Response estimation to efficiently capture dynamic response gain changes in multifocal responses

Abstract

A system and method are disclosed for estimating the response of multiple portions of the visual field of one or both eyes to simultaneously displayed multifocal stimuli from a recording of the response of one or both pupils. The collection of multifocal stimuli is controlled by separate pseudorandom sequences, one sequence for each stimulated visual field region. Given the pseudorandom nature of the stimuli, the density of the stimuli varies over time. The system and method includes estimating, for each component portion of the visual field, i.e., for each sequence, both a temporal impulse response and a gain kernel that characterizes the dynamic changes in the pupil response driven by short-term variations in the overall stimulus density. While a gain kernel for each stimulus region can be estimated, it is demonstrated that a smaller number of kernels can be estimated by grouping the stimuli into symbolic stimulus groups and estimating only the gain kernel for each group. Certain symbolic stimulus groups, especially those capturing switch eye configurations, have been shown to be very efficient. The ability to estimate more reliable responses with a relatively small number of extra gain kernel coefficients is a means by which less data needs to be collected and thus the test period can be shortened or the response can be more accurately estimated over a period of time.

Description

[Technical field]

[0001] The present invention relates generally to methods and systems for simultaneously assessing functional aspects of many portions of a human or animal's visual field. More specifically, the present invention relates to determining how a history of stimuli displayed over approximately one second affects responses to subsequent stimuli, particularly when the balance of simultaneously displayed stimuli switches from one eye to the other. [Background technology]

[0002] Any discussion of background throughout this specification should not be taken as an admission that such background is prior art or that such background is widely known or forms part of the common general knowledge in the field.

[0003] Multifocal stimulation of the nervous system refers to the display of different stimulus sequences to different parts of the sensory organ over time. For example, four buzzers can be attached to different parts of the skin of the forearm. Four different time sequences can then determine when each buzzer stimulates the skin at its location. This is a multifocal display of simultaneous tactical stimulation across the entire sensory field of the forearm. In the case of vision, the visual field of one or both eyes is divided into multiple regions, and the visual properties of each region (e.g., color, contrast, brightness, or texture) are independently adjusted over time according to a set of time sequences, one for each region of the visual field or visual field. The stimulus sequence is a record of the stimulation history of each stimulated region. In both examples, some recording means is used to record the pooled response of the nervous system to the multifocal stimulation.

[0004] The classical recording method is to record the evoked electrical response of the nervous system to a stimulus. In the touch and vision examples just given, the brain's electrical response can be recorded by placing electrodes on the scalp to capture the brain's response to the stimulus. These recordings include a version of the sum of all responses to multiple stimuli presented simultaneously. If only one area is stimulated at a time, the brain activity reflects the response of only one area, and the response can be estimated by averaging with respect to the onset of each stimulus in the sequence. If multiple stimuli are presented simultaneously, the problem arises of how to estimate the separate average response to the stimuli delivered to each sensory area. This can be achieved if the time sequences driving the stimulus presentation in each area are statistically independent, i.e., substantially uncorrelated in time. The average response can then be obtained in all stimulated areas by methods such as cross-correlation between the evoked response and the stimulus history, or some form of multiple regression between the evoked response and the stimulus history, i.e., recording of the stimulus sequences.

[0005] To date, patents related to multifocal methods have concerned spatiotemporal tuning of specific categories of stimulus sequences, or patterns of sequences. Early examples included approximate orthogonal pseudorandom sequences designed to perform response estimation by cross-correlation efficiency (E.E. Sutter, U.S. Pat. No. 6,086,206). These methods did not suggest that a particular temporal rate or density of stimuli was optimal. Later, inventors of the disclosed inventions obtained patents on methods for generating stimuli with specific temporal or spatial densities that were unexpectedly found to generate many times larger and more reliable responses (T. Maddes and A.C. James, U.S. Pat. No. 7,006,863 and U.S. Pat. No. 8,583,223). These patents showed that the average gain, i.e., the average responsiveness of the system, was higher when stimuli were separated in time or space. Thus, for example, U.S. Pat. No. 7,006,863 focused on the idea that "appropriate design of stimulus sequences may enable the nervous system to generate larger and more reliable responses." U.S. Patent No. 8,583,223 focuses on an effect known as "lateral masking," whereby spatially adjacent stimuli can suppress each other's responses; therefore, to minimize the deleterious effects of lateral masking, the patent states that "response size and reliability can be improved by ensuring that simultaneously displayed stimuli are separated in space; such stimuli are thus said to be spatially sparse."

