WO2019178100A1

WO2019178100A1 - Predicting result reversals of glaucoma hemifield tests

Info

Publication number: WO2019178100A1
Application number: PCT/US2019/021852
Authority: WO
Inventors: Mengyu WANG; Lucy Q. SHEN; Louis R. Pasquale; Tobias ELZE
Original assignee: The Schepens Eye Research Institute, Inc.
Priority date: 2018-03-12
Filing date: 2019-03-12
Publication date: 2019-09-19

Abstract

Methods, systems, and devices are provided for determining and/or predicting result reversals and/or false positive results of the glaucoma hemifield test, for example by assessing VF (visual field) mismatch features and quantifying the VF loss patterns in addition to the VF global indices. In general, if the test results generate consecutive ONL results, groups of features can be extracted from the VF results collected, such as the average VF global indices, VF mismatch measures between baseline VFs, and the archetype decompositions of the mean baseline VFs. The VF results can be decomposed into VF patterns (or archetypes). Reversal of GHT results can then be predicted based on an analysis by one or more processors of the data provided, for example by using logistic regression.

Description

Predicting Result Reversals of Glaucoma Hemifield Tests

RELATED APPLICATIONS

[0001] The present application claims priority to US Prov. Patent App. No. 62/641,785, entitled“Predicting Result Reversals Of Glaucoma Hemifield Tests,” and filed on March 12, 2018, which is incorporated herein by reference in its entirety.

FIELD

[0002] The present disclosure relates generally to a glaucoma hemifield test, and more particularly, to determining and/or predicting result reversals and/or false positive results of the glaucoma hemifield test.

BACKGROUND

[0003] Generally, a glaucoma hemifield test (GHT) can be used by clinicians to determine an assessment of a visual field (VF) of a patient where glaucomatous damage is often observed. In some examples, the test can generally compare five corresponding and mirrored areas in the superior and inferior visual fields of the patient. For example, the results can generally be “Outside Normal Limits” (ONL), which represents a significant difference determined in the superior and inferior fields of the patient;“Borderline,” which represents a suspicious difference in the fields but not enough to qualify as ONL; or“Within Normal Limits”

(WNL), which represents no significant differences in the fields. This test is often carried out on standard equipment used by optometrists, orthoptists, and ophthalmologists, such as the Humphrey visual field analyzer (HFA).

[0004] The glaucoma hemifield test can be a measurement used in standard automated perimetry to assist in the interpretation of visual fields measured with the Humphrey visual field analyzer. However, the glaucoma hemifield test frequently can return results that indicate outside normal limits (ONL) for a patient in a first test during a first visit to a clinician and then return results that indicate within normal limits (WNL) in a second test during a second visit. Thus the number of false positives presents a repetitive problem to clinicians when attempting to provide proper care to a patient. Because of this and other reasons, the current approach is not effective for many people.

SUMMARY

[0005] Methods, systems, and devices are provided herein for determining and/or predicting false positive results of the glaucoma hemifield test. For example in one exemplary embodiment, a method is provided for predicting result reversals of a glaucoma hemifield test including testing at least twice, on a visual device, a visual field of the patient such that at least two consecutive visual field results are obtained indicating Outside Normal Limits results. The method also includes decomposing, by a processor, the at least two visual field results of the patient into a plurality of visual field features. The method further includes determining, by the processor, whether the at least two visual field results should be reversed based on analysis of the plurality of visual field features.

[0006] The method can have numerous variations. For example, the plurality of visual field features can include an average visual field global indices, visual field mismatch measures between the at least two visual field results, and archetype decompositions of the at least two visual field results. In another example, the archetype decompositions of the at least two visual field results can include 16 weighted visual field archetypes. The plurality of weighted visual field archetypes can include one weighted visual field archetype configured to represent a normal vision field. In another example, the plurality of weighted visual field archetypes can include one or more weighted visual field archetypes configured to represent a loss of vision field. The one or more weighted visual field archetypes can be configured to represent a loss of vision field include 15 archetypes. In another example, determining whether the at least two visual field results should be reversed can include using logistic regression. Decomposing the at least two visual field results of the patient into the plurality of visual field features can include determining a percentage of each of the archetype decompositions of the at least two visual field results present in the at least two visual field results. The method can also include using unsupervised machine learning. Determining whether the at least two visual field results should be reversed can include identifying false positives. In another example, determining whether the at least two visual field results should be reversed can include predicting a probability of false positives. The processor can be part of the visual device, or the processor can be included in a remote computing device. In another example, the visual device can be a Humphrey visual field analyzer.

[0007] In another aspect, a visual device for determining visual field progression of a patient is provide that includes at least one input, at least one display, at least one sensor, at least one memory, and at least one processor. The visual device is configured to test at least twice a visual field of the patient such that at least two consecutive visual field results are obtained indicating Outside Normal Limits results; to decompose, by the processor, the at least two visual field results of the patient into a plurality of visual field features; and to determine, by the processor, whether the at least two visual field results should be reversed based on analysis of the plurality of visual field features.

