System and method for measuring eye contribution to binocular vision

Abstract

A method of evaluating a binocular imbalance between left and right eyes of a subject, is provided. The method includes selecting a stimulus condition including quality values selected from a numerical scale that quantifies a perceptible quality of an image; concurrently displaying a plurality of images to the subject based on the stimulus condition, the plurality of images including a left-eye image and a right- eye image displayed with their perceptible qualities defined by the quality values; receiving an input from the subject sequentially ranking an apparent perceived quality of the plurality of images; and calculating a scaling factor defining a relative imbalance in subjective quality between the subject's left eye and right eye, based on the sequential ranking. A corresponding system and computer-readable medium are also provided.

Description

SYSTEM AND METHOD FOR MEASURING EYE CONTRIBUTION TO BINOCULAR VISION TECHNICAL FIELD

[0001] The present disclosure generally relates to evaluating human vision. More particularly, the disclosure relates to a system and method for measuring each eye’s contribution to binocular vision to evaluate interocular differences or imbalances between the left and right eyes of a subject. BACKGROUND

[0002] In typical human vision, the inputs from the two eyes of a person are combined in the brain to give a unified perception of the outside world. This process involves aspects of binocular combination, where the information from the two eyes is added together, and interocular suppression, where the inputs from the two eyes compete with each other.

[0003] The unified “binocular” vision that results from this process is superior to the “monocular” vision that occurs when only one eye is used. Visual performance is improved on a variety of tasks. Stereopsis, the comparison of the two eyes’ inputs to make three-dimensional judgements of distance, becomes possible.

[0004] Certain visual disorders can affect the way in which the two eyes’ inputs are processed. A common visual disorder, amblyopia (or “lazy eye”), involves an imbalance in how the inputs from the two eyes are processed. Recent studies provide evidence that the binocular imbalance in amblyopia plays a primary role in the condition. Methods to address this imbalance have been incorporated into novel therapies that attempt to re-train the binocularity of amblyopic patients.

[0005] Therefore, there is a need for a method of determining or quantifying the relative strength and quality of the two eyes’ contribution to binocular vision, so that binocular imbalance can subsequently be treated or corrected. SUMMARY

[0006] According to an aspect of the disclosure, a method of evaluating imbalancebetween a contribution of a left eye of a subject and a contribution of a right eye of a subject to binocular vision is provided. The method comprises a step of selecting a stimulus condition, including a set of images shown to one eye and a second set shown to the other eye. These images vary in a quality of interest (e.g. having different luminance contrast). The method further comprises a step of concurrently displaying a plurality of images to the subject based on the stimulus condition, the plurality of images including a left-eye image set comprising at least one image presented only to a left eye of the subject and a right-eye image set comprising at least one image presented only to a right eye of the subject, each image of the left eye image set being displayed with a quality (e.g. its luminance range for the case of luminance contrast) defined by a corresponding one of the at least one left eye quality values (e.g. a set of luminance contrasts), and each image of the right-eye image set being displayed with a quality defined by a corresponding one of the at least one right eye quality values. The method further comprises steps of receiving an input from the subject sequentially ranking the subjective perceived quality of the plurality of images (e.g. their visibility when the images vary in contrast), and calculating a scaling factor defining a relative imbalance in the visual processing between the subject’s left eye and right eye, based on the sequential ranking of the plurality of images.

[0007] According to an aspect, a method of evaluating a binocular imbalance between left and right eyes of a subject is provided. The method includes: i) selecting a stimulus condition including a left-eye set comprising at least one left quality value, and a right-eye set comprising at least one right quality value, the left quality value and the right quality value being selected from a numerical scale that quantifies a perceptible quality of an image; ii) concurrently displaying a plurality of images to the subject based on the stimulus condition, the plurality of images including a left-eye image set comprising at least one image presented only to a left eye of the subject and a right-eye image set comprising at least one image presented only to a right eye of the subject, each image of the left-eye image set being displayed with the perceptible quality defined by a corresponding member of the at least one left quality value, and each image of the right-eye image set beingdisplayed with the perceptible quality defined by a corresponding member of the at least one right quality value; iii) receiving an input from the subject sequentially ranking an apparent perceived quality of the plurality of images; and iv) calculating a scaling factor defining a relative imbalance in subjective quality between the subject’s left eye and right eye, based on the sequential ranking of the plurality of images.

[0008] According to an aspect, a computing system for evaluating a binocular imbalance between left and right eyes of a subject is provided. The system inclujdes: at least one display; at least one user input; and a processor operatively coupled to the at least one display and the at least one user input, the processor being configured to: i) select a stimulus condition including a left-eye set comprising at least one left quality value, and a right-eye set comprising at least one right quality value, the left quality value and the right quality value being selected from a numerical scale that quantifies a perceptible quality of an image; ii) concurrently display, via the at least one display, a plurality of images to the subject based on the stimulus condition, the plurality of images including a left-eye image set comprising at least one image presented only to a left eye of the subject and a right-eye image set comprising at least one image presented only to a right eye of the subject, each image of the left-eye image set being displayed with the perceptible quality defined by a corresponding member of the at least one left quality value, and each image of the right-eye image set being displayed with the perceptible quality defined by a corresponding member of the at least one right quality value; iii) receive an input from the subject, via the at least one user input, sequentially ranking an apparent perceived quality of the plurality of images; and iv) calculate a scaling factor defining a relative imbalance in subjective quality between the subject’s left eye and right eye, based on the sequential ranking of the plurality of images.

