking algorithms used or accessed are specific to various embodiments of this disclosure.

[0020] Figure 2B The set of all possible test locations that the adaptive algorithm can select for testing is shown, with a horizontal and vertical resolution of 2° between the points.

[0021] Figure 3A An example implementation of the Janssens diagram defined relative to the center of the blind spot (point 310) is shown, specifically according to various embodiments of this disclosure.

[0022] Figure 3B It shows the relationship with Figure 3A Examples of the distribution values ​​related to the Janssens space, specifically according to various embodiments of this disclosure.

[0023] Figure 3C It shows the relationship with Figure 3A Another example of the distribution values ​​related to the Janssens space, specifically according to various embodiments of this disclosure.

[0024] Figure 4 A graph showing the sensitivity of a healthy eye at different angles (in decibels, dB) is illustrated, specifically according to various embodiments of this disclosure.

[0025] Figure 5A The diagram illustrates the relationship between test time (in seconds) and true sensitivity (in dB) in various tests including a variety of experiments, specifically according to various embodiments of this disclosure.

[0026] Figure 5B A second graph showing the relationship between test time (in seconds) and true sensitivity (in dB) in various tests including a variety of experiments is shown, specifically according to various embodiments of this disclosure.

[0027] Figure 6 A flowchart is shown illustrating the use of visual field feedback to provide feedback during the performance of visual field testing, specifically according to various embodiments of this disclosure.

[0028] Figure 7 Example implementations of a visual field feedback (VFF) system for user operation are shown in various embodiments of this paper.

[0029] Figure 8 An example visual feedback (VFF) method for automatically evaluating visual field tests is shown, specifically according to various embodiments of this disclosure.

[0030] Figure 9 Another example visual field analysis method for diagnosing and monitoring glaucoma by implementing adaptive map VF checks and optionally performing visual field tests is shown, specifically according to various embodiments of this disclosure.

[0031] Figure 10A A diagram of a visual field analysis (VFA) system is shown, specifically according to various embodiments of this disclosure.

[0032] Figure 10B An example mean visual hill (HoV) model used in various embodiments of this paper is shown, relative to an age-corrected mean normal vision model.

[0033] Figure 10C Example diagrams illustrating the components of the average visual hill (HoV) model used in various embodiments of this paper are shown.

[0034] Figure 10D Example polynomials for describing the final surface asymmetry are shown in various embodiments of this paper.

[0035] Figure 10E Example personalized visual hill (HoV) models used in various embodiments of this paper are shown, relative to age-corrected average normal visual models.

[0036] The accompanying drawings depict preferred embodiments for illustrative purposes only. Alternative embodiments of the systems and methods shown herein may be employed without departing from the inventive principles described herein. Detailed Implementation

[0037] The virtual reality (VR) system and method described in this paper enables innovative automated adaptive visual field testing, allowing for earlier disease detection and identification of glaucoma disease progression. The adaptive strategy described provides high-fidelity image data insights into visual field (VF) deficiencies, achieving levels of patient care optimization currently unattainable through personalized treatment and monitoring. The VR system and method leverage VR technology to create patient-centered solutions that overcome the inherent limitations of traditional techniques in the field of glaucoma.

[0038] The VR system and method described in this paper implement adaptive map-based visual field testing, an advancement in glaucoma testing. This VR system and method enables automated adaptive visual field testing for real-time diagnosis and testing of glaucoma. Such VR systems and methods can use or can input user-specific data, rather than traditional reference data, which is more prone to false positives.

[0039] Glaucoma presents inconsistently across patients. Therefore, VR systems and methods implement VF testing tailored to specific users or patients. The VR systems and methods described herein implement head-mounted VF testing using an adaptive visual field examination algorithm that is both adaptive and user-specific (e.g., patient-centered). The adaptive mapping visual field examination algorithm provides automation and uniformity through data analysis, as well as personalized and enhanced capabilities for mapping spatial patterns of eye damage to the user (e.g., creating or determining spatial maps), without requiring trained clinicians. The adaptive mapping visual field examination algorithm utilizes anatomical models of retinal neural trajectories (e.g., Janssensky maps) and glaucoma pathophysiology, self-adjusting based on the patient's responses during the test. This differs significantly from traditional VF testing methods, which do not adapt to the patient's disease stage or specific patterns of eye damage. In this way, the VR systems and methods achieve a patient-centered approach to disease characterization.

[0040] In addition to this implementation method, the VR systems and methods described herein also provide active fixation correction for testing and improve reliability through eye tracking and gamified interfaces. For example, gamification in a VR environment makes the testing process easier and more enjoyable for patients, which significantly increases engagement and attention, thereby improving data fidelity and eliminating false positives and / or false negatives or other errors. In contrast, one of the most significant problems with traditional SAP is the reliability of test results. For example, a retrospective study of glaucoma patients found that approximately 48% of tests based on the Humphrey Field Analyzer (HFA) were unreliable. This is usually attributed to fixation loss, which is affected by factors such as fatigue and boredom and directly impacts the results. The more reliable VF results provided by the VR systems and methods described herein may also improve test-retest variability in addition to the specificity of the VF test. The VR systems and methods disclosed herein also address the problem of low engagement by applying various gamification elements, including feedback mechanisms, immersive environments, progress trackers / reflectors, scoring systems, and stylized interfaces. These elements increase user engagement and attention, thereby improving the device's output quality.

[0041] The VR systems and methods described herein are configured to be device-independent, for example, configured to execute on head-mounted displays or other VR devices with different characteristics, hardware, and / or limitations. This is achieved by implementing an adaptive map view inspection algorithm on one or more processors communicatively coupled to the head-mounted device (including VR headsets). The head-mounted device (including VR headsets) can be one of a variety of head-mounted devices, each with different configurations, such as different hardware (e.g., displays, cameras, sensors, mounting hardware, and / or variations thereof). The adaptive map view inspection algorithm adapts to the specific configuration and limitations of the head-mounted display to achieve uniform operation, unaffected by hardware and / or limitations, enabling the VR systems and methods to operate adaptively on various devices. In this way, the adaptive map view inspection algorithm can be programmed, adjusted, or otherwise based on the limitations of each individual device with minimal effort. In various embodiments, the adaptive map view inspection algorithm is configured to identify various types of occult spots, including but not limited to paracentral occult spots, arcuate occult spots, nasal occult spots, vertical occult spots, and other types of occult spots, despite differences in the hardware of one or more VR devices. Additionally, or alternatively, in some implementations, the adaptive map view inspection algorithm may also be configured or programmed to perform on other electronic devices, including but not limited to tablet devices, computer screens, or other electronic devices having screens for displaying or presenting graphic images or content (e.g., VR content) as shown herein.

[0042] VR systems and methods can also provide adaptability across different head-mounted displays and / or VR devices by implementing and using standard databases and / or proprietary standard reference models, such as those described herein. For example, observational studies have been conducted on the eyes of subjects using adaptive map visual field testing algorithms. In various aspects, the adaptive map visual field testing algorithm includes a thresholding algorithm that uses standard data obtained from or output by the ZEST algorithm. That is, in various aspects, the adaptive map visual field testing algorithm can use standard data, such as standard data from standard databases and / or standard models, provided as the output of the ZEST algorithm to identify healthy areas of the retina in previous visual field tests, thereby estimating the patient's general height (GH). For better use, the adaptive algorithm includes estimating GH through rapid, direct measurements of the eye. Unified analytical mapping, tools, and visual aids can be used to display or otherwise interpret visual field results based on the output of the adaptive algorithm.

[0043] Example Adaptive Map Visibility Check Algorithm This section describes an example adaptive map view checking algorithm. It should be noted that the examples in this paper describe one implementation of the adaptive map view checking algorithm, and other different implementations may also be used. In this example, the adaptive map view checking algorithm includes an overthreshold test that requires at least a VF location suspected of having functional impairment as a seed, for example, estimating GH to obtain a reference VF from which a location-related overthreshold level is derived.

[0044] In some implementations, the adaptive map visual field testing algorithm implements a glaucoma pathophysiology model. The algorithm can also integrate active eye correction and use standard models and / or standard databases to select overthreshold visual stimuli. Furthermore, the algorithm can automate the initial VF location selection method. The initial VF location, overthreshold visual stimuli, and / or other parameters can be used as input by the algorithm to iteratively identify the damaged location by presenting or displaying (e.g., via a VR device's display) one or two visual stimuli and recording whether the patient responds to the first trial (in which case the algorithm stops testing at that location) or misses one or two visual stimuli.

[0045] Spatial analysis of damaged location clusters from previous VF SAP tests (e.g., 24-2, 30-2, 10-2, adaptive view inspection) can be used to determine the initial set of test locations for the adaptive map view inspection algorithm. Spatial analysis can be used to obtain damaged location clusters and identify upper and lower boundary locations (e.g., cluster edges), for example, as... Figure 2A As shown and described. For each cluster, the adaptive map view inspection algorithm can also be applied by fitting the retinal nerve fiber bundle trajectories to the cluster edges (as shown and described). Figure 2A To estimate the shape and size of dark spots, thus enabling the Janssens structure-function diagram ( Figure 3A Personalized to suit the patient's anatomy and specific defect areas. This can be done according to a given vertical and horizontal resolution, for example, Figure 2B The 2° setting determines a set of possible initial test locations for use by the adaptive map view check algorithm. The adaptive map view check algorithm can select all or some of these locations as the initial test location pool. The adaptive map view check algorithm iteratively updates the estimated dark spots based on user feedback and can add or remove new untested locations from the initial pool. The resolution between locations can vary between 0.1° (approximately Goldmann I size) and 6° (e.g., intervals in 24-2 and 30-2 regular grids).

[0046] It should be noted that additional and / or different overthreshold values ​​and / or strategies may be used. Furthermore, in some implementations, spatial analysis of damaged site clusters may be performed based on Voronoi mosaicking. Additionally, in some implementations, instead of using Jensen's diagrams, which can provide average bundle paths for different populations, Jensen's diagrams or other values ​​may be used to obtain personalized fits. In some implementations, the selection of new test sites may extend beyond the superior and inferior bundles to ensure that the patient can see the visual stimuli there (e.g., thereby confirming the edges of the scotomas). The adaptive map visual field inspection algorithm may stop when all test sites around the scotomas are determined to be undamaged, or when the operator-defined test area limit is reached.

[0047] Figure 1 An example workflow of the example adaptive map visual field inspection algorithm is illustrated. At box 102, parameters or values, including conventional SAP values, direct eye measurements, or other values ​​related to glaucoma measurements, are captured. These parameters or values ​​are provided as input to the adaptive map visual field inspection algorithm to produce an estimated GH value 104. The average normal value 108 of the visual hill is analyzed against a reference visual field value 110 to generate a decibel difference (∆dB) value 106, which is the attenuation reduction (in decibels) used to generate the overthreshold visual stimulus value 112. The overthreshold visual stimulus value 112 can be used at each location of the VF. At box 114, various techniques 114, including conventional static automated visual field inspection, screening visual field inspection, and other such techniques, can be used to select (box 116) the initial test location. At box 118, the initial test location and visual stimulus value 112 are provided as input to the adaptive map visual field examination algorithm, which can be used for any type of spatial glaucoma injury and utilizes available results from patients who have previously undergone conventional (24-2, 10-2, 30-2) visual field examinations or (beyond the threshold) screening tests. The adaptive map visual field examination algorithm can perform an adaptive map visual field examination using the initial test location. The output of the adaptive map visual field examination can then be analyzed to perform spatial analysis, thereby generating a topographic map 120 with summary statistics of the user's ocular glaucoma injury.

[0048] Example determination of parameters for adaptive map view inspection algorithm Adaptive map vision checking algorithms can be configured to take on various parameters or values ​​to adapt the algorithm to a specific device and / or user. In this way, when implemented by one or more processors, the adaptive map vision checking algorithm can be device-independent and configure the VR device to be patient-centric / user-specific. In some implementations, the adaptive map vision checking algorithm accesses standard datasets or databases (e.g., referred to herein as the Envision standard dataset or database) and / or standard models generated from standard datasets or databases. For example, standard datasets or databases can be used for (1) age-corrected mean normal visual field, and (2) standard references for detecting VF locations with abnormally low visual sensitivity. Age-corrected spatial models can be generated for normal visual hills and / or normal isometry values.

[0049] In some respects, the model can use quantile surfaces and correlation values, which can provide a standard reference for the detection of anomalous VF locations. This approach can improve statistical power and avoid the limitations of empirical standard references used in traditional SAP. Personalized reference VFs can be derived from the standard model. This derivation of reference VFs is based on the fact that, after adjusting for age effects, there are many inter-individual differences in overall sensitivity (GH) and visual hill shape among healthy eyes. Figure 4 As shown, for 95% of healthy eyes, sensitivity can vary from 5 dB to 8 dB (406) or even higher, not only due to ocular opacities (e.g., cataracts) but also due to differences in attention and standards. The reference VF can be obtained by subtracting the difference between the user's GH and the average normal GH from the average normal visual field.

[0050] Figure 4 An example of age-corrected sensitivity is shown in a trained, healthy eye. Figure 4 The data shown are from the State University of New York-Indiana University dataset. The x-axis (angle 404) shows the horizontal visual angle. The y-axis (sensitivity (dBs) 402) shows the sensitivity of 91 healthy eyes from 91 subjects at a y-angle of 3°. Curves (408c1 and 408c2) show the 95% confidence interval for each x-value on the x-axis 404. Interval lengths are shown in dB 406. The rectangular area 410 represents the blind spot.