[0008] The device can have numerous variations. For example, the plurality of visual field features can include an average visual field global indices, visual field mismatch measures between the at least two visual field results, and archetype decompositions of the at least two visual field results. The archetype decompositions of the at least two visual field results can include 16 weighted visual field archetypes. In another example, the plurality of weighted visual field archetypes can also include one weighted visual field archetype configured to represent a normal vision field. The plurality of weighted visual field archetypes can include one or more weighted visual field archetypes configured to represent a loss of vision field. In still another example, determining whether the at least two visual field results should be reversed further includes using logistic regression. Decomposing the at least two visual field results of the patient into the plurality of visual field features can further include determining a percentage of each of the archetype decompositions of the at least two visual field results present in the at least two visual field results. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The invention will be more fully understood from the following detailed description taken in conjunction with the accompanying drawings, in which:

[0010] FIG. 1 illustrates an embodiment of a visual field analyzer;

[0011] FIG. 2 illustrates an analyzer printout produced by the visual field analyzer of FIG. 1;

[0012] FIG. 3 illustrates an example diagrammatic view of a device architecture;

[0013] FIG. 4 illustrates one embodiment of a procedure for predicting result reversal of visual field results;

[0014] FIG. 5 illustrates an embodiment for performing the procedure for determining visual field results of FIG. 4;

[0015] FIG. 6 illustrates a possible data exchange step of the procedure of FIG. 4;

[0016] FIG. 7 illustrates an example of 16 visual field archetypes;

[0017] FIG. 8 illustrates an example of visual field decomposition into weighted visual field archetypes;

[0018] FIG. 9 illustrates parameter coefficients of a logistic regression model based on the procedure of FIG. 4;

[0019] FIG. 10 illustrates a receiver operating characteristic curve based on the procedure of FIG. 4;

[0020] FIG. 11 illustrates results of an exemplary patient with 3 consecutive GHT ONE results;

[0021] FIG. 12 illustrates results of an exemplary patient with GHT results reversal for MD of -3 dB or more; and [0022] FIG. 13 illustrates a receiver operating characteristic curve that illustrates

performance of an approach based on the procedure of FIG. 4 on a validation subset of data used herein.

[0023] The patent or application file contains at least one drawing executed in color. Copies of this patent or patent application publication with color drawing(s) will be provided by the Office upon request and payment of the necessary fee.

[0024] It should be understood that the above-referenced drawings are not necessarily to scale, presenting a somewhat simplified representation of various preferred features illustrative of the basic principles of the disclosure. The specific design features of the present disclosure, including, for example, specific dimensions, orientations, locations, and shapes, will be determined in part by the particular intended application and use environment.

DETAILED DESCRIPTION

[0025] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. As those skilled in the art would realize, the described embodiments may be modified in various different ways, all without departing from the spirit or scope of the present disclosure. Further, throughout the specification, like reference numerals refer to like elements.

[0026] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. As used herein, the singular forms “a,”“an,” and“the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms“comprises” and/or “comprising,” when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. As used herein, the term“and/or” includes any and all combinations of one or more of the associated listed items. The term“coupled” denotes a physical relationship between two components whereby the components are either directly connected to one another or indirectly connected via one or more intermediary components.

[0027] It is understood that the term“patient” or other similar term as used herein is inclusive of any subject— human or animal— on which an ocular assessment could be performed. The term“user” as used herein is inclusive of any entity capable of interacting with or controlling a device. The“user” may also be the“patient,” or the“user” and“patient” may be separate entities, as described herein.

[0028] Additionally, it is understood that one or more of the below methods, or aspects thereof, may be executed by at least one processor. The processor may be implemented in various devices, as described herein. A memory configured to store program instmctions may also be implemented in the device(s), in which case the processor is specifically programmed to execute the stored program instructions to perform one or more processes, which are described further below. Moreover, it is understood that the below methods may be executed by a specially designed device, a mobile device, a computing device, etc.

comprising the processor, in conjunction with one or more additional components, as described in detail below.

[0029] Furthermore, the methods, or aspects thereof, of the present disclosure may be embodied as non-transitory computer readable media on a computer readable medium containing executable program instructions executed by the processor. Examples of the computer readable mediums include, but are not limited to, ROM, RAM, compact disc (CD)- ROMs, magnetic tapes, floppy disks, flash drives, smart cards and optical data storage devices. The computer readable recording medium can also be distributed in network coupled computer systems so that the computer readable media is stored and executed in a distributed fashion, e.g., by a telematics server or a Controller Area Network (CAN).

[0030] Referring now to embodiments of the present disclosure, the accurate diagnosis of glaucoma relies heavily on the use of standard automated perimetry to measure VF loss. The GHT can be used to assist in the interpretation of VFs of a patient (for example, measured with the Humphrey visual field analyzer as discussed above and illustrated in FIG. 1). The GHT can compare symmetric VF sectors between the upper and lower hemifields. In some examples of the GHT, the GHT generally can have 6 possible outcomes: within normal limits (WNL), borderline, outside normal limits (ONL), general reduction of sensitivity, abnormally high sensitivity, and borderline or general reduction of sensitivity. As one possible set of the outcomes, outside normal limits can appear when the differences between a matched pair of mirrored zones exceeds the differences of about 99% of individuals in a normal population or both members of 2 paired zones are more abnormal than about 99.5% of individuals in a normal population. Borderline can denote when 2 paired zones are more abnormal than about 97% of the individuals, whereas the abnormality of the paired zones do not meet criteria for ONL. General reduction of sensitivity can appear when both conditions for ONL are not met and the best region of the VF is more abnormal than about 99.5% of the individuals in a normal population. Abnormally high sensitivity denotes that the best region of the VF has higher sensitivity than about 99.5% of the individuals in a normal population, which may indicate low reliability of the VF test. Within normal limits is assigned to the VF when none of those aforementioned conditions are met.