[0009] According to an aspect, a non-transitory computer-readable medium is provided. The medium stores instructions thereon which, when executed by at least one processor of a computing system, cause the computing system toperform a method comprising: i) selecting a stimulus condition including a left-eye set comprising at least one left quality value, and a right-eye set comprising at least one right quality value, the left quality value and the right quality value being selected from a numerical scale that quantifies a perceptible quality of an image; ii) concurrently displaying a plurality of images to the subject based on the stimulus condition, the plurality of images including a left-eye image set comprising at least one image presented only to a left eye of the subject and a right-eye image set comprising at least one image presented only to a right eye of the subject, each image of the left-eye image set being displayed with the perceptible quality defined by a corresponding member of the at least one left quality value, and each image of the right-eye image set being displayed with the perceptible quality defined by a corresponding member of the at least one right quality value; iii) receiving an input from the subject sequentially ranking an apparent perceived quality of the plurality of images; and iv) calculating a scaling factor defining a relative imbalance in subjective quality between the subject’s left eye and right eye, based on the sequential ranking of the plurality of images. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 is a flow chart illustrating a method for measuring eye contribution to binocular vision, according to an embodiment.

[0011] Figure 2 is a schematic representation of a system for measuring eye contribution to binocular vision, according to an embodiment.

[0012] Figure 3A is a graphical representation of images displayed according to different stimulus conditions, and Figure 3B is a graphical representation of different stimulus conditions, according to an embodiment.

[0013] Figure 4 is a graphical representation of a set of images displayed as part of the method of Figure 1.

[0014] Figure 5 is a visual representation of a look-up table and a likelihood distribution after a given iteration and sequential ranking, according to the method of Figure 1.DETAILED DESCRIPTION

[0015] As will be described in further detail below, the proposed system and method, according to one aspect of the present description, aims to evaluate the relative strength and quality of the input from each eye of a subject contributing to binocular vision, using an innovative dichoptic ordering test. The proposed system and method are adapted to make behavioral measurements of the human visual system by tasking a subject with ranking images according to perceived subjective quality.

[0016] The proposed system and method advantageously allow for identifying subjects for whom a binocular imbalance is likely to be the cause of their visual impairment. The binocular imbalance may be the result of interocular suppression, difference in input gain, distortion of the image, loss of fine detail received by the eyes of the subject, or other factors. Further, the proposed system and method allow for tuning eventual treatments prescribed to subjects by evaluating a specific degree of imbalance that needs to be corrected, and for tracking changes in binocular balance that result from said treatments. For example, the proposed system and method may be used as an end-point in clinical trials of new treatments.

[0017] The method generally relies on measuring each eye’s input by tasking a subject, or user, with ranking concurrently displayed images, or visual stimuli. A dichoptic presentation of the images is used to concurrently display a set of images, such as “sprite” images of fruits, where a subset of images is visible only to the left eye, and another subset of images is visible only to the right eye. The subject is tasked with ranking the images based on a perceived quality of each image, e.g., for the case of luminance contrast, ranking based on how visible the images appear to the subject. In that case, the subject may be tasked with ranking, or selecting, the images in an order from a highest to lowest apparent or perceived contrast. Based on the ranking of the dichoptic images, a contrast scaling factor, or interocular contrast ratio, between the eyes of the subject would then be calculated indicative of the binocular imbalance between the eyes of the subject.In the more general case, the relative imbalance in any other visual quality such as blurriness or distortion could be calculated for images varying in that quality.

[0018] The method described herein is designed to select sets of image qualities to be displayed in a manner that will efficiently allow calculating imbalances in the relative strength or quality of the input from the two eyes. This is achieved by adopting numerical scales or numerical quality values that quantify various perceptible qualities of an image such as their contrast, blur, or distortion, among others. Such numerical quality values can correspond to mathematical relations established between the perceived contrast, blur, distortion or other qualities of a stimulus and the physical properties of the stimulus. The intensity of a particular quality of interest (e.g. contrast, blur, distortion, etc.), and the scale associated with that quality (as there are multiple ways one could approach quantifying these aspects via a quality value) are interchangeable.

[0019] Sets of images varying in the quality of interest are shown to the two eyes and the subject is tasked with ranking the images based on their perceived quality (e.g. their visibility when the images vary in contrast, their sharpness when they vary in blur, or their fidelity when they vary in distortion). A scaling factor adjusts the balance in numerical quality presented to the two eyes, searching for a compensation that counter-acts any imbalance. For example, if stimuli shown to one eye are perceived as more visible, the contrast of the images shown to that eye will be reduced. If the stimuli shown to one eye are seen as sharper, then the images shown to that eye will be blurred. If the stimuli shown to one eye are seen as being less distorted, the images shown to that eye will be distorted by a scrambling algorithm. To find the scaling factor, a series of trials, or iterations, are performed. On each iteration, a new stimulus condition, comprising a set of quality values for each eye, is selected for displaying the images, with the particular stimulus condition selected by an entropy-minimizing procedure that selects a stimulus condition allowing the determination of the most likely contrast scaling factor, indicative of the binocular imbalance, in a small number of iterations.