[0051] Adaptive map view inspection algorithms can be implemented based on these parameters or values. For example, a 24-2 total deviation (TD) probability map can be used to identify healthy areas of view (VF) at 10 appropriately selected locations from the healthy areas (e.g., Figure 2AA fast (e.g., 10 renders) simplified thresholding algorithm is run on the gray points in the map to estimate the GH. The reference field of view can be obtained by subtracting the difference between the average normal GH and the user's GH from the age-corrected average normal field of view. This personalization achieved through the reference field of view is expected to improve the sensitivity of map VF inspections.

[0052] In some implementations, a standard dataset or database can be used to determine specificity at different overthreshold levels. A function can be implemented to determine the overthreshold level to use at any given specificity (e.g., 95%). If no prior VF SAP data is available, an initial screening test can be performed with overthreshold stimuli corresponding to the 0.05 and 0.01 percentiles (probability levels) for each location, according to the Envision Standard Reference. For example, such a screening procedure can be implemented as follows: A stimulus corresponding to the 0.05 percentile is presented or displayed (e.g., on a VR device) at each location. If seen or otherwise detected, the adaptive map vision check algorithm stops, and the location is marked as healthy. If not seen, a stimulus corresponding to the 0.01 percentile is presented or displayed (e.g., on a VR device). If seen or otherwise detected, the location is assigned a probability level of 0.05. If not seen or detected, a probability level of 0.01 is assigned. This implementation provides a more efficient and / or faster test than SAP's thresholding algorithm. Initial locations are then obtained from the TD probability map described herein.

[0053] Figure 2A An example of spatial analysis for identifying clusters of damaged locations is shown. For a field of view of 200 at a center + / - 30°, spatial analysis can be used to identify clusters of damaged locations, for example, as described herein, for a specific VF test of a 24-2 pattern with defects. This implementation can be performed to determine the starting location of the map view inspection described herein. This analysis and grouping can be applied or implemented (e.g., by an adaptive map view inspection algorithm) to plot dark spots using the map view inspection. Figure 2A As shown, the edges of the two clusters (e.g., cluster 201 defined by position "1" and cluster 202 defined by position "2") are fitted with paths from the Janssens structure-function map, thus estimating the shape and size of the two sclera in a manner consistent with glaucoma pathophysiology. Figure 2B As shown, the implementation of map view inspection can then be used to confirm defects within dark spots seen in the 24-2 pattern based on their location within the original estimate, and to test outside of them to determine a more accurate shape estimate than using the traditional 24-2 pattern alone.

[0054] More generally, this calculation can be based on a square grid with a specific resolution (e.g., Figure 2B This is achieved using 2° (in the formula). For example, Figure 2B The square grid illustrates square grid portion 251. In some embodiments, the shell surrounding a given cluster may be positioned or fixed at half the resolution of the given square grid (e.g., square grid portion 251), for example, Figure 2A Cluster 201 is shown as a square region labeled "1". Additionally, or alternatively, the edge location or edge of a given cluster can be used to determine the upper and lower edges of each dark spot. Furthermore, or alternatively, the cluster edges of one or more clusters can be used to fit a Janssens path surrounding the site of eye damage (e.g., as shown in the image). Figure 2A (As shown).

[0055] Figures 3A-3C This paper discusses the implementation of the Jensen's structure-function diagram. Typically, the Jensen's diagram uses a polar coordinate set (r, φ) to mathematically describe the trajectory of retinal nerve fiber bundles in a modified Euclidean space via nonlinear functions; this space is referred to as the Jensen's space. The equations used to describe the paths are as follows: (1) In equation (1), r0 is the radius of the circle at the starting point of the path (interpreted as the optic nerve head, ONH, typically a 4° angle of view). φ0 = φ(r0; φ0) is the angular position of the trajectory at the starting point (interpreted as the angle of incidence on the ONH). b and c are parameters that depend on the retinal region (superior, inferior, temporal, nasal).

[0056] After obtaining the (r, φ) pairs of the path with the incident angle φ0, these values ​​can be converted back to Cartesian coordinates in retinal space, which is inverted relative to VF coordinates. The conversion between retinal space and Janssenis space depends on the location of the blind spot center (e.g., Figure 3A Blind spot 310 shown. For example, Figure 3A The blind point 310 is shown to be located at (15, -2). Parameters b and c depend on the incident angle φ0 on the ONH.

[0057] This model can be implemented in any programming language, such as R or C#, and can be implemented as follows: Figure 3A The depiction is shown below. Figure 3A This includes VF coordinates in degrees (e.g., -30 to +30 degrees on x-axis 304 and -30 to +30 degrees on y-axis 302). The model supports the application of map-based visual field examination to glaucoma, particularly but not limited to wedge-shaped and arcuate scotomas. The model can be enhanced by making it user-specific / patient-centered.

[0058] For each cluster of damaged locations, by fitting equation (1) to the upper and lower edge points of the cluster (after transforming their coordinates from the view space to the Janssensky space, as shown in the figure) Figure 2A (As shown), to estimate the values ​​of parameters b and c that define the upper and lower edge bundles of the blind spot. In one implementation, the patient's blind spot center location can be input into a transformation in Janssensky space to adapt the implementation to the user.

[0059] Additionally, or alternatively, other user-specific implementations may include updating or altering anatomical features, such as the angle of the temporal retinal suture and assumptions about how ganglion cell bundles propagate in the retina.

[0060] After selecting a starting position from the initial position pool, for example, Figure 2B The initial location pool can be used as part of the adaptive map view check to iteratively update the dark spot estimates. Each iteration can be as follows: present the visual stimulus and record the results of the initial location; after collecting all results, perform cluster analysis and fit using equation (1) to obtain updated dark spot estimates; modify the pool of possible test locations based on these estimates; and select new untested locations. The adaptive algorithm can be repeated for each dark spot until the estimated dark spot does not change significantly relative to the previous iteration, or all surrounding locations are determined to be undamaged, or the operator decides to interrupt the test and save the results.

[0061] In some implementations, to avoid extreme and / or unstable fitting, the value of parameter b can be limited by an algorithm to produce or output reasonable values. The 95% distribution of parameter b can be determined as a function of φ0 (i.e., phi0). Figure 3B and 3C This was explained, with curves 356ll and 386ll representing the lower limit, and curves 356ul and 386ul representing the corresponding upper limit. Figure 3B ) and below ( Figure 3C The upper limit of half field of view. Figure 3B and 3C Each of these corresponds to a half of the visual field, for example, one of the two halves of the visual sensory field.

[0062] exist Figure 3B and 3C In both cases, the upper limit (356 ul and 386 ul) can be obtained by multiplying the lower limit (356 ll and 386 ll) by a factor e. 1.2 To calculate.

[0063] The outputs of these function fits can be obtained using the least squares method. If the fit itself is unreliable (e.g., based on reliability criteria), the average value of β can be used (e.g., -1.9 and 0.7 for the upper and lower half-fields, respectively).

[0064] Related aspects of adaptive map view inspection algorithms The disclosure described herein provides implementations of device-independent software (e.g., adaptive map field of view checking algorithms) for implementing or performing clinical functional tests in head-mounted displays (e.g., VR devices). This section provides example implementations or aspects of device-independent software (e.g., adaptive map field of view checking algorithms) and related devices (e.g., VR devices).

[0065] Traditional SAP Traditional SAP can be improved and automated through Bayesian strategies (e.g., ZEST algorithm implementation). Since the Swedish Interactive Threshold Algorithm (SITA) standard remains the most commonly used traditional strategy, it can be referenced for measurement or improvement. Hardware differences between VR devices can be quantified to further measure the differences between the ZEST and SITA algorithms and their respective retest variability. Thresholding methods can be used to compare hardware and setup differences for specific field meters, such as starting position.

[0066] The implementation can be used or based on the specific context of a given device (e.g., a VR device) (e.g., 10 candela / m² (cd / m²)). -2 Test location and test size (e.g., Goldman III size; diameter 0.43°).

[0067] Some aspects may include: 1. The output of the adaptive map field of view check algorithm is comparable to that of the Humphrey FieldAnalyzer (HFA), and the consistency limit is no greater than that between the HFA and any other commercial field of view meter.

[0068] 2. The output of the ZEST algorithm can be comparable to that of an HFA using the SITA standard, for example: The testing time for healthy eyes and glaucoma eyes is comparable (or shorter). The retest variability was the same as or less than that of HFA using the SITA standard.

[0069] Envision's ZEST-based algorithm As described in this paper, adaptive map view inspection can output results from the ZEST algorithm. ZEST is a thresholding algorithm equivalent to the SITA standard. The adaptive map view inspection algorithm is a super-thresholding method used to measure the extent and shape of dark spots, rather than the depth that ZEST or SITA attempts to measure. The adaptive map view inspection algorithm implemented using the ZEST algorithm may be referred to in this paper as Envision's ZEST-based algorithm or Envision's ZEST algorithm.

[0070] Additionally, or alternatively, in some implementations, a portion of the HFA dynamic range can be used to perform the test to make the test shorter or more efficient. For example, as Figure 5A and 5B Each of them, as shown, can be used in the range of 20 dB to 40 dB. Figure 5A and 5B The chart values ​​for each of these values ​​are generated by simulating test times (502, 552) at each location, where the full dynamic range of the Zester algorithm (e.g., as...) is... Figure 5A The true sensitivity (dB) value in 504 is shown in the figure, and the useful dynamic range (e.g., as shown in the figure) is also shown in the figure. Figure 5B The true sensitivity (dB) value of 554 in the figure is compared with that shown in the figure. Figure 5A and 5B The true sensitivity values ​​for each of these values ​​(504 and 554, respectively) are the same or similar within a range of approximately 20 dB to 40 dB. Therefore, limiting testing to this range can improve efficiency or reduce testing time by eliminating tests outside this range. It should be understood that this range may also include extended values, additional values, and / or different values.

[0071] Standard Database Standard databases or other standard datasets can be used to develop device-independent analysis methods for the ZEST algorithm and adaptive map view inspection algorithms. Standard models can be built from standard datasets, which, after device-specific calibration, can be used for different hardware configurations and limitations. In some implementations, quantile regression can be used to achieve standard values.

[0072] Some aspects may include: 1. Standard values ​​can be generated using quantile regression.

[0073] 2. Standard references can be adjusted through external biometrics (e.g., refractive power, axial length, corneal curvature, etc.) because the statistical sensitivity of the test is expected to improve as more additional (independent) information is incorporated.

[0074] 3. Models (e.g., standard models) can also be developed for calibrating standard references used in different devices, maintaining the device independence of the software.

[0075] Progress Analysis In some implementations, the Permutation of Pointwise Linear Regression (PoPLR) can be used in traditional SAP derived from ZEST tests. Permutation analysis of spatial damage progression can be used in adaptive map view inspection algorithms.

[0076] Some aspects may include: 1. PoPLR is used to provide progression analysis, identifying or detecting focal loss in VF. It can provide higher sensitivity compared to global indicators such as mean bias (MD) and other complex progression models such as Analysis with Non-Stationary Weibull Error Regression and Spatial Enhancement (ANSWERS).

[0077] Map view check In some implementations, the adaptive map visual field inspection algorithm can be implemented as a superthresholding method to define the extent of glaucoma scotomas, which is an improvement on the depth of defects at predetermined locations (e.g., 24-2 tests). Implementing map visual field inspection may include determining a starting test point. Such a starting point may be based on or derived from previous visual field inspection tests, such as 24-2, optical coherence tomography (OCT) results, or obtained through any other implementation (e.g., manually selected from fundus images) to obtain a region of interest or another 24-2 superthreshold screening test. Implementing map visual field inspection may also include determining a reference visual field, which can be obtained by directly estimating the GH of the VF.

[0078] Some aspects of map view inspection may include: 1. Implement and execute personalized models based on mathematical graphs of nerve fiber pathways (e.g., Janssens diagrams) that can be used to describe the pathophysiology of glaucoma.

[0079] 2. Map view check provides adaptive and self-correcting output because the test location is selected at runtime based on test results and status.

[0080] 3. Map view inspection can be performed efficiently by a processor (e.g., the processor of a VR device) because only the relevant area of ​​the view is tested, ignoring areas without glaucoma damage.

[0081] 4. Unlike any other visual field testing method, map visual field testing can use healthy areas of the retina as input to obtain a personalized reference visual field for each patient.

[0082] 5. The map-based visual field examination is performed in a patient-centered manner, where the reference visual field can be adjusted based on standard data for each patient, thus the overthreshold stimulation is personalized for each subject. Furthermore, the personalized model for describing scotomas based on the Janssens diagram is also personalized for each patient.

[0083] 6. Map view inspection can be used to derive important spatial metrics and view impairment features, similar to the metrics and features of dynamic view inspection (e.g., line-of-sight and dark spot mapping, such as sensitivity).

[0084] 7. Map-based field of view inspection eliminates three major limitations of traditional dynamic field of view inspection, which are direct results of manual operation. For example, the following can be eliminated: high retest variability; high inter-operator variability; and / or lack of standardized analytical methods.

[0085] 8. Map view inspection analysis can be based on novel spatial and statistical methods in view inspection, such as adaptive map view inspection algorithms.

[0086] 9. Because the assessment shifts from evaluating defect depth to evaluating the shape and extent of dark spots, map-based visual inspection is expected to provide a superior structure-function correlation compared to traditional visual inspection tests.

[0087] Implementation methods and modeling One or more processors may implement computational instructions including the software or algorithms described herein (e.g., an adaptive map view inspection algorithm). One or more processors may be processors of a head-mounted display device (e.g., a VR device) and / or processors of a computing device (e.g., a remote computing device) communicatively coupled to (e.g., via a network) a head-mounted display device and configured to send data to and receive data from the head-mounted display device.