[0031] To reduce false discovery in some examples, a clinician can wait until receiving 2 consecutive GHT ONL results before considering a diagnosis of glaucomatous VF loss. However, other ranges are possible, for example receiving between 1 and 10 or 1, 2, 3, 4, 5,

6, 7, 8, 9, 10, etc. consecutive results. In addition, the sensitivity of GHT for early glaucomatous VF loss can be limited, compared to the sensitivity of the GHT for the full range of glaucomatous VF loss that can be high. Assuming that glaucomatous VF loss is irreversible, a conversion from 2 consecutive GHT ONL results to WNL results can be considered a GHT results reversal. Again, though, other ranges are possible, for example receiving between 1 and 10 or 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, etc. consecutive results.

[0032] When measuring the visual field of a patient, the results of a test (for example performed by a visual field analyzer 2, such as a Humphrey visual field analyzer like the Humphrey HFA Il-i - Perimetry as illustrated in FIG. 1) can be provided to identify normal vision or a type of vision defect, condition, disease state, etc. in an eye of a patient. The results provide information regarding various defects, for example glaucoma. Each result is generated on an analyzer printout 4 that provides a variety of standard information, as illustrated in FIG. 2. The information provided can relate generally to reliability indices 10, numerical displays 20, grey scale 30, total deviation 40, probability display 50, pattern deviation 60, global indices 70, glaucoma hemifield test 80, and visual field index 90.

[0033] A clinician can examine the results in an attempt to diagnose the type of vision loss present, if any. As discussed above, the clinician can especially focus on the visual field index 90 to identify glaucoma. However, the prevalence of false positive test results may cause many clinicians to be unwilling to diagnose glaucoma without repeated retesting.

[0034] Thus, an approach provided herein can be used to predict the occurrence of GHT results reversal to WNL using VF features. The VF features include the VF global indices, VF mismatch measures between baseline VFs, and computationally derived representative VF loss patterns (archetypes), discussed below. The VF mismatch measures capture the variation and similarity between the 2 baseline VFs, and the archetype decompositions quantify the spatial patterns of VF loss. However, different numbers of baselines can be used in some examples, such as 1-10 or 1, 2,3 ,4 ,5, etc. baselines from different testing. Thus by using this approach, clinicians can be supported quantitatively in the decision of whether 2 consecutive ONL GHT results will revert to WNL results.

[0035] While a Humphrey visual field analyzer is discussed above, a variety of visual devices can be used herein. FIG. 3 illustrates an example diagrammatic view of an exemplary device architecture according to embodiments of the present disclosure. As shown in FIG. 3, a device 109 may contain multiple components, including, but not limited to, a processor (e.g., central processing unit (CPU) 110, a memory 120, a wired or wireless communication unit

130, one or more input units 140, and one or more output units 150. It should be noted that the architecture depicted in FIG. 3 is simplified and provided merely for demonstration purposes. The device architecture depicted in FIG. 3 should be treated as exemplary only and should not be treated as limiting the scope of the present disclosure.

[0036] The processor 110 is capable of controlling operation of the device 109. More specifically, the processor 110 may be operable to control and interact with multiple components installed in the device 109, as shown in FIG. 3. For instance, the memory 120 can store program instructions that are executable by the processor 110 and data. The process described herein may be stored in the form of program instructions in the memory 120 for execution by the processor 110. The communication unit 130 can allow the device 109 to transmit data to and receive data from one or more external devices via a communication network. The input unit 140 can enable the device 109 to receive input of various types, such as audio/visual input, user input, data input, and the like. To this end, the input unit 140 may be composed of multiple input devices for accepting input of various types, including, for instance, one or more cameras 142 (i.e., an“image acquisition unit”), touch panel 144, microphone, sensors 146, one or more buttons or switches, and so forth. The input devices included in the input 140 may be manipulated by a user. Notably, the term“image acquisition unit,” as used herein, may refer to the camera 142, but is not limited thereto. The output unit 150 can display information on the display screen 152 for a user to view. The display screen 152 can also be configured to accept one or more inputs, such as a user tapping or pressing the screen 152, through a variety of mechanisms known in the art. The output unit 150 may further include a light source 154.

[0037] The device 109 can thus be programmed in a manner allowing it to perform the techniques for measuring and analyzing a visual field of a patient, as described herein.