[0020] In the case of contrast, the system and method described herein are based on the principle that when dichoptic visual stimuli, or images, are presented with equal contrast to a subject, the image presented to the weaker eye, i.e., the amblyopic eye, will be perceived by the subject as having a lower contrast, or visibility, than the image presented to the stronger eye, i.e., the “fellow” eye.

[0021] In the present disclosure, the term “stimulus” or “visual stimulus” refers to an image having a determined quality value and that is presented or displayed to a subject. For example, visual stimuli may include a pair of fruit images displayed to a subject, comprising a visual stimulus for each eye.

[0022] The term “dichoptic” is used to refer to a way of displaying a stimuli or images such that each image is visible only to a specific eye. For example, dichoptic stimuli can refer to a pair of images displayed to a subject where one of the images is only seen by the left eye, and the other image is only seen by the right eye of the subject. Dichoptic stimuli, or dichoptic images, can be displayed using a range of technologies, such as red-cyan images displayed and visualized using matching red-green anaglyph glasses, also known as “3D glasses”, lenticular displays, presentation of the two eye’s images on a divided display or two separate displays (such as in a stereoscope), and head-mounted display technologies, where each eye views a separate image. Other technologies can be contemplated without departing from the present disclosure.

[0023] The term “luminance contrast”, or simply contrast, refers to a range of luminance contained in a visual stimulus or image. The highest contrast image possibly displayed on a screen is one which ranges from the highest pixel luminance, i.e., white, to the lowest pixel luminance, i.e., black, that can be shown. The lowest contrast image can be understood as an image in which all of the pixels have the same luminance, which effectively renders the image invisible, provided there is no difference in the color used across the image. Further, the contrast of an image can be determined relative to a background of the image. Therefore, on a black background, a high contrast image may appear white, while a low contrast image may faintly appear as a black-on-black image. For complex stimulusimages, it is common to use the root-mean-square (RMS) contrast metric, which is based on the standard deviation of the pixel luminances. There is a logarithmic relationship between physical contrast and the perception of contrast. This is exemplified by the just-noticeable difference for a change in contrast. The minimum change in contrast that is needed for a stimulus to appear different to a reference image is proportional to the contrast of the reference image. Therefore, the numerical scale for contrast used as an input as part of the method described herein can be based on the logarithm of the physical contrast.

[0024] The term “blur” generally refers to processes, typically optical in nature, that reduce the detail seen in an image. Different processes can result in different types of blur, and different types of blur can be quantified in different ways for establishing a numerical scale of quality. For example, a scattering of light by passing through a cloudy medium like a cataract in the eye will result in a reduction in the contrast of finer details (higher spatial frequencies). This type of blur can be quantified by the standard deviation of a Gaussian process. This Gaussian blur is distinct from that experienced as a result of optical defocus, which also shifts the location of some features (e.g. points of light seen out of focus in a camera become small "rings"). This blur is better modelled by a Sinc function. The relationships between physical and perceived blur (both Gaussian and Sinc) are similar to those for contrast. On this basis, the methods described herein can operate based on the logarithm of the blur magnitude.

[0025] The term “distortion” refers to spatial infidelity or scrambling in the mapping between locations in the input image and locations in the percept. Although distortions can simply be optical (e.g. "fish eye lens" effect), more complex distortions can result from miswiring or uncertainty in the connections between different visual field maps in the visual system. Such a scrambling is proposed as an important factor in the visual deficits seen in amblyopia. There are different approaches that can be taken to generate stimuli to explore scrambling. One approach is to have an algorithm that generates a regular stimulus, such as a set of small elements (e.g. dots) arranged in a line or a grid. This can be distorted orscrambled by shifting the position of these elements (e.g., randomly, or deterministically), with the magnitude of this shift determining the degree of distortion and the corresponding numerical quality value. Another approach is to start with a more complex image (e.g. the letters used in eye exam charts) and apply a distortion algorithm to those. For this approach, there is also a logarithmic relationship between physical scrambling of the stimulus and perceived scrambling. Therefore, although the numerical scale for each case may need to be individually validated, a general starting principle can be of a logarithmic relationship between physical stimulus and perception.

[0026] It will be appreciated that, for simplicity and clarity of illustration, where considered appropriate, reference numerals may be repeated among the figures to indicate corresponding or analogous elements or steps. In addition, numerous specific details are set forth in order to provide a thorough understanding of the exemplary embodiments described herein. However, it will be understood by those of ordinary skill in the art, that the embodiments described herein may be practiced without these specific details. In other instances, well-known methods, procedures and components have not been described in detail so as not to obscure the embodiments described herein. Furthermore, this description is not to be considered as limiting the scope of the embodiments described herein in any way but rather as merely describing the implementation of the various embodiments described herein.

[0027] One or more systems described herein may be at least partly implemented in computer programs executing on microcontrollers, computing devices or processing devices, each comprising at least one processor, a data storage system (including volatile and non-volatile memory and / or storage elements), at least one input device, and at least one output device. The terms “processing device” or “computing device” encompass computers, servers and / or specialized electronic devices which receive, process and / or transmit data. “Processing devices” and “computing device” are generally part of “systems” and include processing means, such as microcontrollers and / or microprocessors, CPUs or areimplemented on FPGAs, as examples only. For example, and without limitation, the processing device may be a programmable logic unit, a mainframe computer, server, and personal computer, cloud based program or system, laptop, personal data assistance, cellular telephone, smartphone, wearable device, tablet device, video game console, or portable video game devices. In some embodiments, a non-transitory computer-readable medium can be provided with instructions stored thereon which, when executed, cause a computing system to carry out the methods as described herein.