[0088] In various implementations, visual stimuli, such as light stimuli, can be visualized or displayed on the head-mounted display device. Additionally, or alternatively, immersive multimedia images can also be visualized in the head-mounted display device. For example, immersive multimedia images are images acquired and / or generated for depicting or presenting on the display of a VR device. For example, a user, such as a patient, can view VR visualizations through a VR device. In various embodiments, VR can refer to an immersive user experience, where a user can experience the sensations of a three-dimensional (3D) environment. For example, in one embodiment, the visualization of immersive multimedia images can be used to create a real-world experience for a user, such as a patient. Immersive multimedia images can guide a user to look in a specific direction or move their eyes to achieve a VF test.

[0089] More generally, head-mounted display devices can include any computing device capable of visualizing or generating visual stimuli and / or immersive multimedia images to create a visual experience for the user. In some embodiments, for example, the head-mounted display device can be any commercially available electronic display device with a display screen. In the aspect where the head-mounted display device is a VR device, such VR devices can include any of the PICO device, Google Cardboard device, Google Daydream View device, Apple Vision device, Oculus Rift device, PlayStation VR device, Samsung Gear VR device, or HTC VIVE device. Each of these electronic display devices can use one or more processors capable of visualizing light, visual stimuli, and / or immersive multimedia images in VR. For example, the Google Cardboard VR device includes a VR headset that uses one or more processors of an embedded smartphone (e.g., a smartphone) to visualize immersive multimedia images in VR; in some embodiments, the smartphone can be a smartphone based on Google Android or Apple iOS, or other similar computing devices. Other VR devices (such as the Oculus Rift) can include VR headsets that use one or more processors of a computing device (such as a personal computer / laptop) to visualize light, visual stimuli, and / or immersive multimedia images in VR. A personal computer / laptop may include one or more processors, one or more computer memories, and software or computer instructions for performing visualization, annotation, or transmission of light, visual stimuli, and / or immersive multimedia images or VR visualizations as described herein. Additionally, other electronic display devices may include one or more processors as part of a head-mounted display, which may operate independently of the processors of different computing devices to visualize light, visual stimuli, and / or immersive multimedia images in VR. Electronic display devices may also include software or computer instructions for capturing, generating, annotating, enhancing, transmitting, interacting with, or otherwise manipulating such visualizations.

[0090] In some embodiments, the VR device's headset may include a focal length lens (e.g., a 40 mm focal length lens) to focus the user's gaze on VR visualization content (e.g., exemplary VR visualization content or images). For example, the distance from the VR headset shell to the screen interface (such as the smartphone screen interface in a Google Cardboard headset) when viewed through the focal length lens can create a VR experience for a user (e.g., a patient).

[0091] In embodiments where VR is not implemented, displays such as computer monitors, tablet screens, television screens, or other such screens may be used to present light, visual stimuli, and / or other images and / or visual effects, as described herein.

[0092] In various embodiments, the electronic display device may include embedded sensors and / or cameras that track the user's head movements and adjust the viewing angle to simulate an environment or other visual field testing area, giving the user the feeling of looking around in a 3D world or other testing area. In some embodiments, the embedded sensor may be a sensor associated with a mobile device or other computing device embedded in the electronic display device's head-mounted display. In other embodiments, the sensor may be part of the electronic display device itself. In some aspects, the camera and / or sensor may capture images of the user's eyes.

[0093] In various embodiments, the electronic display device may include or be communicatively coupled to an input control. For example, in some embodiments, the input control may be a button located on the head-mounted display device. In other embodiments, the button may include a magnet attached to the housing of the head-mounted display device, wherein the magnet interacts with a computing device embedded in the head-mounted display, such as a smartphone, enabling the computing device to sense (e.g., via a magnetometer located in the computing device) the movement of the magnet when it is pressed, thereby acting as an input source for the head-mounted display device. In other embodiments, the input control may include a separate joystick or wired or wireless controller that can be manually operated by a user to control the head-mounted display device and / or the visualization of the head-mounted display device. Furthermore, in other embodiments, the head-mounted display device or its associated smartphone or personal computer may allow commands to be input via voice or body gestures to control the head-mounted display device and / or the visualization of the head-mounted display device.

[0094] In various embodiments, input controls on the head-mounted display (HMD) device allow a user to interact with the visualization, whereby a user wearing the HMD device can provide input to analyze, view, enhance, annotate, or otherwise interact with the visualization. In some embodiments, a user can use input controls to select from menus or lists displayed on the HMD display screen. For example, the displayed menus or lists may include options for navigating or highlighting certain views or features of the HMD visualization. In other embodiments, graphics or items may be interactive or selectable with the HMD visualization. Furthermore, in other embodiments, a user can provide text, graphics, video, or other input to the HMD visualization to enhance or annotate it. In some embodiments, enhancement or annotation of the HMD visualization will result in the same enhancement or annotation appearing in the visualization (e.g., such as an immersive multimedia image), and / or vice versa. In various aspects, a user can participate in VF testing and can provide input via input controls and / or the display screen to indicate when an optical stimulus, visualization, immersive multimedia image, and / or other aspect or object presented on the display screen is seen or otherwise detected. For example, in various aspects, the HMD device or other electronic display device may include an HFA device for measuring a user's field of vision. In these respects, visual testing may include instructing the patient (e.g., via a head-mounted display device or other electronic display device) to fixate on a central target. The user can then use input controls to indicate when they see a visualization (e.g., a light stimulus). This visual field test can assess the user's retina's ability to detect visualizations or other stimuli at specific points within the visual field.

[0095] In various embodiments, the input control may be used in conjunction with a crosshair or other indicator visible to the user within the head-mounted display visualization, allowing the user to hover the crosshair or other indicator over a menu, list, graphic, text, video, or other item within the visualization, enabling the user to interact with the item, for example, by clicking, pressing, or otherwise selecting the input control to confirm the selection or otherwise manipulate the item focused by the crosshair or other indicator.

[0096] In various embodiments, the visualizations and / or immersive multimedia images may be submitted to a healthcare provider, such as a hospital or health service company, via a computer network. The computer network can be any computer network, such as the Internet (with or without security protocols such as Secure Sockets Layer and / or Transport Layer Security), a private network operated by the healthcare provider, or a secure Virtual Private Network (VPN) operated by the healthcare provider on the Internet. In some embodiments, users, such as patients or operators, may transmit immersive multimedia images and / or visualizations directly from a computing device, such as a smartphone and / or electronic display device, which may have been used to capture or generate immersive multimedia image visualizations and / or related data, such as VF test data.

[0097] In various embodiments, immersive multimedia images may be stored in one or more memories of a computing device. For example, in some embodiments, immersive multimedia images or videos may be stored in various image or video file formats, such as Joint Photographic Experts Group (JPEG), Portable Network Graphics (PNG), Tagged Image File Format (TIFF), bitmap (BMP), Moving Picture Experts Group (MPEG), or QuickTime (MOV).

[0098] Basic settings In some implementations, immersive multimedia images or videos can be displayed on the VR device or headset's display using a 3D engine or model. For example, the Unity engine or model can be used. The adaptive map field-of-view checking algorithm can be implemented to configure the VR device or headset as device-independent. The adaptive map field-of-view checking algorithm can load or implement device-specific parameters or values ​​defined according to physical and psychophysical units, such as viewing angle, cd / m², etc. For example, in some implementations, the dB scale can be based on the Weber contrast ratio or value, thus allowing for a device-independent implementation, unlike HFA and any other commercial perimeters that use a device-related dB scale based on maximum luminance difference attenuation. Furthermore, the presentation of stimuli can be configured to be at infinity, thus eliminating the need for lens correction or accommodation, unlike HFA, which requires compensation for the fact that stimuli are presented on the VR device's display (e.g., 30 centimeters or closer).

[0099] In various implementations, 3D modeling and / or VR techniques can be used to calculate the viewpoint of each stimulus in the x and y dimensions. For example, 3D modeling can be performed using VR to calculate all parameters identical to those in the real world, such as size and position. A 3D-to-two-dimensional (2D) planar world image (e.g., an immersive multimedia image) can be fitted with 3D parameters. These 3D parameters can be mapped to the virtual world to provide information and / or other visual aspects in the immersive multimedia image described herein.

[0100] In some implementations, the virtual projector can be rotated to an angle on the x and y axes and visualize or display stimuli at a distance that the user can visualize on the VR device's display.

[0101] Some aspects may include: 1. Define the virtual world using physical units (e.g., meters or degrees, but any unit can also be used) and distance to replicate the real world in a virtual world (e.g., VR space). This implementation reduces the modeling and conversion from physical units (e.g., meters or degrees) to computational units (e.g., image pixel positions on a screen). Furthermore, this implementation makes the definition of the virtual world device-independent, as each device can render or model the same units on its hardware, such as a display.

[0102] 2. Furthermore, the implementation of the virtual projector simplifies processor execution by reducing mathematical calculations and computational load, enabling the software to run efficiently and quickly. This implementation ensures that, provided the setup and calibration data are correct, the stimulus will be placed in the correct location, such as on the display of the VR device.

[0103] Other forms of visual field examination In some aspects, other forms of visual field inspection can be implemented. For example, adaptive map visual field inspection algorithms can be customized to replicate other types of visual field inspections, such as frequency-doubled visual field inspection and SWAP, although standard values ​​for these inspections are not included. In some implementations, dynamic automatic visual field inspection (with the same parameters as HFA and Octopus) can be implemented, but standard values ​​are not included. Furthermore, in some implementations, hybrid dynamic and / or static visual field inspections can be performed. Additionally, in some implementations, the visual stimulus can also be defined as anti-blurring.

[0104] Wider impact The VR systems and methods for diagnosing and monitoring glaucoma using adaptive VR analytics described in this article promote the health and well-being of patients using these innovative technologies. This includes early detection and monitoring of the disease. Its advantages include early detection of glaucoma and monitoring of its progression, for example, the ability to detect glaucoma-related visual field defects earlier and more accurately. The innovative technologies described in this article can identify and detect the disease at its earliest stages, when intervention and treatment are most effective. This translates into improved visual outcomes and quality of life for affected individuals.

[0105] Furthermore, the systems and methods described herein provide user-specific and / or personalized healthcare because the adaptive nature of our technology allows for personalized assessment of VF for each patient. By tailoring diagnostic and monitoring strategies based on individual patient characteristics, including baseline visual field data, this technology can help optimize patient care and inform treatment decisions.

[0106] Furthermore, the systems and methods described in this article can reduce healthcare costs. Early diagnosis and intervention for glaucoma can reduce healthcare costs associated with the management of advanced glaucoma, including surgical procedures and costs associated with severe low vision.

[0107] Furthermore, the systems and methods described in this article improve the field of eye diseases and care, advancing the scientific understanding of glaucoma, its progression, and the relationship between anatomical changes and functional impairments. This knowledge can drive further research and development in the field, potentially leading to breakthroughs in glaucoma treatment and other related eye diseases.

[0108] Furthermore, the systems and methods described in this article can improve quality of life by protecting vision, which is crucial for maintaining overall quality of life, independence, and productivity. People with low vision have a higher incidence of depression and anxiety. By helping with the early detection and management of glaucoma, this technology can help reduce the incidence of blindness.

[0109] Overall, the technologies described in this article have great potential to have a positive impact on patients' health and well-being.

[0110] View feedback / gamification Visual testing of the eye can be achieved by providing feedback to the test subject during an ophthalmological examination. This implementation may be referred to herein as gamification. More generally, visual field testing is an examination procedure and protocol that involves assessing the test subject, for example, a patient's visual field, by presenting stimuli at varying intensities within their line of sight. Traditionally, visual function testing procedures are error-prone and rely on the patient's attention. However, test subjects may lose focus due to difficulty in concentrating their eyes or attention, reducing the validity of test results. Furthermore, the lack of feedback during testing can lead to inaccurate data collection, resulting in complete failure of the visual field test.

[0111] Computational systems for visual field testing that incorporate visual field feedback (VFF) may include devices (e.g., virtual reality (VR) headsets) and / or one or more processors for providing feedback to test subjects to indicate their performance during the test. Such systems are referred to herein as VFF systems. Visual field feedback, or other feedback, may be provided to test subjects in response to their input following the presentation of stimuli. For example, in visual field testing, stimuli may take the form of different visual objects presented to the test subject at different light intensities. In various aspects, VFF may include feedback presented to test subjects during visual field testing and may take the form of one or more types of feedback, such as visual feedback, audio feedback, tactile feedback, somatosensory feedback, or a combination of multiple types of feedback. Furthermore, feedback may be used to assess the progress of test subjects, for example, how much test content remains for a given visual field test. In this way, test subjects can assess their test progress without the assistance of a test administrator. However, it should be understood that test administrators (e.g., medical professionals) may perform the tests without departing from the disclosure herein.

[0112] The VFF system and methods shown can be used for visual field testing, including but not limited to rehabilitation, diagnosis and / or treatment purposes after disease or injury, visual performance enhancement and / or visual function training.

[0113] In some respects, the feedback presented to a test subject (e.g., a patient) is related to the accuracy of the subject's response. In one example, positive feedback can be generated if the subject responds correctly to a stimulus. Positive feedback can be displayed (e.g., on a VR headset). Furthermore, the cumulative amount or percentage of positive feedback can be tracked (e.g., stored in computer memory).

[0114] Conversely, if the response is incorrect, negative feedback can be generated and displayed (e.g., on a VR headset). Furthermore, the cumulative amount or percentage of negative feedback can be tracked (e.g., stored in computer memory). In this way, feedback changes depend on the outcome. In various aspects, the VFF system is capable of providing feedback in real-time or near real-time, allowing test subjects to understand their performance as the visual field feedback system outputs feedback. In some aspects, identifiable patterns and patterns of feedback repetition can be generated based on the cumulative amount and / or percentage of positive and / or negative feedback. In some aspects, the cumulative amount and / or percentage of positive and / or negative feedback can be used to generate scores based on the quantity and / or comparison of positive and / or negative feedback. In some aspects, scores can be displayed to test subjects in real-time or near real-time as they perform visual field tests. Furthermore, in some aspects, the cumulative amount and / or percentage of positive and / or negative feedback can be used to generate graphical charts or images to graphically depict the cumulative amount and / or percentage of positive and / or negative feedback to the test subject. In some respects, specific graphs can be displayed (e.g., via a VR headset) based on the cumulative amount and / or percentage of positive and / or negative feedback, where one graph can be used for positive feedback or its cumulative value, and another graph can be used for negative feedback and / or its cumulative value.