[0038] To this end, techniques are disclosed herein relating to predicting false positive results of the glaucoma hemifield test and as further explored in Wang et ak, Reversal of Glaucoma Hemifield Test Results and Visual Field Features in Glaucoma, Ophthalmology,

Volume 125, Number 3, March 2018, incorporated herein by reference in its entirety. [0039] The various approaches provided herein can be used to determine and/or predict GHT result reversals by assessing VF mismatch features and quantifying the VF loss patterns in addition to the VF global indices, discussed in detail below. In general as illustrated in FIGS. 4 and 5, a clinician 102 tests the visual field of a patient 104 by using a visual field analyzer 106, such as a Humphrey visual field analyzer, in step 100. The test can be performed more than one time to generate one or more viable VF results for the patient 104 at step 200. For example, the approach can generate 2 VF results for the patient, however other numbers of VF results are possible as discussed above. If no glaucoma is detected, the test can be terminated. If the test results generate one or more ONL results, for example 2 consecutive ONL results, 3 groups of features can be extracted from the VF results collected in step 100: the average VF global indices, VF mismatch measures between baseline VFs, and the archetype decompositions of the mean baseline VFs at step 300. However, different numbers of consecutive results and/or different numbers of groups of figures, such as 1-10 or 3, 4, 5, 6, etc., can be used. The VF results can be decomposed into 16 VF patterns (or archetypes) that have been previously computationally derived, as illustrated in FIG. 4 and discussed in detail below. However, different numbers of VF patterns are possible, such as from 10-30 or 10, 15, 20, 25, etc. In some examples, a normal archetype is represented in AT1, and or or more (such as 15) visual field loss archetypes are represented in various AT values (such as AT2-AT16). The 16 visual field archetypes can be generated using visual field information from glaucoma patients. However, the same archetypes can be used for a variety of other eye conditions, such as stroke, pituitary disease, age-related macular degeneration, cataract, etc. Alternatively, new archetypes can be identified using visual field information from other eye conditions, such as stroke, pituitary disease, age-related macular degeneration, cataract, etc., and either used with glaucoma patients and/or used to analyze the one or more other eye conditions. The results can also include various mean deviation or defect (MD) values and/or various pattern standard deviation (PSD) values and/or various Decibel (dB) values. In some examples, MD can be a weighted mean value of all test points in a total deviation plot, and/or PSD can be a metric that indicates a difference in a sensitivity of adjacent tested points, and/or dB can be a logarithmic scale of stimulus intensity. The data can either be analyzed on the visual field analyzer 106 or transferred to another computing device 108, such as a desktop computer as illustrated in FIG. 6. Reversal of GHT results can then be predicted based on an analysis by one or more processors of the data provided. For example, logistic regression can be applied to predict GHT results reversal to WNL after 2 consecutive GHT ONL results using the VF features as independent variables in step 400. Again, however, different numbers of consecutive results can be used as discussed above. Various exemplary steps of one embodiment of the approach discussed herein will now be discussed in detail.

[0040] DETAILS ON ANALYSIS:

[0041] While specific values and analysis is provided below, the present disclosure is not limited thereto. The analysis provided herein is exemplary to illustrate the applicability of the approach discussed herein. VF results used herein were obtained by the Glaucoma Research Network, a consortium including the following glaucoma centers: Massachusetts Eye and Ear (MEE), Wilmer Eye Institute, New York Eye and Ear Infirmary, Bascom Palmer Eye Institute, and Wills Eye Hospital. VF data on various eyes with at least 3 reliable consecutively measured VFs were selected. The reliability criteria for VF selection were fixation loss of 33% or less, false-negative rates of 20% or less, and false-positive rates of 20% or less. A subset of eyes was selected additionally such that the GHT results for the first 2 VFs were ONL and the GHT results of the third VF were any of WNL, borderline, or ONL. The total deviation (TD) values from each of the 52 locations tested in the 24-2 pattern were extracted and used to derive the VF mismatch features and the VF loss patterns.

[0042] Initially, the proportions of eyes with GHT results reversal from ONL at baseline to

WNL on the second test for all VF loss severities were calculated. For the subset with 2 consecutive ONL results, the proportions of eyes with GHT results reversal on the third measurement to WNL for all VF loss severities also were evaluated. All statistical analyses were performed using R software (Version 3.3.1, R Foundation, Vienna, Austria). For the subset of eyes with 2 consecutive ONL results, 3 groups of features were extracted from baseline VFs: the average VF global indices, VF mismatch measures between baseline VFs, and the archetype decompositions of the mean baseline VFs. The global indices extracted included the mean deviation (MD), the pattern standard deviation (PSD), and the MD and PSD differences between the second and first VFs. The VF mismatch measures calculated include the standard deviation of the TD difference in all 52 locations between baseline VFs and the similarity index of the TDs between baseline VFs measured by the cosine similarity, a standard similarity measure between 2 vectors that measures the cosine of the angle between them.