[0028] Software modules, or simply modules which are implied to be software, may be represented herein as any combination of flowchart elements or other elements indicating performance of process operations and / or textual description. Such modules may be executed by hardware that is expressly or implicitly shown, the hardware being adapted to (made to, designed to, or configured to) execute the modules. Moreover, it should be understood that module may include for example, but without being limitative, computer program logic, computer program instructions, software, stack, firmware, hardware circuitry or a combination thereof which provides the described capabilities.

[0029] The system and method described herein when used to measure contrast imbalance relates to an innovative tool, the Dichoptic Contrast Ordering Test (DiCOT), adapted for measuring the degree of binocular imbalance of a subject, which may include a difference in interocular suppression or binocular combination. As mentioned above, the DiCOT system and method is based on the principle that an imbalance in input strength and / or eye suppression, between the eyes of a subject will cause the apparent contrast perceived by one eye to be higher or lower than the apparent contrast perceived by the other eye, for a same contrast. The system and method can determine a binocular imbalance by finding a contrast scaling factor that equalizes the apparent contrast perceived by the two eyes. When evaluating other subjective qualities, the system and method can similarly determine a binocular imbalance by finding a scaling factor for that quality that equalizes the apparent quality perceived by the two eyes.

[0030] Referring to Figs. 1 to 4, method 10 and system 50 generally include presenting or displaying dichoptic images, e.g., cartoon fruit sprites, to a subject via a touch-enabled display which allows interacting with the subject. The subject, presented with the dichoptic images, is tasked with sequentially ranking, or sequentially selecting, the images in a given order of perceived quality, such as by highest to lowest apparent quality. In the exemplary embodiment described below, six images, grouped in a left-eye image set and a right-eye image set, are concurrently displayed on the touch-enabled display. Once a sequential ranking is performed by the subject, an estimate of the binocular imbalance is calculated along with a measure of the variability associated with the measurement informing the subject of the level of confidence to be had in the estimate. For example, a likely scaling factor indicative a relative binocular imbalance of the subject may be calculated.

[0031] The system and method described herein are preferably iterative, meaning that once the images are sequentially ranked by the subject and once an initial binocular imbalance is determined, a new stimulus condition may be selected for adjusting the relative quality of new images presented to the subject, and the new images may be generated and displayed based on the new stimulus condition. This allows for iteratively determining the binocular imbalance by updating the estimated scaling factor based on new sequential rankings.

[0032] In the embodiment of Figs.1 to 4, the subjective quality to be measured is perceived contrast. As such, the dichoptic images are presented or displayed to the subject with a quality corresponding to a luminance range defined by quality values corresponding to a set of luminance contrasts. The binocular imbalance is determined by estimating a contrast scaling factor. It will be appreciated, however, that this is for exemplary purposes only, and that the described principles can apply when measuring other qualities such as blur and distortion. For example, in such cases the dichoptic images can be displayed with a blur and / or a distortion defined by their own corresponding sets of quality values, and the estimated scaling factors can correspond to a blur scaling factor and / or a distortion scaling factor.

[0033] Turning to Fig.1, the method 10 for evaluating binocular imbalance between left and right eyes of a subject includes steps 12-22, which correspond to a first iteration, or first trial, of the method. Further iterations comprise performing step 26 and steps 14-22, described in detail below; the method 10 ends at step 24.

[0034] The method 10 includes a first step 12 of selecting an initial stimulus condition. The stimulus condition includes a left-eye contrast set and a right-eye contrast set that both comprise at least one luminance contrast as a ratio indicative of a given contrast relative to a maximum contrast possible, for example. In a preferred embodiment, the left-eye contrast set includes three left luminance contrasts, while the right-eye contrast set includes three right luminance contrasts. The initial stimulus condition may include similar luminance contrasts for both of the left-eye and right-eye contrast sets, such that a first iteration of images is presented or displayed, at step 14 described below, with left and right images having the same contrast range. For example, the initial stimulus condition may include ratios {100%; 53%; 27%} in both the left-eye and the right-eye contrast sets, each associated with one of six visual stimuli, or images, to be displayed.

[0035] At step 14, the images, including a left-eye image set and a right-eye image set, are displayed. In an embodiment, each of the left-eye and the right-eye image sets includes three images. Therefore, each image of the left-eye image set is displayed with a luminance range defined by a corresponding luminance contrast of the left-eye contrast set, and each image of the right-eye image set is displayed with a luminance range defined by a corresponding luminance contrast of the right- eye contrast set. The images are displayed concurrently in a dichoptic manner, meaning that while all the images are displayed at the same time, the images of the left-eye image set are only visible to the left eye of the subject, and the images of the right-eye image set are only visible to the right eye.