[0115] In some embodiments, an artificial intelligence model can be trained and executed to generate feedback based on responses given by a test subject. In these respects, the model can be trained using training data in the form of test subject responses from current or previous tests (e.g., technical feedback, such as one or more button presses, eye movements, and / or other gestures by the test subject) to generate positive and / or negative feedback. More generally, in various embodiments, supervised or unsupervised machine learning procedures or algorithms can be used to train the machine learning model. The machine learning procedure or algorithm can employ a neural network, which can be a convolutional neural network, a deep learning neural network, or a combined learning module or program that learns from two or more features or feature datasets (e.g., response data from multiple users). The machine learning procedure or algorithm can also include natural language processing, semantic analysis, automated reasoning, regression analysis, support vector machine (SVM) analysis, decision tree analysis, random forest analysis, K-nearest neighbor analysis, Naive Bayes analysis, clustering, reinforcement learning, and / or other machine learning algorithms and / or techniques. In some embodiments, artificial intelligence and / or machine learning-based algorithms can be included as libraries or packages executed on one or more processors described herein. For example, libraries can include TensorFlow-based libraries, PyTorch libraries, and / or scikit-learn Python libraries.

[0116] Machine learning can include identifying and recognizing patterns in existing data (e.g., training an AI model based on user feedback indications or responses when a user uses a system) to facilitate prediction or recognition of subsequent data (e.g., applying the model to new responses from new individuals as input to the model to determine the model's output, i.e., feedback predictions or indications, such as a prediction accuracy value corresponding to the percentage of displayed stimuli correctly responded to by a test subject). As another example, training data can include feature data, such as user responses, including button presses (e.g., pressing a button on a VR device), eye movements of a test subject leaving or towards a specific area of ​​the VR device screen, correctly recognizing objects within the VR device, or other actions of the test subject. Additionally, or optionally, feature data can also include spatial mapping data, which can include mappings of spatial patterns of eye damage in users (e.g., test subjects). Spatial mapping data can be used to approximate visual information (e.g., the user's field of vision) to define whether a user is likely to see objects from damaged and / or undamaged areas of the eye. AI models can be trained using a variety of spatial mapping data from multiple users' eyes along with labeled data (e.g., test user responses, defining whether the user is able to see or detect a given stimulus). The output of an AI model can be an accuracy prediction based on spatial mapping data and / or test responses. The output of an AI model can also be used to update the AI ​​model.

[0117] Tag data may include feedback given to the user (e.g., positive and / or negative feedback as described herein, which may include positive feedback instructing the test subject to correctly identify the object through the VR device, negative feedback when the test subject misidentifies the object through the VR device, negative feedback when the test subject's eyes leave the VR device when a response should have been provided but was not; or other feedback instructing the test subject to negative or positive interaction during visual testing).

[0118] Machine learning models, such as the vision feedback test model described in some embodiments of this document, can be created and trained based on example data (e.g., “training data,” such as user response data) or data (which may be referred to as “features” and “labels”) to make effective and reliable predictions on new inputs (such as test-level or production-level data or inputs). In supervised machine learning, a machine learning program running on a server, computing device, or other processor can be provided with example inputs (e.g., “features”) and their associated or observed outputs (e.g., “labels”) so that the machine learning program or algorithm can determine or discover rules, relationships, patterns, or other machine learning “models” that map these inputs (e.g., “features”) to outputs (e.g., labels), for example, by determining and / or assigning weights or other metrics to the various feature categories of the model. Subsequent inputs can then be provided to these rules, relationships, or other models so that the model running on the server, computing device, or other processor can predict expected outputs based on the discovered rules, relationships, or models.

[0119] In unsupervised machine learning, servers, computing devices, or other processors may need to find their own structure in unlabeled example inputs. For example, servers, computing devices, or other processors may perform multiple training iterations to train multiple generations of models until a satisfactory model is generated, such as a model that provides sufficient predictive accuracy when provided with test-level or production-level data or inputs. The disclosures in this paper may use one or both of supervised and unsupervised machine learning techniques.

[0120] In various aspects, the feedback (e.g., stimulus) presented to the test subject can be a identifiable response, a change in environment, and / or a change in the test subject's field of vision due to the test subject's interaction. Furthermore, feedback can take any form, including visual, audio, tactile, or somatosensory feedback. Examples of visual feedback include shapes, images, videos, colors, and changes in light intensity. For example, this can manifest as small, localized flashes of colored light or images at the location of the stimulus (e.g., within or generated by a visual feedback system, such as the display and / or speakers of a VR headset or device). Examples of audio stimuli include sound effects, music, and speech. For example, this can manifest as a deep echo effect. Examples of somatosensory feedback are vibrations or other physical sensations. For example, this can manifest as a slight, controlled vibration of a controller. Additionally, feedback can be formatted as a data format (e.g., a normalized data format) that can be stored on a server and provided to the test subject via a computer network.

[0121] Figure 6 A flowchart 600 is shown illustrating the use of visual field feedback to provide feedback during the performance of visual field testing, according to various embodiments of the present disclosure. For example, Figure 6An exemplary implementation of the algorithm executed on the VFF system (block 608) based on the performance of the visual field test is shown. Figure 6 As shown in flowchart 600, stimuli can be presented on a device display (e.g., a VR headset display) (box 604) (box 602), and the test subject can respond to the presented stimuli (box 606). A response can take the form of the stimulus being recognized or otherwise acknowledged by the test subject through interaction with a designated controller (e.g., a designated controller of display device 604). A response can also take the form of the stimulus not being recognized by the test subject by not interacting with the controller. The VFF system then records and identifies whether the response is a negative recognition (box 610) or a positive recognition (box 614) of the presented stimulus. Examples of data collected during testing are listed below, but other available and appropriate data can also be collected. The VFF system can then provide the test subject with visual, audio, or somatosensory responses (e.g., negative and / or positive feedback). For example, if the recognition is negative, negative feedback is provided to the test subject (box 612). If the recognition is positive, positive feedback is provided to the test subject (box 614). For example, in... Figure 6 During the process, feedback can be positive or negative. If there are still unfinished parts of the visual test, the process can return to the starting box where the stimulus was presented to the test subject.

[0122] The feedback recorded and interpreted by the VFF system from a given test subject (e.g., a patient) may include, but is not limited to, the following data: (a) whether the test subject (e.g., a patient)’s response was successful or unsuccessful; (b) the test subject’s (e.g., a patient’s) response time; (c) the location of the stimulus; (d) the test subject’s (e.g., a patient’s) gaze location; (e) the test subject’s eye movements; and (f) the response time (e.g., the response time of the VFF system).

[0123] The accuracy of the test subject's response can also be determined by analyzing the above responses and comparing them with the actual location of the stimulus or the predicted response. Predictive responses can be generated using artificial intelligence models trained on response data and / or relevant feedback and / or indications from the current test subject or other test subjects, for example, as described in this paper.

[0124] For example, as shown in box 602, stimuli are presented to the patient using a device display (e.g., a VR headset). For instance, the stimuli may take the form of various visual objects presented to the test subject at different light intensities.

[0125] Furthermore, at box 604, one or more processors can be used as part of the device display to present stimuli to the patient. For example, the device display could be a VR headset, a computer or tablet screen, or other types of displays.

[0126] At box 606, one or more processors can be configured to receive patient responses. Responses can take the form in which the test subject interacts with a designated controller to indicate that they have identified and acknowledged the form of the stimulus presented. Responses can also take the form in which the test subject does not interact with the controller to indicate that they have not identified the form of the stimulus.

[0127] At box 608, one or more processors may be configured as part of a VFF system to record and identify responses as negative responses (box 610) or positive responses (box 612).

[0128] At box 612, one or more processors can be configured to present negative feedback to the patient. At box 614, one or more processors can be configured to present positive feedback to the patient. For example, the feedback may include visual feedback, audio feedback, haptic feedback, or tactile feedback.

[0129] If further testing is required, the algorithm in flowchart 600 can initiate a new session by returning the implementation to box 604 to present the next stimulus to the patient. If no further testing is required, the method terminates, and the testing is complete.

[0130] Figure 7 Example implementation 700 of the VFF system during user operation is shown in various embodiments of this document. For example, Figure 7 An exemplary implementation of the VFF system is shown, in which the feedback is both audio and visual. In some aspects, such as... Figure 7 As shown, the feedback is positive, including audible responses to correct ones and visual representations of the patient's response. In some respects, the visual representations may be color-coded to indicate correct and / or incorrect responses. For example, as shown in box 702, a stimulus is presented to the patient. The stimulus can be any one or more of the stimuli described herein.

[0131] Furthermore, at box 704, one or more processors can be used as part of a device display to present stimuli to the patient. For example, the device display can be a VR headset, a computer or tablet screen, or a display of other types. The one or more processors can be the processor of the VR headset itself and / or the processor of a computing device communicatively coupled to the VR headset.

[0132] At box 706, one or more processors may be configured to record a patient's response to a displayed stimulus. The patient's response may be recorded by collecting user input (e.g., input device), tracking the patient's head position or gaze, or other input that collects and records the patient's response to the stimulus.

[0133] At box 708, one or more processors may be configured as part of the VFF system to record and identify responses; for example, recording may occur in computer memory. In these respects, responses may be identified as positive responses, which may include the patient correctly recognizing an object presented on the display.

[0134] At box 710, one or more processors may be configured to determine that positive feedback should be presented to the patient. Positive feedback may include the feedback described herein.

[0135] At box 712, one or more processors may be configured to provide positive feedback to the patient. The feedback may take the form of any feedback or instruction described herein.

[0136] Figure 8 An example visual feedback (VFF) method 800 for automatically evaluating visual field tests is illustrated, specifically according to various embodiments of this disclosure. For example, as shown in block 810 of method 800, one or more processors may be configured to display stimuli to a test subject (e.g., a patient) on a head-mounted display device (e.g., a virtual reality (VR) device). For example, the one or more processors may be one or more processors of the visual feedback system described herein, or one or more processors of a personal computing device or other similar processing device.

[0137] Furthermore, at box 820, method 800 includes recording the test subject's response to the displayed stimulus via one or more processors. For example, in some aspects, the test subject's response may include recognizing and acknowledging the stimulus by interacting with a controller, or not recognizing or acknowledging the stimulus by not interacting with the controller. Additionally, the test subject's response may include (a) whether the test subject's (e.g., a patient's) response is a hit or a miss; (b) the test subject's (e.g., a patient's) response time; (c) the location of the stimulus; (d) the test subject's (e.g., a patient's) fixation location; (e) the test subject's eye movements; and (f) the response time (e.g., the response time of a VFF system).

[0138] Furthermore, at box 830, method 800 includes analyzing a test subject's response to a displayed stimulus using a visual feedback system (e.g., which may include a VR headset) with the stimulus location. For example, the test subject's response can be analyzed based on the location of the displayed stimulus and can be fed into an artificial intelligence model executing on one or more processors. The artificial intelligence model can be trained using response feedback from the test subject (e.g., button presses, eye movements, or other such feedback provided by the user, including feedback described herein), wherein the artificial intelligence model is configured to output a prediction that defines an accuracy value corresponding to the percentage of times the test subject correctly responds to the displayed stimulus. The artificial intelligence model can generate feedback for the test subject based on the prediction that defines the accuracy value.

[0139] Furthermore, at box 840, method 800 includes providing feedback to the test subject based on response and analysis. For example, the feedback can be positive or negative feedback as described herein. Additionally, the feedback can include any of visual, audio, tactile, or sensory feedback. In another example, the feedback can be related to the accuracy of the test subject's response. In another example, positive feedback is displayed if the test subject responds correctly, and negative feedback is displayed if the test subject responds incorrectly. In yet another example, the feedback can be formatted as it can be stored on a server and provided to the test subject via a computer network, as described herein.

[0140] The visual feedback method shown can be used for any of the following purposes: visual field testing, rehabilitation after disease or injury, diagnosis and / or treatment, visual performance enhancement, or visual function training.

[0141] Figure 9 Another example visual field analysis method 900 for glaucoma diagnosis and monitoring is illustrated by implementing adaptive map visual field inspection and optionally performing VF testing, specifically according to various embodiments of this disclosure. Method 900 can be implemented, for example, on one or more processors described herein. This may include one or more processors of a VR headset and / or one or more processors of a computing device communicatively coupled to a VR headset. The disclosure herein applies to... Figure 9 The publicly available content.

[0142] refer to Figure 9 At block 910, method 900 includes performing a VF test on the display screen of an electronic display device (e.g., a virtual reality (VR) device). As described herein, the electronic display device may include a display screen positioned near or within the user's eye range, and the electronic display device is communicatively coupled to one or more processors. Furthermore, the VF test may be adapted to the region of the display screen based on values ​​in a standard database or model, wherein the VF test is presented as the same or similar visualization to one or more different electronic display devices having different displays with different corresponding shapes, formats, sizes, and / or resolutions, wherein the adaptive map view inspection algorithm is configured to access or implement the standard database or model to make the electronic display device device-independent relative to one or more differently configured electronic display devices.

[0143] At box 920, method 900 includes receiving visual test data indicating the user's field of vision.

[0144] At box 930, method 900 includes detecting one or more user-specific initial test locations based on visual test data, the one or more initial test locations defining one or more healthy clusters indicating no scatoma (error words) and one or more damaged clusters indicating dark spots.