[0043] For the archetype decomposition to quantify the VF spatial patterns, the average VFs

(i.e., average TD values at all 52 locations) of the first 2 VFs were decomposed into 16 VF patterns (archetypes) computationally derived (as illustrated in FIG. 7). Again, however, the approach is not limited thereto. The VF loss patterns then were represented by the decomposition coefficients, which sum up to 100% (as illustrated in FIG. 8). Thus FIGS. 7 and 8 illustrate quantifying VF loss patterns with archetypes (ATs), such that FIG. 7 illustrates the 16 computationally derived archetypes and FIG. 8 illustrates an example of the

VF decomposition to the VF archetypes. In short, the 16 VF archetypes were identified by an unsupervised machine learning method (archetypal analysis) based on more than 13,000 reliable VFs and represent one or more weighted visual field archetypes configured to represent a normal vision field (for example, AT1) and one or more weighted visual field archetypes configured to represent loss vision field (for example, AT2-AT15). Nine of those archetypes represent clinically recognizable glaucomatous patterns with similarity to patterns determined by manual inspection of VF data and confirmed by clinical correlation:

archetypes 8 and 13 (altitudinal VF loss); archetypes 9, 10, and 16 (partial arcuate defects); archetypes 3 and 5 (nasal step); and archetypes 14 and 16 (paracentral). Archetype 2 was associated with both glaucomatous VF loss and a higher occurrence of ptosis. Archetype 1 represents the normal VF. All other archetypes represent clinical conditions different from glaucoma, such as hemianopia (archetypes 12 and 15). However, as discussed above, the approach discussed herein is not limited to glaucoma, and instead can be used when analyzing a variety of eye conditions, such as stroke, pituitary disease, age-related macular degeneration, cataract, etc.

[0044] Logistic regression was applied to predict GHT results reversal to WNL after 2 consecutive GHT ONL results using the VF features as independent variables, however other consecutive results numbers can be used. The technique of weighted error penalization was used to mitigate the underestimation of GHT results reversals because of an imbalanced dataset. Stepwise regression was performed to select the optimal feature combination that predicts the GHT results reversal based on Bayesian information criterion. The regression analyses were implemented for eyes with MD of -3 dB or more and MD of -6 dB or more and less than -3 dB, respectively. These ranges were used for this exemplary analysis and do not limit applicability of the approach herein. Ten-fold cross-validation was applied to evaluate the predictive model performance by the area under the receiver operating characteristic curve (AUC). The AUCs of the embodiment of the approach discussed herein to predict GHT results reversal were compared with the AUC performance of models that included only VF global indices and models that also included the VF global indices plus VF mismatch measures. Cross-validation was used to test the performance of the approach on the data that are not used in model training.

[0045] The dataset herein was partitioned into parts, such as 10 parts for this exemplary analysis, and each of the 10 subsets was used once as testing partitions, whereas the model was trained on the 9 remaining partitions. Thus, the AUCs for model evaluation were calculated on different data subsets than those used for generating the models. Because clinical data were available only in the MEE dataset, it was excluded from the training dataset, and instead, its clinical data was used to test the robustness of the approach discussed herein. The AUC performance of the model was evaluated. The jackknife resampling method was used to compute the AUC confidence interval (Cl). For a subset of the MEE data, an assessment of glaucoma status at the time of the third VF test was made based on the consensus of 2 glaucoma specialists masked to study results by reviewing the fundus photography for glaucomatous optic disc changes and OCT images for characteristic nerve fiber layer thinning closest to the test date of the third VF. When structural data were equivocal, a reliable VF that postdated the third test was used to confirm the presence or absence of glaucoma. In addition, eye surgical history was extracted from medical records for the subset. Thus, instead of defining GHT results reversal as 2 ONL test results followed by WNL results on the third test, the GHT results reversal was alternatively defined as 2 ONL results and absence of clinical glaucoma. The AUC performance and prediction accuracy was evaluated using the approach discussed herein that was trained to predict GHT results reversal defined as 2 GHT ONL results followed by WNL results. Again as discussed above, the analysis is not limited thereto, though.

[0046] RESULTS OF TEST DATA:

[0047] Data was used for 44,503 eyes of 26,130 patients (mean age, 63.8 ± 14.3 years) with at least 3 reliable annually measured VFs. However, similar approaches can be used with patients of various ages, such as between about 15 and 100 years old, or between about 40 and 80 years old, or between about 50 and 80 years old, or between about 60 and 80 years old, such as use with elderly populations at personal homes or in nursing homes, etc. Of these, 16,604 eyes of 12,688 patients (mean age, 66.4 ± 13.5 years) demonstrated GHT ONL results on the first 2 tests during the first 2 visits. The GHT results reversal prevalence to WNL results after 2 consecutive ONL determinations increased from 0.1% for MD less than - 12 dB to 13.8% for MD of -3 dB or more. However, again, a variety of consecutive test results can be used and resulting in a range of MD and dB values. The GHT results are especially relevant for the diagnosis of glaucomatous VF loss at a mild stage (MD, >-6 dB). Therefore, data from all eyes with mild VF loss and the first 2 VFs with GHT ONL results were analyzed. However, one or both eyes can be analyzed, and 2 or more VFs can be analyzed as discussed above. For the resulting 6,481 eyes, 9.2% reversed to WNL results at the third visit. An MD of -3 dB or more with the first 2 VFs showing GHT ONL results were then selected. The dataset yielded 2,199 eyes of 2,077 patients (mean age, 64.3 ± 12.0 years) with 13.8% showing GHT results reversals. For example, FIGS. 9 and 10 illustrate the best predictive model for MD of -3 dB or more selected by stepwise regression to predict the GHT results reversals with the optimal parameter combination. The figures thus show the best predictive model selected by stepwise regression to predict glaucoma hemifield test results reversals for mean deviation (MD) of -3 dB or more, with FIG. 9 illustrating parameter coefficients of the logistic regression model and FIG. 10 illustrating receiver operating characteristic curve. The blue cross illustrates the decision threshold for a fixed false-positive rate of one third, as described herein. AT is the archetype, AUC is the area under the receiver operating characteristic curve; Cl is the confidence interval; PSD is the pattern standard deviation; SD-TDD is the standard deviation of the total deviation difference between the 2 baseline visual fields; and SI-TD is the similarity index of the total deviations between the 2 baseline visual fields.