[0036] Displaying three images to each eye allows for ranking multiple images having different luminance contrasts on a single iteration of method 10. For example, at each iteration, both of the left-eye and right-eye image sets may include a first image, or high-contrast image, having corresponding luminancecontrast of 1.00, or 100%, that is therefore presented to each of the left and right eyes at a maximum contrast (i.e., a maximum quality). The luminance contrast of the other two images of each of the left-eye and right-eye image sets may vary according to the stimulus condition, selected by a selection algorithm described in more detail below. Having a high-contrast image at each iteration helps preventing any eye-based contrast adaptation or renormalisation that may occur in a subject performing a number of iterations. Furthermore, using a plurality of images to be displayed to each eye during a single iteration results in the subject implicitly making multiple comparisons of apparent contrasts between all of the images on display, thereby resulting in more information about the correct scaling, or binocular imbalance, being received from a single iteration. Yet further, by including multiple images with different contrasts in each of the left-eye and the right eye image sets, there is an implicit “correct” order in which the images for one eye should be ranked. This can be used to validate whether the subject understands the task.

[0037] However, it should be understood that in alternative embodiments, the number of images of each of the left-eye and the right-eye image sets, and correspondingly the number of quality values (such as luminance contrasts) of each of the left-eye and the right-eye sets, may be modified without departing from the present disclosure. For example, each iteration may include only one left-eye image and one right-eye image concurrently presented to the subject, whom may then be tasked with selecting the image with a higher apparent quality.

[0038] Step 16 includes receiving a sequential ranking of the images from the subject or determining that the images were sequentially ranked by the subject. For example, in the DiCOT, the subject is tasked with selecting the images in a decreasing order of apparent contrast perceived, and a sequential ranking is therefore generated. When all the images have been selected, step 20 is performed. For example, the sequential ranking, or ranking response, made by the subject may be received as a series of selections, e.g., a response of “R1, L1, R2, L2, R3, L3” to a first iteration, indicative that the subject selected the images froma highest to a lowest on-screen contrast, alternating between a right-eye image and then a left-eye image for each contrast level.

[0039] In some iterations, lower quality images may not be visible to a disfavored eye (for example an eye affected by amblyopia). Therefore, when a pre- determined number of images are sequentially ranked, the method 10 may comprise a step 18 of skipping selection of the remaining unselected images. At this step, the method randomly selects the remaining unselected images, such that the unselected images are randomly ordered, to generate a complete sequential ranking, using the selection of images made by the subject and the randomly generated selection. The images may therefore be displayed until a complete sequential ranking is made.

[0040] Once the sequential ranking is made, the method includes, at step 20, calculating, or estimating, the underlying scaling factor (r) between the two eyes, that defines the binocular imbalance in visual processing between the eyes of the subject. The contrast scaling factor can be determined as the most likely value, or likelihood, from a scaling factor likelihood distribution based on the history of sequential rankings made in step 16, in response to the different stimulus conditions determined in steps 14 and / or 26: log L( ^^^^) =∑^^log L( ^^^^|condition^^, response^^), (1)where log L( ^^^^) is log-likelihood of the ^^^^ℎscaling factor parameter value, calculated as the sum of the log-likelihoods based on the responses given to the conditions over each j of the iterations performed. These likelihoods values are obtained from pre-generated look-up tables, discussed in more detail below.

[0041] After converting back to likelihood, the value is normalized as L': L'( ^^^^) = L( ^^ ^^)⁄∑ ^^ L( ^^ ^^)(2)

[0042] In an embodiment, a 1% probability of lapsing, i.e., of a subject randomly selecting the images in an iteration, is also incorporated into the calculation at eachiteration. Without this parameter, the method would assume that humans always respond perfectly without misplacing their attention or pressing the wrong button. Of course, it should be understood that a higher or lower probability of lapsing may be incorporated without departing from the present disclosure.

[0043] In the present embodiment, and a shown in Fig.3B, the calculation of the scaling factor is based on look-up tables 53 that are generated prior to performing the method 10, for each possible stimulus condition. The look-up tables 53 store likelihood distributions of scaling factors for each possible sequential ranking associated with a stimulus condition, across a range of discrete scaling factors. A likelihood distribution is generated by calculating a probability that the scaling factor for the subject is a given discrete scaling factor given a sequential ranking, for each discrete scaling factor of a range of possible scaling factors. The probabilities are calculated for each possible arrangement of a sequential ranking of a given stimulus condition. In the DiCOT, a range of possible contrast scaling factors, e.g., a range of {-18dB; +18dB} (where 0 dB is an equal contribution from the two eyes and every 6 dB step in either direction is approximately a factor of two), is divided in discrete contrast scaling factors increments, e.g., in 0.5dB increments, and the probability of the contrast scaling factor being a given discrete contrast scaling factor is calculated for each of those increments.

[0044] In some embodiments, the look-up tables are generated through Monte- Carlo simulations. For example, the Monte-Carlo simulations can be parametrized to perform 100,000 simulations per stimulus condition and per parameter value, e.g., contrast scaling factor, which is stepped from -18 dB to +18 dB in 0.5 dB intervals.

[0045] In the simulations, the input quality of the two eyes varies according to the scaling factor, with the assumption of a linear mapping between the numerical quality value and the perceived quality, affected by a 10% internal additive noise. Varying this noise parameter has the effect of making the likelihood distributions overall broader, for large values of the noise parameter, or narrower, for small values of the noise parameter. The value of 10% is not selected on the assumptionthat it is a good estimate of the amount of noise affecting this process in human vision. Rather, it is a value which advantageously gives reasonably broad likelihood distributions without challenging the precision limit for representing and manipulating probability values used by the method. In the look-up tables 53, each column gives the relative likelihood or probability that the scaling factor is a specific discrete scaling factor, for each the 720 possible sequential rankings (6 x 5 x 4 x 3 x 2 x 1 possible rankings) of a stimulus condition.