[0145] At box 940, method 900 includes generating a spatial map that identifies one or more locations of damaged clusters based on one or more initial test locations.

[0146] Optionally, method 900 includes blocks 950-980.

[0147] At box 950, method 900 includes displaying stimuli to a test subject (e.g., a patient) on a head-mounted display device (e.g., a virtual reality (VR) device).

[0148] At box 960, method 900 includes recording the test subject's response to the displayed stimulus via one or more processors.

[0149] At box 970, method 900 includes analyzing a test subject's response to a displayed stimulus using a visual feedback system (e.g., a VR device) with stimulus location and / or spatial mapping. For example, the stimulus location could be within or generated by a VFF system, such as the display screen and / or speakers of a VR headset or device. Examples of stimuli (e.g., feedback) could include small, localized flashes of color or images at the stimulus location. Examples of audio stimuli include sound effects, music, and speech. For example, this could manifest as a deep echo effect. Examples of somatosensory stimuli (e.g., feedback) are vibrations or other physical sensations. For example, this could manifest as a slight, controlled vibration of a controller. Furthermore, such feedback could be formatted as a data format (e.g., a normalized data format) that could be stored on a server and made available to the test subject via a computer network.

[0150] Spatial mapping can include mappings of spatial patterns of eye damage in users (e.g., test subjects) as described herein. Spatial mapping can be generated by an adaptive map vision inspection algorithm that uses anatomical models of retinal nerve trajectories (e.g., Janssens diagrams) and glaucoma pathophysiology, self-adjusting based on the patient's responses during the process.

[0151] Analysis of the test subject's response includes displaying one or more stimuli in the test subject's impaired visual field, determined based on a spatial mapping. This analysis may include confirming the spatial mapping based on the user's test response. For example, an initial spatial mapping may be generated for the test subject using the algorithms described herein or other methods. Stimuli may then be displayed in visual fields predicted by the analysis based on the initial spatial mapping to be impaired, blurred, or otherwise invisible to the test subject. The user may provide a test response (or fail to provide such a response). If the user fails to respond, the initial spatial mapping or its associated metadata may be updated to create an enhanced spatial mapping that confirms the area of ​​the user's impaired eye associated with that visual field. However, if the user responds successfully, the initial spatial mapping or its associated metadata may be updated to create an enhanced spatial mapping that confirms the area of ​​the user's unimpaired eye associated with that visual field. If the user responds or does not respond as expected (based on the initial spatial mapping and the resulting expected user visual field), the initial spatial mapping (or a further updated version or metadata thereof) may be updated to reflect whether the user's eye is impaired or not (as appropriate).

[0152] At box 980, method 900 includes providing feedback to the test object based on response and analysis.

[0153] Asymmetry in the visual shape of hills Individual differences in the shape of the visual hill within the body can lead to varying degrees of asymmetry, particularly nasal asymmetry between individuals. By calculating the differences in slope between the upper and lower sides, and the mean normal values, as well as the differences in slope between the nasal and temporal sides, and the mean normal values, the upper and lower, and nasal and temporal asymmetries of average normal eye sensitivity at 75 years of age were assessed. The results showed that these asymmetries were subtle and insignificant. Linear regression of the slope change in the upper-lower difference as a function of eccentricity showed a change of -0.006 (dB / decade) / degree or -0.1 dB / decade from the macula to the periphery. The slope difference between the temporal and nasal sides was similar (-0.09 dB / decade from the macula to the periphery). Both upper / lower (SI) and nasal / temporal (NT) asymmetries of mean sensitivity were statistically significant, with slope differences of -0.12 dB, or approximately -2 dB, for both upper / lower and temporal / nasal sides.

[0154] The presence of this asymmetry makes it difficult to accurately model and analyze the stimulus response of test subjects. To address this issue, the asymmetry can be modeled and considered during testing. The shape of the visual hill is typically not linear, and sensitivity decreases on average with eccentricity across all locations. This simple fit reasonably describes the average eccentricity effect, with a maximum difference of less than 0.4 dB between the average eccentricity and the fitted eccentricity. However, due to the asymmetry, the fit is not ideal, with a difference as high as 2.1 dB. The average eccentricity model obtained as a function of eccentricity for the average sensitivity s is: s = 33.5 - 0.35 e + 0.013 e 2 -0.0039 e 3 .

[0155] More complex models can consider each eye quadrant separately. Specifically, to account for differences between quadrants, residuals (sensitivity at each eccentricity minus the cubic fit value) are calculated, and a quadratic function is fitted to these residuals. The coefficient values ​​for each quadrant are then added to the average fit to obtain a unique fit for each quadrant. Quadrant 1 (x>0, y>0): s=34.1-0.59e+0.020e 2 -0.00039 e 3 ; Quadrant 2 (x<0, y>0): s=34.2-0.46e+0.014e 2 -0.00039 e 3 ; Quadrant 3 (x<0, y<0): s=33.1-0.18e+0.007e 2 -0.00039 e 3 ; Quadrant 4 (x>0, y<0): s=33.7-0.40e+0.016e 2 -0.00039 e 3 .

[0156] However, this approach still has problems. Specifically, the mean normal foveal sensitivity predicted by the upper and lower modeling differs by up to 1 dB, and quadrant models need to be stitched together to obtain a single visual field model. Another issue is that, although considering only the age slope seems to be fine, the mean normal sensitivity fit varies greatly between 50 and 100 years of age, so the age factor must be taken into account.

[0157] Another possible approach to modeling asymmetry is to use polar coordinates from -180° to 180° and capture the asymmetry after setting an age. The model also includes age and the effect of age and eccentricity. The age-corrected average sensitivity model for a subject aged a years at a location with eccentricity r and angle θ (in radians) is as follows: .

[0158] There is no systematic or automatic method for determining this fit. Given that sensitivity decreases with age and eccentricity, and that the age effect becomes more pronounced with increasing eccentricity, age and eccentricity are obvious choices. A cubic polynomial is chosen for the eccentricity because the visual hill is known to have two inflection points. Trigonometric functions (e.g., Zernike polynomials) are included to account for the asymmetry of the visual hill.

[0159] However, the model has a problem: the intercept is too large (e.g., 38.3 dB).

[0160] Instead, an improved Visual Field Analysis (VFA) system is needed that can model the visual hill of vision (HoV) and asymmetry step by step (e.g., first fitting the age effect, then fitting the overall (symmetric) shape of the visual hill, and finally fitting the asymmetry using Zernike polynomials).

[0161] Visual Field Analysis (VFA) System Now for reference Figure 10A A field-of-view analysis system 1000 for generating improved personalized visual hill models is illustrated. The field-of-view analysis system 1000 includes a computing system 1002 and a head-mounted display device 1004. The computing system 1002 includes at least one processor 1005 and a memory 1006, which stores an average visual hill (HoV) model and instructions executable by the at least one processor 1005 to perform the various methods described herein. The head-mounted display device 1004 may include one or more of the various display systems described herein, such as combined with… Figure 6 and Figure 7 The device display described.

[0162] In operation, the computational system 1002 generates an eye difference estimate for the test subjects relative to reference data of average healthy eyes obtained from a standard dataset. The eye difference estimate indicates the difference in overall sensitivity or overall height (GH) relative to the reference data, as well as the rate at which sensitivity decays with respect to eccentricity (distance from the fovea of ​​the retina). The computational system 1002 generates a personalized visual HoV model based on the eye difference estimate and an average HoV model 1008. As described in more detail below, the personalized HoV model integrates personalized GH and eccentricity parameters into the average HoV model 1008. Specifically, the definition of the individual eccentricity component allows its slope to effectively capture the difference from the age-corrected average normal visual field. This definition makes a negative GH represent below-average normal overall sensitivity, and a positive GH represent above-average normal overall sensitivity. Furthermore, a negative eccentricity slope represents a steeper HoV than average normal, and a positive eccentricity slope represents a gentler HoV than average normal.

[0163] The computing system 1002 initiates the display of stimuli on the head-mounted display device 1004. Specifically, the head-mounted display device 1004 displays corresponding stimuli at multiple test locations. The computing system 1002 receives the responses from the test subject wearing the head-mounted display device 1004 to each corresponding stimulus and stores the responses in the memory 1006.

[0164] The computing system 1002 analyzes the response to determine the corresponding sensitivity value of the test subject at each of the multiple test locations, and determines the total deviation value of the test subject by subtracting the corresponding sensitivity value of each of the multiple test locations from the corresponding value of the personalized HoV model. The computing system 1002 can also analyze the total deviation value and provide feedback to the test subject based on the analysis.

[0165] The personalized HoV model generated by the computing system 1002 can also be used to readjust the reference field of view used on the head-mounted display device 1004. For example, the derivation of the overthreshold values ​​described herein can initially be based on the average HoV, and as testing progresses using the VFA system 1000, these values ​​can be replaced with the best estimate of the personalized HoV. The personalized HoV can also be used to assess whether certain test locations should be retested with different overthreshold values ​​after estimating the reference field of view of the eye to confirm damage (e.g., if the entire visual hill is found to be very low, this could be the result of cataracts, etc.). The personalized HoV can also be used to infer confidence intervals for individual GH and eccentricity effects, as well as personalized standard values ​​from these individual GH. The personalized HoV model can also serve as a reference for spatial statistical analysis of test results. The personalized HoV model can also be used to calculate standard references that are mathematically more reasonable, detecting anomalous test locations by calculating quantile surfaces. These standard references can be applied to any custom grid, as long as the location in the grid is within a 30° range of the center of the field of view.

[0166] The computing system 1002 can employ different processes to generate the aforementioned eye difference estimate. For example, the computing system 1002 can first determine a set of preliminary test locations and display a visual psychophysical algorithm (e.g., stepwise method, full thresholding method, ZEST, etc.) at these preliminary test locations on the head-mounted display device 1004 to obtain preliminary sensitivity. Then, the computing system 1002 can determine the overall sensitivity or GH of the eye difference estimate and the rate at which the sensitivity decays with eccentricity based on the preliminary sensitivity.

[0167] In other embodiments, the computing system 1002 may generate an initial estimate of the overall sensitivity or GH and the rate at which sensitivity decays with eccentricity, and obtain the temporary response of the test subject to the corresponding stimulus displayed at multiple test locations. The computing system 1002 may then analyze the temporary response to adjust the initial estimate of the overall sensitivity or GH and the rate at which sensitivity decays with eccentricity, obtain confidence intervals, and adjust the level of the corresponding stimulus displayed on the head-mounted display device based on the adjusted initial estimate to obtain a corresponding sensitivity value for the test subject at each of the multiple test locations. The computing system 1002 may then determine the overall sensitivity or GH and the rate at which sensitivity decays with eccentricity based on the corresponding sensitivity value for the test subject at each of the multiple test locations, representing an eye-difference estimate. The initial estimate may include an age-corrected mean normal value, a value generated from analysis of one or more previous visual field tests, or a value from a preliminary test performed on the test subject.

[0168] In some embodiments, the test subject's response to corresponding stimuli displayed at multiple test locations on the head-mounted display device 1004 is a visual field. In these embodiments, the computing system 1002 can generate an eye difference estimate by identifying damaged points in the visual field, removing damaged points from the visual field, analyzing the remaining points in the visual field to generate an eye difference estimate, and fitting a personalized HoV model.

[0169] Now for reference Figure 10B The image shows a graphical representation of an example of the average HoV model 1008, next to a graphical representation of the age-corrected average normal vision model 1010. Figure 10B As shown, the root mean square error (RMSE) of the average HoV model 1008 is lower than that of the age-corrected average normal vision model 1010, and the largest difference between the average HoV model 1008 and the age-corrected average normal vision model 1010 occurs at the edge or periphery.

[0170] In some embodiments, the average HoV model 1008 may include three different combination elements or models. Specifically, these combination elements or models may include an intercept / age model 1012, an eccentricity or radial mean shape model 1014, and a visual field asymmetry model 1016, such as... Figure 10C As shown. Typically, the intercept / age model 1012 describes the theoretical sensitivity of a 0-year-old patient at the visual center (fovea of ​​the retina) and records age-related eccentricity differences, which are functions of the subject's age and distance from the fovea (eccentricity). Eccentricity models typically record a linear decay of sensitivity with eccentricity, while visual field asymmetry models use Zernike polynomials to record asymmetries between the superior and inferior, and nasal and temporal sides of the visual field.

[0171] To find the intercept and age effect of the average HoV model 1008, a quadratic function was fitted to a first set of visual field standard data (e.g., 10⁻² visual field standard data) and independently fitted to a second set of visual field standard data (e.g., 24⁻² visual field standard data). For the second set of normal data, a cross component was included to account for the known fact that the age effect increases with eccentricity. Finally, a linear component of the average (symmetric) shape of the HoV was added. A linear model was used because the radial eccentricity effect can be well described by a linear model. For the fitting of the second set of normal data, positions corresponding to the blind spot, the four outermost upper positions (most affected by eyelid artifacts), and the two outermost nasal positions (most affected by marginal lens artifacts) were removed.

[0172] In some embodiments, the average HoV model 1008 is constructed by fitting the intercept, age, eccentricity, and age-eccentricity effect, and then a visual field asymmetry model 1016 is generated using Zernike coefficients for all locations except two in the blind spot (including four eyelid locations and two outermost nasal locations). The age and age-eccentricity coefficients given by this method are as follows: Age coefficient = -0.0402 dB / year Age - Eccentricity coefficient = -0.0013 dB / (year × degree) In this age model, the age slope at the fovea of ​​the retina is -0.04 dB / year, the slope at the location (±10°, ±10°) is -0.06 dB / year, and the slope at an eccentricity of 30° is -0.08 dB / year.