[0048] Mean deviation was associated positively and PSD was associated negatively with

GHT results reversals for this example analysis; however other analyses can use different criteria. The standard deviation of the TD difference was associated positively and the similarity index of TDs was associated negatively with GHT results reversals. Based on stepwise regression, 8 of the original 16 archetypes were selected in this example, while other numbers of selected VFs can be used in other examples. Seven archetypes (archetypes 2, 4,

5, 7, 9, 11, and 12) were associated positively and 1 archetype (archetype 16) was associated negatively with GHT results reversals. The MD and PSD difference between baseline VFs did not remain in the optimal feature combination. Compared with the model with VF global indices, the AUC performance of cross-validation to predict GHT results reversals increased significantly from 0.669 (95% Cl, 0.668-0.671) to 0.745 (95% Cl, 0.744-0.746; P < 0.001) by adding VF mismatch features. Furthermore, the AUC performance of the models increased significantly to 0.770 (95% Cl, 0.769-0.772; P < 0.001) by adding the archetype features. If a false-positive rate was chosen for the approach discussed herein of 33% (for example, as denoted by the blue cross in FIG. 10) for instance, 74.5% of the GHT results reversals could be predicted correctly. The corresponding probability threshold was 0.51. For eyes with MD of -6 dB or more and less than -3 dB, the AUC for the best predictive model was 0.820 (95% Cl, 0.819-0.820).

[0049] Exemplary results for an exemplary patient are provided in FIG. 11. However, again as discussed above, other patients can be analyzed in different was. For example, FIG. 11 shows an exemplary patient with 3 consecutive GHT ONL results, and FIG. 12 shows an exemplary patient with GHT results reversal for MD of -3 dB or more. FIGS. 11 and 12 thus illustrate examples of visual fields, with FIG. 11 illustrating 3 consecutive glaucoma hemifield test results outside normal limits (ONL) and FIG. 12 illustrating GHT results reversal from 2 consecutive GHT ONL results. Visual fields are decomposed into the combination of archetypes, and nonconsidered archetypes (ATs) are those archetypes that are not in the parameter set of the best predictive model. The Avg. is the average; MD is the mean deviation; PSD is the pattern standard deviation; SD-TDD is the standard deviation of the total deviation difference between the 2 baseline visual fields; SI-TD is the similarity index of the total deviations between the 2 baseline visual fields; and WNL represents within normal limits. Consistent with the relationship between the VF features and the occurrence of GHT results reversals shown in FIG. 9, the standard deviation of the TD difference is higher and the similarity index of TDs between the baseline VFs of the patient with GHT results reversal is lower than those of the patient with 3 consecutive GHT ONL results. Furthermore, the substantial presence of archetype 11 (9.7%) suggests that lens rim artifacts also contribute to the GHT results reversal as shown in FIG. 11. Likewise, the presence of archetypes only related to glaucomatous loss, namely, archetype 2 (21.5%), archetype 4 (4.7%), and archetype 5 (15.8%), provides evidence that the third GHT is likely to reveal ONL results after 2 GHT ONL results. In contrast, the average MD and PSD of the patient with GHT results reversal for the baseline VFs are not higher and lower than the patient with 3 GHT ONL results as expected, respectively. 644 eyes of 576 patients (mean age, 64.2 ± 14.3 years) with MD of -6 dB or more from MEE were used to test the approach discussed herein performance trained with the data from the other Glaucoma Research Network sites. 50 of 644 eyes (7.8%) showed GHT results reversals from 2 ONL results to WNL results.