[0046] Therefore, at step 20, a scaling factor distribution for a given iteration is selected from the look-up table of the selected stimulus condition as the likelihood distribution associated with the sequential ranking made by the subject. The scaling factor is calculated as the most likely scaling factor from the scaling factor distribution. In subsequent iterations through steps 14-22 and 26, the look-up table associated with each selected stimulus condition is used to update the scaling factor by adding the likelihood distribution associated with a given sequential ranking in the look-up table with previously determined likelihood distributions. In other words, each iteration updates an ongoing estimate of the scaling factor of the subject’s underlying binocular imbalance being one of a set of “potential” values which range from highly favoring the left eye to highly favoring the right eye, with a middle of the range being associated with a general binocular balance.

[0047] Turning to Fig. 5, when a subject completes a sequential ranking in response to the stimulus condition, the corresponding row in the relevant look-up table is selected, e.g., row 6. As mentioned above, this indicates the likelihood distribution for each of the possible quantized or discrete scaling factors, on the basis of the sequential ranking given in that iteration. The log of each value of the likelihood distribution is added to the cumulative history of all of the previous log likelihoods, to obtain an overall distribution 55, the peak of which is the most likely scaling factor.

[0048] At step 24, once a termination threshold is reached, the final most likely scaling factor, based on the overall distribution 55, is determined as the binocular imbalance between the left and right eyes of the subject. The termination thresholdmay be a maximum number of iterations, or a confidence level, for example. In a preferred embodiment of the DiCOT, the termination threshold is the first of: a 95% confidence interval width limit of 3 dB (determined by bootstrapping described in step 41) reached after a minimum of 10 iteration, or a maximum of 20 iterations when the confidence intervals do not become sufficiently narrow.

[0049] Further, bootstrapping may be used to deliver this most likely scaling factor along with a 95% confidence interval. An associated measurement error is a fundamental requirement of any clinical test that is involved with therapeutic assessment / endpoint evaluation as it allows statistical evaluations on a single case basis. The 95% confidence interval associated with the determined binocular imbalance is determined through non-parametric bootstrapping (1000 samples), randomly resampling from the individual iterations and selecting the most likely scaling factor from each bootstrapped dataset.

[0050] When, after an iteration, the termination threshold is not reached at step 22, a new stimulus condition is selected at step 26. In a preferred embodiment, the new stimulus condition is selected by an entropy-minimizing procedure that seeks to select a stimulus condition that will allow converging toward the most likely value for the scaling factor in a small number of iterations.

[0051] After each iteration, the updated likelihoods of the discrete scaling factors are used to select the stimulus condition for the next iteration. The selected stimulus condition is one that minimizes the expected posterior entropy for a sequential ranking of a next iteration.

[0052] The entropy, given all previous iterations, is defined as: H( ^^) = − ∑^^L'( ^^^^) × log L'( ^^^^), (3)

[0053] The expected posterior entropy simulates adding one more iteration, and accounting for the various possible sequential rankings that the subject is more or less likely to give based on the current best estimate of their scaling factor, or binocular imbalance. After iteration j, the expected posterior entropy following a next iteration is calculated as above.

[0054] The expected entropy resulting from the j + 1 iteration given a certain condition is the weighted average of the entropies calculated from the likelihood distributions for the different possible sequential rankings.

[0055] The probability of giving a certain sequential ranking has to take into account the relative likelihoods of the different scaling factor options: P(rank^^|condition^^+1) =∑^^L'( ^^^^) × P(rank^^| ^^^^, condition^^+1), (4) where theto a particular stimulus condition under a particular contrast scaling factor can be obtained directly from the Monte-Carlo generated look-up tables.

[0056] From the above, a normalized likelihood function that would be the outcome for each possible sequential ranking for that iteration is determined (according to Bayes’ Theorem) P(rank | ^^ , conditio ′L'( ^^^^|condition^^+1, rank^^) = ^^ ^^ n ^^+1) × ^^ ( ^^ ^^)⁄P(rank ^^|condition ^^+1).(5)

[0057] The conditional entropy of the likelihood function from that single outcome is calculated: ^^^^+1( ^^|condition^^+1, rank^^) = − ∑^^L' ( ^^^^|condition^^+1, rank^^) × log L' ( ^^^^|condition^^+1, rank^^), (6)

[0058] An optimal procedure would include the entire iteration-by-iteration history plus the predicted outcome of the next iteration to calculate the expected entropy to determine the next iteration. From the iteration likelihood, the expected entropy is the weighted sum of the conditional entropies calculated in Equation 6: ^^^^+1( ^^| ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^^+ ^^) = ∑^^^^^^+1( ^^|condition^^+1, rank^^) × P(rank^^|condition^^+1) . (7) The stimulus condition for the next iteration ( ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^^+ ^^) is therefore chosen as the one which minimize the value of Equation 7.

[0059] In some embodiments, training iterations may be performed prior to step 12. In the DiCOT, binocular images are displayed to the subject with fixed contrastswhich the subject must select in an order from high to low contrast. Selecting the images in the correct order may be a requirement to succeed the training trials, ensuring that the subject understands how to perform the task required.