[0173] Figure 10DExample polynomials and the final surface used to describe the asymmetry are shown, as a field-of-view asymmetry model 1016. Specifically, the polynomials used to construct the field-of-view asymmetry model 1016 include representations of vertical tilt 1018, horizontal tilt 1020, oblique astigmatism 1022, horizontal astigmatism 1024, vertical trefoil 1026, horizontal comma 1028, and vertical or oblique quadrefoil 1030. The Zernike coefficients for each of these polynomials used to model the asymmetry are as follows: Vertical tilt: −0.958 dB / degree; Horizontal tilt: 0.517 dB / degree; Oblique astigmatism: −0.221 dB / degree 2 ; Horizontal astigmatism: 0.650 dB / degree 2 ; Vertical trefoil: 0.185 dB / degree 3 ; Horizontal comma shape: 0.541 dB / degree 3 ; Oblique quadrilateral: −0.452 dB / degree 4 .

[0174] Finally, by adjusting for age and asymmetry and fitting a simple linear model as a function of the radius of the fovea, an intercept of 36.1 dB and an eccentricity parameter of -0.12 dB / degree were derived from the original second set of visual field standard data (e.g., 24-2 visual field standard data). A similar intercept of 36.2 dB was derived from the first set of visual field standard data (e.g., 10-2 visual field standard data).

[0175] Refer again Figure 10C The figure shows that the model predicts a decrease of more than 3 dB in the radial reduction at the outermost positions on the north and west sides of the fovea. It should be understood that the model can be simplified to a fewer number of polynomials using methods different from the Zernike polynomial.

[0176] Now for reference Figure 10E A graphical representation of an example of a personalized HoV model 1034 is shown, next to a graphical representation of an age-corrected average normal vision model 1010. As described above, the personalized HoV model 1034 is generated by a computing system 1002 ( Figure 10A Based on the average HoV model 1008 ( Figure 10B This is generated for a specific test subject wearing the head-mounted display device 1004. For example... Figure 10EAs shown, by recalculating the intercept and eccentricity parameters, the personalized HoV model 1034 reduces the root mean square error from 0.42 dB for the average HoV model 1008 to 0.28 dB for the personalized visual hill model 1034.

[0177] Specifically, the intercept and eccentricity effect parameters were estimated based on the established average HoV model 1008 and the age effect, and were recalculated individually for each test subject. For example, the age effect and asymmetry of all visual fields in the standard dataset were removed. Then, for each visual field, the intercept and eccentricity parameters were obtained. Figure 10E In the example personalized HoV model 1034 shown, the weighted average intercept is 36.5 dB, and the weighted average eccentricity parameter is -0.159 dB / degree. This intercept is 0.4 dB larger than that of the average HoV model 1008, but -0.04 dB / degree lower than the average eccentricity parameter.

[0178] When fitted to individual visual fields, the mean RMSEs for the age-corrected mean normal model 1010 and the mean HoV model 1008 are 2.1 dB and 2.0 dB, respectively. For the age-corrected mean normal model, the RMSE is equivalent to the root mean square TD value. Similarly, the RMS for the PD value is 2.5 dB. The personalized approach of the personalized HoV model 1034 has the same objective as the conventional model PD value, but utilizes reference deviations (RD) that take into account the eccentricity effect. RD is obtained after adjusting each visual field to the individual HoV height and eccentricity effect (strongly assuming that all individuals have the same asymmetry). This approach produces a mean RMS PD of 1.6 dB, a reduction of 0.9 dB or 36% relative to the model deviation value. Example personalized HoV model 1034 adjusts for age, asymmetry, and mean intercept and eccentricity effect, and then uses a linear model to obtain the intercept and eccentricity effect for each individual visual field. In this way, the estimate of GH (e.g., the overall sensitivity difference from mean normal to individual visual field) is used as the intercept for fitting.

[0179] The eccentricity effect estimate recorded by the model is defined as the slope of the fit (e.g., the difference in visual hill steepness from mean normal to individual visual field). Furthermore, the advantage of the personalized HoV model 1034 over alternative methods can be evaluated by estimating the inter-individual differences in GH and eccentricity effect and comparing the estimates to the conventional model by subtracting the 7th most sensitive TD for the visual field. The central 95% distribution of GH in the example personalized HoV 1034 is within ±2.3 dB, and the central 95% distribution of the eccentricity effect is within ±0.11 dB / degree, meaning that the difference at 30° is within ±3.3 dB solely due to the eccentricity effect. The average difference is 1.8 dB compared to the conventional estimate (using the 7th most sensitive TD), and the difference between the GH estimate and GH7 varies significantly with changes in the estimated eccentricity effect, where a flatter visual field produces a smaller GH7 estimate, while a steeper visual field produces a larger GH7 estimate compared to the example personalized HoV 1034 GH estimate.

[0180] Invention aspect The following aspects are provided as examples based on the disclosure herein and are not intended to limit the scope of this disclosure. Furthermore, aspects 1-9 relate to systems and methods for glaucoma diagnosis and monitoring by implementing adaptive map visual field testing; aspects 10-30 relate to systems and methods for automatically evaluating visual field tests; aspects 31-44 relate to systems and methods for diagnosing and monitoring glaucoma by implementing adaptive map visual field testing; and aspects 45-64 relate to visual field analysis systems and methods for automatically evaluating visual field tests.

[0181] Aspect 1. A visual field (VF) analysis system configured for diagnosing and monitoring glaucoma by implementing adaptive map visual field testing, the VF analysis system comprising: a head-mounted device (e.g., a virtual reality (VR) device) including a display screen positioned near or within visual range of a user's eyes, the head-mounted device being communicatively coupled to one or more processors; and an adaptive map visual field testing algorithm including computational instructions stored in memory accessible to the one or more processors; a standard database or model, wherein the adaptive map visual field testing algorithm is configured to access or implement the standard database or model to make the head-mounted device device independent of one or more head-mounted devices with different configurations; wherein the computational instructions of the adaptive map visual field testing algorithm, when executed by the one or more processors, are configured to cause the one or more processors to implement VF testing. The VF test includes: performing a VF test on the display screen of the head-mounted device, wherein the VF test is adapted to the area of ​​the display screen based on values ​​in the standard database or model, wherein the VF test is presented as a visualization identical or similar to that of one or more different head-mounted devices, the one or more different head-mounted devices having different displays with different corresponding shapes, formats, sizes and / or resolutions; receiving visual test data indicating the user's field of vision; detecting one or more user-specific initial test locations based on the visual test data, the one or more initial test locations defining one or more healthy clusters indicating no scatoma and one or more damaged clusters indicating dark spots; and generating a spatial map identifying the locations of the one or more damaged clusters based on the one or more initial test locations.

[0182] Aspect 2, the VF analysis system according to aspect 1, wherein the spatial mapping positions are spaced 0.5 degrees or 0.5 resolution apart in the vertical and / or horizontal angles.

[0183] Aspect 3. The VF analysis system according to Aspect 1 or 2, wherein the standard database or model includes standard values ​​generated by quantile regression, wherein the quantile regression includes generating standard values ​​from standard reference values, wherein the standard reference values ​​include biometric values, such as refractive power, axial length, corneal curvature, or other independent variables including user eye biometric values, wherein the head-mounted device is updated with or has access to the standard database or model to calibrate the head-mounted device to be device-independent when implementing the adaptive map vision inspection algorithm.

[0184] Aspect 4. A visual field (VF) analysis method for diagnosing and monitoring glaucoma by implementing an adaptive map visual field test, the visual field analysis method comprising: performing a VF test on a display screen of an electronic display device (e.g., a virtual reality (VR) device), wherein the electronic display device includes a display screen positioned near or within a user's eye or visual distance, the electronic display device being communicatively coupled to one or more processors, and wherein the visual field test is adapted to a region of the display screen based on values ​​in a standard database or model, wherein the visual field test is presented as a visualization identical or similar to that of one or more different electronic display devices having different displays screens having different corresponding shapes, formats, sizes, and / or resolutions, wherein an adaptive map visual field test algorithm is configured to access or implement the standard database or model to make the electronic display device device independent of the one or more different configured electronic display devices; receiving visual test data indicating the user's visual field; detecting one or more user-specific initial test locations based on the visual test data, the one or more initial test locations defining one or more healthy clusters indicating no scatoma and one or more damaged clusters indicating scotomas; and generating a spatial map identifying the locations of the one or more damaged clusters based on the one or more initial test locations.

[0185] Aspect 5, according to the VF analysis method of aspect 4, wherein the spatial mapping positions are spaced 0.5 degrees or 0.5 resolutions apart in the vertical and / or horizontal angles.

[0186] Aspect 6. The VF analysis method according to Aspect 4 or 5, wherein the standard database or model includes standard values ​​generated by quantile regression, wherein the quantile regression includes generating standard values ​​from standard reference values, wherein the standard reference values ​​include biometric values, such as refractive power, axial length, corneal curvature, or other independent variables including user eye biometric values, wherein the electronic display device is updated with or has access to the standard database or model to calibrate the electronic display device to be device-independent when implementing the adaptive map field of view inspection algorithm.

[0187] Aspect 7. A tangible, non-transitory computer-readable medium storing instructions for diagnosing and monitoring glaucoma by implementing adaptive map vision tests, the instructions, when executed by one or more processors, causing the one or more processors to: implement a VF test on a display screen of an electronic display device (e.g., a virtual reality (VR) device), wherein the electronic display device includes a display screen positioned near or within the user's eye's viewing distance, the electronic display device being communicatively coupled to the one or more processors, and wherein the vision test is adapted to a region of the display screen based on values ​​in a standard database or model, wherein the vision test is presented as a visualization identical or similar to that of one or more different electronic display devices, the one or more different electronic display devices... The sub-display device has different displays with different shapes, formats, sizes, and / or resolutions, wherein the adaptive map view inspection algorithm is configured to access or implement the standard database or model to make the electronic display device device device-independent relative to the one or more differently configured electronic display devices; receive visual test data indicating the user's view; detect one or more user-specific initial test locations based on the visual test data, the one or more initial test locations defining one or more healthy clusters indicating no scatoma and one or more damaged clusters indicating dark spots; and generate a spatial map identifying the locations of the one or more damaged clusters based on the one or more initial test locations.

[0188] Aspect 8. The tangible, non-transitory computer-readable medium according to aspect 7, wherein the spatially mapped positions are spaced 0.5 degrees or 0.5 resolutions apart in the vertical and / or horizontal angles.

[0189] Aspect 9. A tangible, non-transitory computer-readable medium according to aspect 7 or 8, wherein the standard database or model includes standard values ​​generated by quantile regression, wherein the quantile regression includes generating standard values ​​from standard reference values, the standard reference values ​​including biometric values, such as refractive power, axial length, corneal curvature, or other independent variables including user eye biometric values, wherein the electronic display device is updated with or has access to the standard database or model to calibrate the electronic display device to be device-independent when implementing the adaptive map field of view inspection algorithm.

[0190] Aspect 10. A visual feedback (VFF) system configured for automatically evaluating visual field tests, the VFF system comprising: a head-mounted device (e.g., a virtual reality (VR) device) including a display screen positioned near or within a user's eyes, the head-mounted device being communicatively coupled to one or more processors; and computational instructions stored in a computer memory, the computational instructions, when executed by the one or more processors, causing the one or more processors to: display a stimulus to a test subject (e.g., a patient); store the test subject's response to the displayed stimulus in the computer memory; analyze the response based on the location of the displayed stimulus; and provide feedback to the test subject based on the response and analysis.

[0191] Aspect 11. The VFF system according to aspect 10, wherein the response of the test subject may include identifying and acknowledging the stimulus by interacting with the controller, or not identifying or acknowledging the stimulus by not interacting with the controller.

[0192] Aspect 12. The VFF system according to aspect 10 or 11, wherein the response of the test subject may include (a) whether the response of the test subject (e.g., a patient) is successful or unsuccessful; (b) the response time of the test subject (e.g., a patient); (c) the location of the stimulus; (d) the gaze location of the test subject (e.g., a patient); (e) the eye movement of the test subject; and / or (f) the response time (e.g., the response time of the visual feedback system).

[0193] Aspect 13. The VFF system according to any one of Aspects 10-12, wherein the feedback may include any one of visual feedback, audio feedback, haptic feedback, and somatosensory feedback.

[0194] Aspect 14. The VFF system according to any one of Aspects 10-13, wherein the feedback is related to the accuracy of the test object's response.

[0195] Aspect 15. The VFF system according to any one of Aspects 10-14, wherein the feedback may be positive or negative.

[0196] Aspect 16. The VFF system according to any one of Aspects 10-15, wherein positive feedback is displayed if the test object responds correctly, and negative feedback is displayed if the test object responds incorrectly.

[0197] Aspect 17. The VFF system according to any one of Aspects 10-16, wherein the system is used for any one of visual field testing, rehabilitation after disease or injury, diagnostic and / or therapeutic purposes, visual performance enhancement or visual function training.

[0198] Aspect 18. The VFF system according to any one of Aspects 10-17, wherein the feedback is formatted in a format that can be stored on a server and provided to the test subject via a computer network.

[0199] Aspect 19. The VFF system according to any one of Aspects 10-18, wherein analyzing the response based on the location of the displayed stimulus comprises: inputting the response into an AI model executed on the one or more processors, wherein the AI ​​model is trained using training data of response feedback from a test subject (e.g., button press, eye movement, or other such feedback provided by the user), wherein the AI ​​model is configured to output a prediction, the prediction defining an accuracy value corresponding to the percentage of the test subject that correctly responds to the displayed stimulus; and generating feedback for the test subject by the AI ​​model based on the prediction defining the accuracy value.

[0200] Aspect 20. A visual feedback (VFF) method for automatically assessing visual field tests, the method comprising: displaying a stimulus to a test subject (e.g., a patient) on a head-mounted device (e.g., a virtual reality (VR) device); recording the test subject's response to the displayed stimulus via one or more processors; analyzing the test subject's response to the displayed stimulus using the stimulus location via a visual feedback system; and providing feedback to the test subject based on the response and analysis.