The AUC performance to predict the GHT results reversals was 0.870 (95% Cl, 0.870- 0.870). As shown in FIG. 13, 92.0% (95% Cl, 92.0%-92.0%) of the GHT results reversals were predicted correctly by the approach discussed herein, with the tradeoff of 33.3% of the patients with 3 consecutive GHT ONL results to be misclassified. Thus FIG. 13 illustrates a receiver operating characteristic curve that illustrates performance of the approach discussed herein on a validation subset of the data used herein, including 644 eyes with mean deviation of -6 dB or more. The blue cross illustrates the decision threshold for a fixed false-positive rate of one third. AUC is the area under the receiver operating characteristic curve, and Cl is the confidence interval. The corresponding probability threshold was 0.44. At this specificity level, any VFs with predicted probability larger than 0.44 were classified to be GHT results reversals. 97 eyes were additionally selected that included 48 eyes with GHT results reversals and 49 eyes without GHT results reversals. Of the 40 eyes diagnosed with glaucoma based on clinical data, 20.0% showed GHT results reversal. Of the 57 eyes without glaucoma, 70.2% showed GHT results reversal. The AUC for predicting the GHT results reversals for the subset of 97 eyes was 0.774 (95% Cl, 0.773-0.775). The approach discussed herein correctly predicted 68.8% (95% Cl, 68.6%-68.9%) of the GHT results reversals, with the tradeoff of 33.3% of misclassification for the patients with 3 consecutive GHT ONL results. The AUC for predicting the GHT results reversals defined by 2 ONLs with absence of glaucoma was 0.773 (95% Cl, 0.772-0.774). The approach discussed herein correctly predicted 87.7% (95% Cl, 87.6%-87.8%) of the GHT results reversals, with the tradeoff of 33.3% of misclassification. Although the AUC performance to predict 2 ONL results with absence of glaucoma is not significantly different (P ¼ 0.18) from the AUC performance to predict the GHT results reversals, the prediction accuracy for 2 ONL results with absence of glaucoma was significantly higher (P < 0.001) than that of predicting the GHT results reversals. The GHT is a standard parameter included in the Humphrey Field Analyzer that aims to aid clinicians in the diagnosis of glaucomatous VF loss. The approach discussed herein demonstrates that in VFs with mild severity, GHT can revert from 2 consecutive ONL results to WNL results in a significant portion of eyes (13.8% for MD > -3 dB).

[0050] Thus the inclusion of computationally derived VF mismatch and archetype features significantly improves the prediction of whether 2 consecutive GHT ONL results will revert to WNL results compared with models with VF global indices alone. GHT results reversals are related positively to MD and negatively to PSD. The standard deviation of the TD difference and the similarity index of TDs that characterize the consistency of the baseline VFs are related positively and negatively to GHT results reversals, respectively. For eyes with MD of -3 dB or more, the occurrence of GHT results reversals is more likely to be associated with archetypes related to nonglaucomatous, severe widespread VF loss and lens rim artifacts, and less likely to be associated with archetypes related to typical early glaucomatous VF loss (as shown in FIGS. 11 and 12). For example, the 3 archetypes with larger positive coefficients are archetypes 7, 11, and 12. Archetype 7 denotes central VF defects that are more typical for macular disorders. Archetype 11 typically is associated with VF measurement rim artifacts related to the use of high hyperopic correcting lenses.

Archetype 12 is representative of hemianopia, which typically is caused by stroke. The high and positive coefficients of the nonglaucomatous archetypes 7, 11, and 12 in the predictive model of GHT results reversals therefore are explained. The 4 archetypes with smaller positive coefficients are archetypes 2, 4, 5, and 9, and the only archetype with negative coefficient is archetype 16. Archetypes 2, 4, 5, 9, and 16 are all related to early glaucomatous VF loss.

[0051] The approach discussed herein thus can provide additional aid to clinicians when interpreting GHT ONL results, as an exemplary use of the approach discussed herein. The approach generates probabilities for GHT results reversals from VF features calculated from baseline VFs. As an exemplary guideline, for a given false-positive rate of 33.3%, the decision probability thresholds are 0.48 (MD > -3 dB) and 0.51 (-6 dB < MD < -3 dB), respectively. A value greater than the respective threshold would falsely predict a GHT results reversal in 33.3% of the cases with 3 consecutive GHT ONL results, but correctly predict it in 74.5% for MD of -3 dB or more and in 83.9% for MD of -6 dB or more and less than -3 dB of the GHT results reversals, respectively. The AUC and accuracy for predicting GHT results reversals in the MEE validation data were 0.870 (95% Cl, 0.870-0.870) and 92.0% (95% Cl, 92.0%-92.0%), respectively, with a specificity of 66.7% prescribed, and thus demonstrate that the approach discussed herein are robust. The AUC performance of the approach on the MEE validation data was better than the AUC performance. The reason for the better predicted results may speak to the robustness of the approach with a prediction that was more than the upper bound of the 95% Cl. For the additional subset of 97 eyes with clinical data, the model performance (0.774; 95% Cl, 0.773-0.775) was significantly lower (P < 0.001) than the model performance with all MEE data. The lower model performance was expected, because the GHT results reversal frequency was set to be 50% in this subset and is significantly higher than the GHT results reversal frequency of the overall MEE data (7.8%). Instead of predicting GHT results reversals defined solely based on VFs, the GHT results reversals may also be defined by 2 ONL results with negative glaucoma diagnosis. The accuracy to predict the GHT results reversals defined based on glaucoma diagnosis significantly outperformed (P < 0.001) the predicting accuracy for GHT results reversals. Interestingly, the approach trained based on fitting the GHT results reversals defined solely by VFs was better at predicting patients with 2 ONL results and no glaucoma diagnosis. The favorable results from this approach show that a model purely based on parameters from 2 previous VFs can predict the glaucoma diagnosis. Although some clinicians naturally rely on other clinical data to make glaucoma management decisions, especially information about optic nerve integrity, the approach provided herein based purely on VF features may be used instead or may augment the clinical decision-making process.