[0060] In some embodiments, a new stimulus condition may be selected after each selection of one image by the subject. For example, the subject may be tasked with selecting the highest-contrast image from the images presented, and once the selection is received, a new stimulus condition is calculated, and new images are presented. Therefore, rather than waiting for a sequential ranking of the displayed images from the subject before estimating the binocular imbalance, followed by selecting new stimulus conditions, the method is adapted to “continuously” estimating and / or updating the binocular imbalance based on a selection of one image, and thereafter selecting a new stimulus condition.

[0061] Turning now to Fig. 2, the method 10 described hereinabove may be implemented by system 50. In the illustrated embodiment, the system 50 includes computing device 52 comprising a processor operatively coupled to a display and a user input. In the present embodiment, the display and user input are provided in the form of a touch-enabled display device 64, although it is appreciated that other configurations are possible. The computing device 52 may be a backend server or a cloud server, for example, communicating with the display device 64. The computing device 52 includes a look-up generation module 56 adapted to create the look-up tables 53 described herein above. The computing device 52 also includes an evaluation module 62 adapted to receive the sequential ranking from the subject and to estimate the contrast scaling factor based on above- mentioned contrast scaling factor distributions. The evaluation module 62 may also be adapted to select a new stimulus condition for subsequent iterations of method 10. The computing device 52 also includes storage 54 for storing the look-up table, the contrast scaling factor distributions, the estimated contrast scaling factor, or binocular imbalance, and instructions executable by the processor to implement modules 56, 58, 62 and / or carry out the steps of the methods as described above.

[0062] The computing device 52 further includes a User Interface (UI) module 58 adapted to manage interactions with the subject via the touch-enabled display device 64. For example, the UI module is adapted to transmit display information for displaying the images 68 on the display device 64. The display information may include the generated images, or information for generating the images such as image type, image contrast and image positioning, for example. The UI module 58 is also adapted to receive a sequential ranking from the subject through touch buttons. The UI module 58 may also be adapted to display a skip button 66 when a pre-determined number images have been selected by the subject, and a reset button (not shown) such that a subject they can reset an individual iteration.

[0063] In an embodiment, the computing device 52 is integrated with the touch- enabled display device 64 and includes and a processing device and a non- transitory medium storing processor-readable instructions storing instructions that, when read by the processing device, perform the steps of method 10 described above. For example, the computing device 52 may be a handheld personal device such as a touchpad. The display device 64 is adapted to display color anaglyphs representing dichoptic images to the subject. For example, images for one eye are rendered in red, and the other in cyan, on a white background. The subject may wear red / green “3D glasses” when using the display device 64. Therefore, through the “red lens” eye, the red images blend into the background and are not visible, and the cyan images appear dark and are visible. Conversely, through the “green lens” eye, the reverse is true. The red images are visible, and the cyan images are not.

[0064] The touch-enabled display device 64 may generate feedback when actions are taken by the subject. For example, tapping an image may generate both a visual feedback, e.g., the image is crossed-out, and an auditory feedback, e.g., a positive “click” sound.

[0065] In other embodiments, the images may be presented or displayed using other technologies to generate dichoptic images. For example, the touch-enabled display device 64 may be a lenticular display or tablet, adapted to show dichopticfull-color images without the need for glasses. Other embodiments may include display device 64 having a first display for displaying images to the subject’s left eye, and a second display for displaying images to the subject’s right eye, such as a head-mounted virtual reality, augmented reality, and / or extended reality device. In some embodiments, the display device 64 may be provided with polarized, anaglyph or active-shutter glasses configured to cooperate with the display device 64 to selectively display the at least one left image only to the left eye of the subject, and to selectively display the at least one right image only to the right eye of the subject Alternative embodiments may include a printed physical medium instead of a touch-enabled display.

[0066] In Fig. 3A, an exemplary panel displayed on the touch-enabled display device 64 includes dichoptic images where the images of the left-eye image set have contrasts of 100%, 35%, and 12% and the images of the right-eye image set contrasts of 100%, 81%, and 65%. The contrasts are determined by the stimulus condition associated with look-up table 53.

[0067] In Fig.4, both the left-eye luminance contrasts and the right-eye luminance contrasts are similar, at 100%, 75%, and 50%, the display 68 displaying the left- eye image set 69a and the right-eye image set 69b for sequential ranking by the subject. For example, this may correspond to the initial stimulus condition.

[0068] The method and system described above result in a simple, game-like iterative process, lasting as little as 2-3 minutes in some embodiments, and involving touching elements on the screen in order of their perceived visibility. This task can be accomplished by children as young as 3 years old.

[0069] This inventive subject matter has been described in terms of specific embodiments, and configurations which are intended to be exemplary only. Persons of ordinary skill in the art will appreciate, having read this disclosure, that many obvious variations, modifications and refinements may be made without departing from the inventive concepts presented herein.

Claims

CLAIMS 1. A method of evaluating a binocular imbalance between left and right eyes of a subject, the method comprising steps of: i) selecting a stimulus condition including a left-eye set comprising at least one left quality value, and a right-eye set comprising at least one right quality value, the left quality value and the right quality value being selected from a numerical scale that quantifies a perceptible quality of an image; ii) concurrently displaying a plurality of images to the subject based on the stimulus condition, the plurality of images including a left-eye image set comprising at least one image presented only to the left eye of the subject and a right-eye image set comprising at least one image presented only to the right eye of the subject, each image of the left- eye image set being displayed with the perceptible quality defined by a corresponding member of the at least one left quality value, and each image of the right-eye image set being displayed with the perceptible quality defined by a corresponding member of the at least one right quality value; iii) receiving an input from the subject sequentially ranking an apparent perceived quality of the plurality of images; and iv) calculating a scaling factor defining a relative imbalance in subjective quality between the subject’s left eye and right eye, based on the sequential ranking of the plurality of images.