[0201] Aspect 21. The VFF method according to aspect 20, wherein the response of the test subject may include identifying and acknowledging the stimulus by interacting with the controller, or not identifying or acknowledging the stimulus by not interacting with the controller.

[0202] Aspect 22. The VFF method according to aspect 20 or 21, wherein the response of the test subject may include (a) whether the response of the test subject (e.g., a patient) is successful or unsuccessful; (b) the response time of the test subject (e.g., a patient); (c) the location of the stimulus; (d) the gaze location of the test subject (e.g., a patient); (e) the eye movement of the test subject; and (f) the response time (e.g., the response time of the visual feedback system).

[0203] Aspect 23. The VFF method according to any one of Aspects 20-22, wherein the feedback may include any one of visual feedback, audio feedback, haptic feedback, and somatosensory feedback.

[0204] Aspect 24. The VFF method according to any one of Aspects 20-23, wherein the feedback is related to the accuracy of the test object's response.

[0205] Aspect 25. The VFF method according to any one of Aspects 20-24, wherein the feedback may be positive or negative.

[0206] Aspect 26. The VFF method according to any one of Aspects 20-25, wherein positive feedback is displayed if the test object responds correctly, and negative feedback is displayed if the test object responds incorrectly.

[0207] Aspect 27. The VFF method according to any one of Aspects 20-26, wherein the method is used for any one of visual field testing, rehabilitation after disease or injury, diagnostic and / or therapeutic purposes, visual performance enhancement or visual function training.

[0208] Aspect 28. The VFF method according to any one of Aspects 20-27, wherein the feedback is formatted in a format that can be stored on a server and provided to the test subject via a computer network.

[0209] Aspect 29. The VFF method according to any one of Aspects 20-28, wherein analyzing the response based on the location of the displayed stimulus comprises: inputting the response into an AI model executed on the one or more processors, wherein the AI ​​model is trained using training data of response feedback from a test subject (e.g., button press, eye movement, or other such feedback provided by the user), wherein the AI ​​model is configured to output a prediction, the prediction defining an accuracy value corresponding to the percentage of the test subject that correctly responds to the displayed stimulus; and generating feedback for the test subject by the AI ​​model based on the prediction defining the accuracy value.

[0210] Aspect 30. A tangible, non-transitory computer-readable medium storing instructions for automatically assessing visual field tests, the instructions, when executed by one or more processors, causing the one or more processors to: display stimuli to a test subject (e.g., a patient) on a head-mounted device (e.g., a virtual reality (VR) device); record the test subject's response to the displayed stimuli via the one or more processors; analyze the test subject's response to the displayed stimuli using the location of the stimulus via a visual field feedback system; and provide feedback to the test subject based on the response and analysis.

[0211] Aspect 31. A visual field (VF) analysis system configured to diagnose and monitor glaucoma by implementing an adaptive map visual field test, the visual field analysis system comprising: a head-mounted device (e.g., a virtual reality (VR) device) including a display screen positioned near or within a user's eye range, the head-mounted device being communicatively coupled to one or more processors; and an adaptive map visual field test algorithm including computational instructions stored in memory accessible to the one or more processors; a standard database or model, wherein the adaptive map visual field test algorithm is configured to access or implement the standard database or model to make the head-mounted device device independent of one or more head-mounted devices with different configurations; wherein the computational instructions of the adaptive map visual field test algorithm, when executed by the one or more processors, are configured to cause the one or more processors to perform a VF test, the VF test comprising: performing a VF test on the display screen of the head-mounted device, wherein the VF test is based on the... Values ​​from a standard database or model are adapted to the area of ​​the display screen, wherein the VF test is presented as a visualization identical or similar to one or more different head-mounted devices having different displays with different corresponding shapes, formats, sizes, and / or resolutions; receiving visual test data indicating the user's field of vision; detecting one or more user-specific initial test locations based on the visual test data, the one or more initial test locations defining one or more healthy clusters indicating no scatoma and one or more damaged clusters indicating scotomas; generating a spatial map identifying the locations of the one or more damaged clusters based on the one or more initial test locations; displaying stimuli to a test subject (e.g., a patient); storing the test subject's response to the displayed stimuli in computer memory; analyzing the response based on the location of the displayed stimuli and / or the spatial map; and providing feedback to the test subject based on the response and analysis.

[0212] Aspect 32, the VF analysis system according to aspect 31, wherein the spatial mapping positions are spaced 0.5 degrees or 0.5 resolution apart in the vertical and / or horizontal angles.

[0213] Aspect 33. The VF analysis system according to aspect 31 or 32, wherein the standard database or model includes standard values ​​generated by quantile regression, wherein the quantile regression includes generating standard values ​​from standard reference values, the standard reference values ​​including biometric values, such as refractive power, axial length, corneal curvature, or other independent variables including user eye biometric values, wherein the head-mounted device is updated with or has access to the standard database or model to calibrate the head-mounted device to device independence when implementing the adaptive map vision inspection algorithm.

[0214] Aspect 34. The VF analysis system according to any one of Aspects 31-33, wherein the response of the test subject may include identifying and acknowledging the stimulus by interacting with the controller, or not identifying or acknowledging the stimulus by not interacting with the controller.

[0215] Aspect 35. The VF analysis system according to any one of Aspects 31-34, wherein the response of the test subject may include (a) whether the response of the test subject (e.g., a patient) is successful or unsuccessful; (b) the response time of the test subject (e.g., a patient); (c) the location of the stimulus; (d) the fixation location of the test subject (e.g., a patient); (e.g., the eye movement of the test subject; and / or (f) the response time (e.g., the response time of the visual field feedback system).

[0216] Aspect 36. The VF analysis system according to any one of Aspects 31-35, wherein the feedback may include any one of visual feedback, audio feedback, tactile feedback, and somatosensory feedback.

[0217] Aspect 37. The VF analysis system according to any one of Aspects 31-36, wherein the feedback is related to the accuracy of the test object's response.

[0218] Aspect 38. The VF analysis system according to any one of Aspects 31-37, wherein the feedback may be positive or negative.

[0219] Aspect 39. The VF analysis system according to any one of Aspects 31-38, wherein positive feedback is displayed if the test object responds correctly, and negative feedback is displayed if the test object responds incorrectly.

[0220] Aspect 40. The VF analysis system according to any one of Aspects 31-39, wherein the system is used for any one of visual field testing, rehabilitation after disease or injury, diagnostic and / or therapeutic purposes, visual performance enhancement or visual function training.

[0221] Aspect 41. The VF analysis system according to any one of Aspects 31-40, wherein the feedback is formatted in a format that can be stored on a server and provided to the test subject via a computer network.

[0222] Aspect 42. The VF analysis system according to any one of Aspects 31-41, wherein analyzing the response based on the location of the displayed stimulus comprises: inputting the response into an AI model executed on the one or more processors, wherein the AI ​​model is trained using training data of response feedback from a plurality of test subjects (e.g., button presses, eye movements, or other such feedback provided by the user) and / or spatial mappings of a plurality of test subjects, wherein the AI ​​model is configured to output a prediction, the prediction defining an accuracy value corresponding to the percentage of the percentage of the test subject that correctly responds to the displayed stimulus based on the spatial mapping of the test subject; and generating feedback for the test subject by the AI ​​model based on the prediction defining the accuracy value.

[0223] Aspect 43. A visual field (VF) analysis method for diagnosing and monitoring glaucoma by implementing an adaptive map visual field test, the VF analysis method comprising: performing a visual field test on a display screen of an electronic display device (e.g., a virtual reality (VR) device), wherein the electronic display device includes a display screen positioned near or within a user's eye or visual distance, the electronic display device being communicatively coupled to one or more processors, and wherein the VF test is adapted to a region of the display screen based on values ​​in a standard database or model, wherein the VF test is presented as the same or similar visualization to one or more different electronic display devices having different displays screens having different corresponding shapes, formats, sizes and / or resolutions, wherein an adaptive map visual field test algorithm is configured to access or implement the standard database or model to enable the electronic display screen to perform a visual field test on a region of the display screen. The display device is device-independent relative to the one or more differently configured electronic display devices; receives visual test data indicating a user's field of vision; detects one or more user-specific initial test locations based on the visual test data, the one or more initial test locations defining one or more healthy clusters indicating no scatoma and one or more damaged clusters indicating scotomas; generates a spatial map based on the one or more initial test locations to identify the locations of the one or more damaged clusters; displays stimuli to a test subject (e.g., a patient) on a head-mounted device (e.g., a virtual reality (VR) device); records the test subject's response to the displayed stimuli via one or more processors; analyzes the test subject's response to the displayed stimuli using the stimulus locations and / or the spatial map via a visual feedback system (e.g., the VR device); and provides feedback to the test subject based on the response and analysis.

[0224] Aspect 44. A tangible, non-transitory computer-readable medium storing instructions for diagnosing and monitoring glaucoma by implementing adaptive map view inspection, the instructions, when executed by one or more processors, causing the one or more processors to: implement a VF test on a display screen of an electronic display device (e.g., a virtual reality (VR) device), wherein the electronic display device includes a display screen positioned near or within the user's eye or viewing distance, the electronic display device being communicatively coupled to the one or more processors, and wherein the VF test is adapted to a region of the display screen based on values ​​in a standard database or model, wherein the VF test is presented as a visualization identical or similar to that of one or more different electronic display devices having different displays having different corresponding shapes, formats, sizes, and / or resolutions, wherein the adaptive map view inspection algorithm is configured to access or implement the standard... A quasi-database or model is used to make the electronic display device device device-independent relative to the one or more electronic display devices with different configurations; visual test data indicating a user's field of vision is received; one or more user-specific initial test locations are detected based on the visual test data, the one or more initial test locations defining one or more healthy clusters indicating no scatoma and one or more damaged clusters indicating scotomas; a spatial map is generated based on the one or more initial test locations to identify the locations of the one or more damaged clusters; stimuli are displayed to a test subject (e.g., a patient) on a head-mounted device (e.g., a virtual reality (VR) device); the test subject's response to the displayed stimuli is recorded by one or more processors; the test subject's response to the displayed stimuli is analyzed using the stimulus locations and / or the spatial map via a visual feedback system (e.g., the VR device); and feedback is provided to the test subject based on the response and analysis.

[0225] Aspect 45. A visual field analysis (VFA) system configured for automatically evaluating visual field tests, the VFA system comprising: a head-mounted device (e.g., a virtual reality (VR) device) including a display screen positioned near or within visual range of a user's eyes, the head-mounted device being communicatively coupled to one or more processors; an average visual hill model (HoV) stored in a computer memory; and computational instructions stored in the computer memory, the computational instructions, when executed by the one or more processors, causing the one or more processors to: generate an eye difference estimate of a test subject relative to reference data of an average healthy eye obtained from a standard dataset, the eye difference estimate indicating (1) an overall sensitivity or overall height (GH) relative to the reference data. And (2) the difference in the rate at which sensitivity decays with eccentricity (distance from the fovea of ​​the retina); generating a personalized HoV model based on the eye difference estimate and the average HoV model; displaying corresponding stimuli to the test subject at multiple test locations on the head-mounted device; storing the test subject's response to the corresponding stimuli displayed at the multiple test locations in a computer memory; analyzing the response to determine the corresponding sensitivity value of the test subject at each of the multiple test locations; determining the total deviation value of the test subject by subtracting the corresponding sensitivity value at each of the multiple test locations from the corresponding value of the personalized HoV model; analyzing the total deviation value; and providing feedback to the test subject based on the analysis.

[0226] Aspect 46. The VFA system according to aspect 45, wherein, in order to generate the eye difference estimate, the computation instructions, when executed by the one or more processors, cause the one or more processors to: determine a set of preliminary test locations; display a visual psychophysical algorithm at the set of preliminary test locations on the head-mounted device to obtain preliminary sensitivity; and determine, based on the preliminary sensitivity, the overall sensitivity or GH of the eye difference estimate and the rate at which the sensitivity decays with eccentricity.

[0227] Aspect 47. The VFA system according to aspect 45 or 46, wherein, in order to generate the eye difference estimate, the computation instructions, when executed by the one or more processors, cause the one or more processors to: generate an initial estimate of the overall sensitivity or GH and the rate at which sensitivity decays with eccentricity; obtain a provisional response of the test subject to a corresponding stimulus displayed at the plurality of test locations; analyze the provisional response to adjust the initial estimate of the overall sensitivity or GH and the rate at which sensitivity decays with eccentricity, and obtain a confidence interval; adjust the level of the corresponding stimulus displayed on the head-mounted device based on the adjusted initial estimate to obtain a corresponding sensitivity value of the test subject at each of the plurality of test locations; and determine the overall sensitivity or GH and the rate at which sensitivity decays with eccentricity of the eye difference estimate based on the corresponding sensitivity value of the test subject at each of the plurality of test locations.

[0228] Aspect 48. The VFA system according to any one of Aspects 45-47, wherein the initial estimate comprises an age-corrected mean normal value, a value generated from analysis of one or more previous visual field tests, or a value from a preliminary test performed on the test subject.

[0229] Aspect 49. The VFA system according to any one of Aspects 45-48, wherein the response of the test subject to a corresponding stimulus displayed at the plurality of test locations on the head-mounted device is a visual field, and in order to generate the eye difference estimate, the computational instructions, when executed by the one or more processors, cause the one or more processors to: identify damaged points in the visual field; remove the damaged points from the visual field; analyze the remaining points in the visual field to generate the eye difference estimate and fit the personalized HoV model.

[0230] Aspect 50. The VFA system according to any one of Aspects 45-49, wherein one of the least squares algorithm, maximum likelihood algorithm, Bayesian algorithm or machine learning algorithm is used to generate the eye difference estimate and fit the personalized HoV model.