[0052] Studies have shown that eye surgeries, including intraocular pressure-lowering surgeries, cataract extraction, and ranibizumab treatment, can lead to enhanced VF sensitivity, which may trigger GHT results reversals. For the 97 eyes analyzed herein, no significant effects of eye surgeries on GHT results reversals were found. [0053] The approach discussed herein can help clinicians to strike a balance between the cost savings associated with deferral of treatment despite 2 or more consecutive GHT ONL results versus the sight preservation associated with treatment after a second abnormal GHT result is recorded. Thus the occurrence of GHT results reversals can be predicted by assessing VF mismatch features and quantifying the VF loss patterns in addition to the VF global indices. The approach discussed herein can consequently assist clinicians with determining whether GHT ONL results represent true glaucomatous VF loss.

[0054] Thus, a GHT outside normal limits (ONL) result of a patient in a first visit cn potentially be followed by a GHT within normal limits (WNL) in a second visit. GHT reversals can be predicted using the approach herein from positive back to negative to better interpret the GHT results. Additionally, false positive cases defined by two GHT ONL followed by one GHT WNL can be corrected using the approach herein. False positive cases can also be defined by one or more GHT ONL followed by one or more GHT WNL or one GHT ONL followed by one or more GHT WNL. A logistic regression model can be used herein as a classifier, and/or other classifiers can be used, such as random forest, deep neural network, etc.

[0055] While there have been shown and described illustrative embodiments that provide for visual field testing in eyes, it is to be understood that various other adaptations and modifications may be made within the spirit and scope of the embodiments herein. For instance, while a visual device is frequently mentioned throughout the present disclosure, the techniques described herein may also be implemented on various computers or similar machines. Thus, the embodiments of the present disclosure may be modified in any suitable manner in accordance with the scope of the present claims. The foregoing description has been directed to embodiments of the present disclosure. It will be apparent, however, that other variations and modifications may be made to the described embodiments, with the attainment of some or all of their advantages. Accordingly, this description is to be taken only by way of example and not to otherwise limit the scope of the embodiments herein.

Claims

WHAT IS CLAIMED IS:

1. A method of predicting result reversals of a glaucoma hemifield test comprising: testing at least twice, on a visual device, a visual field of the patient such that at least two consecutive visual field results are obtained indicating Outside Normal Limits results; decomposing, by a processor, the at least two visual field results of the patient into a plurality of visual field features;

determining, by the processor, whether the at least two visual field results should be reversed based on analysis of the plurality of visual field features.

2. The method of claim 1, wherein the plurality of visual field features includes an average visual field global indices, visual field mismatch measures between the at least two visual field results, and archetype decompositions of the at least two visual field results.

3. The method of claim 2, wherein the archetype decompositions of the at least two visual field results includes 16 weighted visual field archetypes.

4. The method of claim 2, wherein the plurality of weighted visual field archetypes includes one weighted visual field archetype configured to represent a normal vision field.

5. The method of claim 2, wherein the plurality of weighted visual field archetypes includes one or more weighted visual field archetypes configured to represent a loss of vision field.

6. The method of claim 5, wherein the one or more weighted visual field archetypes configured to represent a loss of vision field include 15 archetypes.

7. The method of claim 1, wherein determining whether the at least two visual field results should be reversed includes using logistic regression.

8. The method of claim 1, wherein decomposing the at least two visual field results of the patient into the plurality of visual field features includes determining a percentage of each of the archetype decompositions of the at least two visual field results present in the at least two visual field results.

9. The method of claim 1, further comprising using unsupervised machine learning.

10. The method of claim 1, wherein determining whether the at least two visual field results should be reversed includes identifying false positives.

11. The method of claim 1, wherein determining whether the at least two visual field results should be reversed includes predicting a probability of false positives.

12. The method of claim 1, wherein the processor is part of the visual device.

13. The method of claim 1, wherein the processor is included in a remote computing device.

14. The method of claim 1, wherein the visual device is a Humphrey visual field analyzer.

15. A visual device for determining visual field progression of a patient comprising: at least one input;

at least one display;

at least one sensor;

at least one memory; and

at least one processor,

wherein the visual device is configured:

to test at least twice a visual field of the patient such that at least two consecutive visual field results are obtained indicating Outside Normal Limits results,

to decompose, by the processor, the at least two visual field results of the patient into a plurality of visual field features, and

to determine, by the processor, whether the at least two visual field results should be reversed based on analysis of the plurality of visual field features.

16. The device of claim 15, wherein the plurality of visual field features includes an average visual field global indices, visual field mismatch measures between the at least two visual field results, and archetype decompositions of the at least two visual field results.

17. The device of claim 16, wherein the archetype decompositions of the at least two visual field results includes 16 weighted visual field archetypes.

18. The device of claim 16, wherein the plurality of weighted visual field archetypes includes one weighted visual field archetype configured to represent a normal vision field.

19. The device of claim 16, wherein the plurality of weighted visual field archetypes includes one or more weighted visual field archetypes configured to represent a loss of vision field.

20. The device of claim 15, wherein determining whether the at least two visual field results should be reversed further includes using logistic regression.

21. The device of claim 15, wherein decomposing the at least two visual field results of the patient into the plurality of visual field features further includes determining a percentage of each of the archetype decompositions of the at least two visual field results present in the at least two visual field results.