2. The method of claim 1, wherein: the left-eye image set comprises a plurality of left-eye images; the right-eye image set comprises a plurality of right-eye images;the left-eye set comprises a plurality of left quality values each associated with a respective one of the plurality of left-eye images; the right-eye set comprises a plurality of right quality values each associated with a respective one of the plurality of right-eye images; and step ii) includes concurrently displaying the plurality of left-eye images each with its associated one of the plurality of left quality values, and the plurality of right-eye images each with its associated one of the plurality of right quality values.

3. The method of claim 1, wherein calculating the scaling factor comprises estimating probabilities of the sequential ranking corresponding to a plurality of potential scaling factor values, and selecting one of the plurality of potential scaling factor values having a maximum likelihood.

4. The method of claim 3, further comprising performing a plurality of iterations of steps i) to iv), wherein in each iteration a new stimulus condition is selected in step i), and wherein step iv) comprises maintaining ongoing estimates of the likelihood of the potential scaling factor values and updating the ongoing estimates after each iteration.

5. The method of claim 4, comprising performing the plurality of iterations with new stimulus conditions until either: (a) a predetermined confidence interval limit of the calculated scaling factor is reached after a predetermined minimum number of iterations; or (b) a predetermined maximum number of iterations is reached.

6. The method of claim 5, wherein in each iteration, the new stimulus condition is selected in step i) using an entropy-minimizing procedure that seeks to minimize the number of iterations required to reach the predetermined confidence interval limit of the calculated scaling factor.

7. The method of claim 3, wherein the probabilities of the sequential ranking corresponding to the plurality of the potential scaling factor values are estimated using pre-generated lookup-tables.

8. The method of claim 2, wherein one of the plurality of left images and one of the plurality of right images have a maximum quality.

9. The method of claim 2, wherein the input from the subject correspond to a ranking of a subset of the plurality of images, and wherein the method further comprises generating a complete sequential ranking of the plurality of images by randomly ranking remaining images of the plurality of images for which an input from the subject was not received.

10. The method of claim 1, wherein the perceptible quality is at least one of contrast, blur, and distortion.

11. A computing system for evaluating a binocular imbalance between left and right eyes of a subject, comprising: at least one display; at least one user input; and a processor operatively coupled to the at least one display and the at least one user input, the processor being configured to: i) select a stimulus condition including a left-eye set comprising at least one left quality value, and a right-eye set comprising at least one right quality value, the left quality value and the right quality value being selected from a numerical scale that quantifies a perceptible quality of an image; ii) concurrently display, via the at least one display, a plurality of images to the subject based on the stimulus condition, the plurality of images including a left-eye image set comprising at least oneimage presented only to the left eye of the subject and a right-eye image set comprising at least one image presented only to the right eye of the subject, each image of the left-eye image set being displayed with the perceptible quality defined by a corresponding member of the at least one left quality value, and each image of the right-eye image set being displayed with the perceptible quality defined by a corresponding member of the at least one right quality value; iii) receive an input from the subject, via the at least one user input, sequentially ranking an apparent perceived quality of the plurality of images; and iv) calculate a scaling factor defining a relative imbalance in subjective quality between the subject’s left eye and right eye, based on the sequential ranking of the plurality of images.

12. The system of claim 11, comprising a touch-enabled display implementing the at least one display and the at least one user input.

13. The system of claim 11, wherein the display comprises a lenticular display.

14. The system of claim 11, wherein the at least one display comprises a first display for displaying images of the left-eye image set to the subject’s left eye, and a second display for displaying images right-eye image set to the subject’s right eye.

15. The system of claim 11, further comprising polarized, anaglyph or active- shutter glasses configured to cooperate with the display to selectively display the images of the left-eye image set only to the left eye of the subject, and to selectively display the images of the right-eye image set only to the right eye of the subject.

16. A non-transitory computer-readable medium storing instructions thereon which, when executed by at least one processor of a computing system, cause the computing system to perform a method comprising: i) selecting a stimulus condition including a left-eye set comprising at least one left quality value, and a right-eye set comprising at least one right quality value, the left quality value and the right quality value being selected from a numerical scale that quantifies a perceptible quality of an image; ii) concurrently displaying a plurality of images to a subject based on the stimulus condition, the plurality of images including a left-eye image set comprising at least one image presented only to a left eye of the subject and a right-eye image set comprising at least one image presented only to a right eye of the subject, each image of the left- eye image set being displayed with the perceptible quality defined by a corresponding member of the at least one left quality value, and each image of the right-eye image set being displayed with the perceptible quality defined by a corresponding member of the at least one right quality value; iii) receiving an input from the subject sequentially ranking an apparent perceived quality of the plurality of images; and iv) calculating a scaling factor defining a relative imbalance in subjective quality between the subject’s left eye and right eye, based on the sequential ranking of the plurality of images.

Images