[0231] Aspect 51. The VFA system according to any one of Aspects 45-50, wherein the mean HoV model comprises a combination of an intercept / age model, an eccentricity model, and a visual field asymmetry model.

[0232] Aspect 52. The VFA system according to any one of Aspects 45-51, wherein the intercept model describes the theoretical sensitivity of a 0-year-old patient at the visual center (fovea of ​​the retina).

[0233] Aspect 53. The VFA system according to any one of Aspects 45-52, wherein the age model illustrates age-related differences arising from the subject's age and distance from the fovea (eccentricity). Aspect 54. The VFA system according to any one of Aspects 45-53, wherein the eccentricity model recording sensitivity decreases linearly with eccentricity.

[0234] Aspect 55. The VFA system according to any one of Aspects 45-54, wherein the visual field asymmetry model uses Zernike polynomials to record the asymmetry between the upper and lower sides of the visual field and between the nasal and temporal sides.

[0235] Aspect 56. A visual field analysis (VFA) method for automatically evaluating visual field tests, the VFA method comprising: generating an eye difference estimate of a test subject relative to reference data of an average healthy eye obtained from a standard dataset, the eye difference estimate indicating differences in (1) overall sensitivity or overall height (GH) and (2) the rate at which sensitivity decays with eccentricity (distance from the fovea of ​​the retina) relative to the reference data; generating a personalized HoV model based on the eye difference estimate and an average HoV model stored in a computer memory; displaying corresponding stimuli to the test subject at multiple test locations on a head-mounted device (e.g., a virtual reality (VR) device) including a display screen positioned near or within the user's eyes; storing the test subject's responses to the corresponding stimuli displayed at the multiple test locations in a computer memory; analyzing the responses to determine a corresponding sensitivity value of the test subject at each of the multiple test locations; determining a total deviation value of the test subject by subtracting the corresponding sensitivity value at each of the multiple test locations from the corresponding value of the personalized HoV model; analyzing the total deviation value; and providing feedback to the test subject based on the analysis.

[0236] Aspect 57. The VFA method according to aspect 56, wherein generating the eye difference estimate includes: determining a set of preliminary test locations; displaying a visual psychophysical algorithm at the set of preliminary test locations on the head-mounted device to obtain preliminary sensitivity; and determining, based on the preliminary sensitivity, the overall sensitivity or GH of the eye difference estimate and the rate at which the sensitivity decays with eccentricity.

[0237] Aspect 58. The VFA method according to Aspect 56 or 57, wherein generating the eye difference estimate comprises: generating an initial estimate of overall sensitivity or GH and the rate at which sensitivity decreases with eccentricity; obtaining a provisional response of the test subject to a corresponding stimulus displayed at the plurality of test locations; analyzing the provisional response to adjust the initial estimate of overall sensitivity or GH and the rate at which sensitivity decreases with eccentricity, and obtaining a confidence interval; adjusting the level of the corresponding stimulus displayed on the head-mounted device based on the adjusted initial estimate to obtain a corresponding sensitivity value of the test subject at each of the plurality of test locations; and determining the overall sensitivity or GH and the rate at which sensitivity decreases with eccentricity of the eye difference estimate based on the corresponding sensitivity value of the test subject at each of the plurality of test locations.

[0238] Aspect 59. The VFA method according to any one of Aspects 56-58, wherein the initial estimate comprises an age-corrected mean normal value, a value generated from analysis of one or more previous visual field tests, or a value from a preliminary test performed on the test subject.

[0239] Aspect 60. The VFA method according to any one of Aspects 56-59, wherein the response of the test subject includes generating the eye difference estimate by: identifying damaged points in the visual field; removing the damaged points from the visual field; analyzing the remaining points in the visual field to generate the eye difference estimate and fitting the personalized HoV model.

[0240] Aspect 61. The VFA method according to any one of Aspects 56-60, wherein one of the least squares algorithm, maximum likelihood algorithm, Bayesian algorithm or machine learning algorithm is used to generate the eye difference estimate and fit the personalized HoV model.

[0241] Aspect 62. The visual field analysis method according to any one of Aspects 56-61, wherein the mean HoV model comprises a combination of an intercept / age model, an eccentricity model, and a visual field asymmetry model.

[0242] Aspect 63. A tangible, non-transitory computer-readable medium storing instructions for automatically evaluating visual field tests, the instructions, when executed by one or more processors, causing the one or more processors to: generate an eye difference estimate of a test subject relative to reference data of an average healthy eye obtained from a standard dataset, the eye difference estimate indicating differences in (1) overall sensitivity or overall height (GH) and (2) the rate at which sensitivity decays with eccentricity (distance from the fovea) relative to the reference data; generate a personalized visual hill model based on the eye difference estimate and an average HoV model stored in a computer memory; and transmit the results to the test subject on a head-mounted device. The test subject is shown corresponding stimuli at multiple test locations, wherein the head-mounted device (e.g., a virtual reality (VR) device) includes a display screen positioned near or within the user's eye range; the test subject's response to the corresponding stimuli displayed at the multiple test locations is stored in a computer memory; the response is analyzed to determine a corresponding sensitivity value for the test subject at each of the multiple test locations; a total deviation value for the test subject is determined by subtracting the corresponding sensitivity value for each of the multiple test locations from the corresponding value in the personalized HoV model; the total deviation value is analyzed; and feedback is provided to the test subject based on the analysis.

[0243] Aspect 64, the tangible, non-transitory computer-readable medium according to aspect 63, wherein the HoV hill model comprises a combination of an intercept / age model, an eccentricity model, and a field-of-view asymmetry model.

[0244] Other considerations Although the following text describes many different embodiments in detail, it should be understood that the legal scope of this description is defined by the wording of the claims and their equivalents set forth at the end of this patent. The detailed description should be interpreted as exemplary and does not describe every possible embodiment, as it would be impractical to describe every possible embodiment. Many alternative embodiments may be implemented using current technology or technology developed after the date of this patent application, and these alternative embodiments will still fall within the scope of the claims.

[0245] The following additional considerations apply to the above discussion. Throughout the specification, multiple instances may implement components, operations, or structures described as single instances. Although the individual operations of one or more methods are shown and described as separate operations, one or more separate operations may be performed concurrently, and there is no requirement that these operations be performed in the order shown. Structures and functionalities presented as separate components in the example configuration may be implemented as composite structures or components. Similarly, structures and functionalities presented as single components may be implemented as separate components. These and other variations, modifications, additions, and improvements fall within the scope of this document's subject matter.

[0246] Furthermore, some embodiments described herein include logic or multiple routines, subroutines, application programs, or instructions. These can constitute software (e.g., code embodied in a machine-readable medium or transmitted signal) or hardware. In hardware, routines, etc., are tangible units capable of performing specific operations and can be configured or arranged in a particular manner. In example embodiments, one or more computer systems (e.g., standalone, client, or server computer systems) or one or more hardware modules of a computer system (e.g., a processor or a group of processors) can be configured by software (e.g., an application program or an application portion) as hardware modules that operate to perform some of the operations described herein.

[0247] The various operations of the example methods described herein can be performed, at least in part, by one or more processors configured, either temporarily (e.g., by software) or permanently, to perform the relevant operations. Whether temporarily or permanently configured, such processors can constitute processor-implemented modules that operate to perform one or more operations or functions. The modules referred to herein may include processor-implemented modules in some example embodiments.

[0248] Similarly, the methods or routines described herein can be implemented at least in part by a processor. For example, at least some operations of the methods can be performed by one or more processors or hardware modules implemented by processors. The execution of certain operations can be distributed among one or more processors, residing not only within a single machine but also deployed across multiple machines. In some example embodiments, the processor or processors can reside in a single location, while in other embodiments, the processors can be distributed across multiple locations.

[0249] The execution of certain operations can be distributed across one or more processors, residing not only within a single machine but also deployed across multiple machines. In some example embodiments, one or more processors or processor-implemented modules may reside in a single geographic location (e.g., within a home environment, office environment, or server cluster). In other embodiments, one or more processors or processor-implemented modules may be distributed across multiple geographic locations.

[0250] This detailed description should be interpreted as exemplary and does not describe every possible embodiment, as describing every possible embodiment, even if possible, is impractical. Many alternative embodiments can be implemented using current technology or technology developed after the date of this application.

[0251] Those skilled in the art will recognize that various modifications, alterations, and combinations can be made to the above embodiments without departing from the scope of the invention, and such modifications, alterations, and combinations should be considered within the scope of the inventive concept.

Claims

1. A visual field (VF) analysis system configured to diagnose and monitor glaucoma by implementing adaptive map visual field examination, the VF analysis system comprising: Head-mounted display devices (e.g., virtual reality (VR) devices) include a display screen positioned near or within the user's eyes and visual range, the head-mounted display device being communicatively coupled to one or more processors; as well as An adaptive map view inspection algorithm, which includes computational instructions stored in one or more processor-accessible memories; Standard database or model, The adaptive map view inspection algorithm is configured to access or implement the standard database or model to make the head-mounted display device device independent of one or more head-mounted display devices with different configurations. Wherein, the computation instructions of the adaptive map view inspection algorithm, when executed by the one or more processors, are configured to cause the one or more processors to perform a view-of-flight (VF) test, the VF test including: A VF test is performed on the display screen of the head-mounted display device, wherein the VF test is adapted to the area of ​​the display screen based on values ​​in the standard database or model, wherein the VF test is presented as a visualization that is the same or similar to one or more different head-mounted display devices, the one or more different head-mounted display devices having different displays, the different displays having different corresponding shapes, formats, sizes and / or resolutions; Receive visual test data indicating the user's field of vision; Based on the visual test data, one or more user-specific initial test locations are detected, wherein the one or more initial test locations define one or more healthy clusters indicating no scatoma and one or more damaged clusters indicating dark spots; and A spatial mapping is generated based on the one or more initial test locations to identify the locations of the one or more damaged clusters.

2. The VF analysis system of claim 1, wherein, The spatially mapped positions are spaced 0.5 degrees or 0.5 resolution apart in the vertical and / or horizontal angles.

3. The VF analysis system according to claim 1, wherein The standard database or model includes standard values ​​generated through quantile regression. The quantile regression involves generating standard values ​​from standard reference values, which include biometric values ​​such as refractive error, axial length, corneal curvature, or other independent variables that include user eye biometric values. The head-mounted display device is updated with or has access to the standard database or model to calibrate the head-mounted display device to ensure device independence when implementing the adaptive map field of view inspection algorithm.

4. A visual field (VF) analysis method for diagnosing and monitoring glaucoma by implementing adaptive map visual field testing, the visual field analysis method comprising: Performing VF testing on the display screen of an electronic display device (e.g., a virtual reality (VR) device), wherein the electronic display device includes a display screen positioned near or within the user's eye range, and the electronic display device is communicatively coupled to one or more processors; and The VF test adapts the region of the display screen to values ​​in a standard database or model. The VF test is presented as a visualization identical or similar to one or more different electronic display devices, each with a different display screen, shape, format, size, and / or resolution. The adaptive map view inspection algorithm is configured to access or implement the standard database or model to ensure device independence for one or more different configurations of the electronic display device. Receive visual test data indicating the user's field of vision; Based on the visual test data, one or more user-specific initial test locations are detected, wherein the one or more initial test locations define one or more healthy clusters indicating no scatomas and one or more damaged clusters indicating the presence of dark spots; and A spatial mapping is generated based on the one or more initial test locations to identify the locations of the one or more damaged clusters.

5. The VF analysis method according to claim 4, wherein the interval between the locations of the spatial mapping in the vertical angle and / or horizontal angle is 0.5 degrees or 0.5 resolution.

6. The VF analysis method according to claim 4, wherein, The standard database or model includes standard values ​​generated through quantile regression; The quantile regression involves generating standard values ​​from standard reference values, which include biometric values ​​such as refractive error, axial length, corneal curvature, or other independent variables that include user eye biometric values. The electronic display device is updated with or has access to the standard database or model to calibrate the electronic display device to ensure device independence when implementing the adaptive map view inspection algorithm.

7. A tangible, non-transitory computer-readable medium storing instructions for diagnosing and monitoring glaucoma by implementing adaptive map vision checks, the instructions, when implemented by one or more processors, causing the one or more processors to perform the following operations: A visual field (VF) test is implemented on a display screen of an electronic display device, such as a virtual reality (VR) device, wherein, The electronic display device includes a display screen positioned near or within the user's eyes and viewing distance, and the electronic display device is communicatively coupled to the one or more processors; and The VF test adapts the region of the display screen to values ​​in a standard database or model. The VF test is presented as a visualization identical or similar to one or more different electronic display devices, each with a different display screen, shape, format, size, and / or resolution. The adaptive map view inspection algorithm is configured to access or implement the standard database or model to ensure device independence for one or more different configurations of the electronic display device. Receive visual test data indicating the user's field of vision; Based on the visual test data, one or more user-specific initial test locations are detected, wherein the one or more initial test locations define one or more healthy clusters indicating no scatomas and one or more damaged clusters indicating the presence of dark spots; and A spatial mapping is generated based on the one or more initial test locations to identify the locations of the one or more damaged clusters.

8. The tangible, non-transitory computer-readable medium of claim 7, wherein, The spatial mapping positions are spaced 0.5 degrees or 0.5 resolution apart in the vertical and / or horizontal angles.

9. The tangible, non-transitory computer-readable medium according to claim 7, wherein The standard database or model includes standard values ​​generated through quantile regression; The quantile regression involves generating standard values ​​from standard reference values, which include biometrics such as refractive error, axial length, corneal curvature, or other independent variables that include user eye biometrics. The electronic display device is updated with or has access to the standard database or model to calibrate the electronic display device to ensure device independence when implementing the adaptive map view inspection algorithm.

Images