[0006] Investigations then moved from evoked electrical responses to using pupillary responses by recording changes in pupil diameter over time. Unlike evoked electrical responses, imaging and video monitoring of the pupil allows for non-contact, multifocal assessment. Due to the connectivity of the neural systems driving the pupils, each pupil reports a version of the sum of the activity of the two retinas. Thus, response estimates from a single pupil recording can produce an average response to stimuli presented simultaneously to multiple stimulus fields in both eyes. Thus, recording both pupils yields two sets of responses for each eye. The move to pupil recordings resulted in two pupil-specific patents.

[0007] The first of these patents involved the so-called luminance balance method: T. Maddes and A.C. James; US Patent No. 8,583,223. This involved keeping the average gain, i.e., responsiveness, approximately equally high across the different test areas. The average gain refers to the average response per unit stimulus intensity obtained across a set of visual field areas stimulated simultaneously during the test period. This technique is based on the discovery that some hyperresponsive areas in the superior temporal visual field respond more naturally than other areas when the same intensity of stimulus is presented to each area. These hyperresponsive areas set the average response gain of the pupillary system lower, decreasing the relative responsiveness of other areas. By reducing the stimulus intensity delivered to the naturally hypersensitive areas, the average gain throughout the test period is maintained higher, thus increasing the response of the naturally less sensitive visual field areas. By inverting the static nonlinear stimulus-response function, the stimulus intensity was balanced across the visual field. Thus, U.S. Patent No. 8,583,223 indicates that "reducing the luminance of the stimuli displayed in the more responsive areas of the visual field reduces their contribution to the pooled drive signal for the (pupil), thereby increasing the absolute response magnitude of the normally less responsive areas."

[0008] The second pupil-specific patent involved so-called clustered-volleys, T. Maddes, CF. Karl and AC. James; US Pat. No. 9,848,771. This involved the discovery that even for luminance-balanced stimuli, there is an additional benefit to displaying stimuli in spatially adjacent clusters within a subset of regions of the visual field. In a non-limiting demonstration, these were chosen to be clusters within the left vs. right, or upper vs. lower halves of the visual field. This was surprisingly the opposite of what was predicted by US Pat. No. 8,583,223. This method kept the number of stimuli displayed at any one time step of the test relatively constant, maintaining a reasonably balanced average gain. This did not happen with previously patented stimulation methods, where the average number of stimuli per time step was more freely varied.

[0009] In the pupillary system, a series of inputs arising on the two retinas proceed to the left and right pretectal olivary nuclei (PON), which then proceed to the left and right Edinger-Westphal nuclei (EWN), each of which sends nerves to the left or right iris to generate the pupil's light response. Each PON also gets visual input from its left or right visual cortex. Each EWN in turn receives input from both PONs, and this is how each pupil responds to the sum of activity on both retinas, as previously described. A series of elaborate experiments, displayed in U.S. Patent No. 9,848,771, concluded that there exists a "gain control mechanism on the EWN or its subsequent pathway to the pupil," and that "this gain control system tends to reduce the pupillary response less when multiple visual stimuli are presented in a cluster volley of spatially adjacent stimuli, compared to previous methods such as stimuli that are sparse in time or space." The patent further presented findings that "suggested that EWN gain control involves a feedback mechanism that is too slow to attenuate the stimuli when presented en masse." Thus, US Patent No. 9,848,771 had two core ideas: that the gain control mechanism that regulates the average gain across the visual field occurs at the level of the EWN, thereby completing the combination of inputs to the two eyes and the visual cortex, and that this mechanism is too slow to be affected by sequences of juxtaposed volleys, which in US Patent No. 9,848,771 occur at a rate of once every quarter of a second.

[0010] It was found that interactions between certain successive volleys can have a significant effect on short-term gain, whereas previous publications such as U.S. Patent No. 9,848,771 showed that the results of all volleys from both retina ensembles are pooled in the EWN, resulting in a gain adjustment that leaves no room for special contributions from each eye.

[0011] All of the above methods specify different types of multifocal stimulus sequences, or different spatiotemporal adjustments of the sequences, whereby all subjects receive the same, optimal stimulus. Similarly, the same response estimation method was used for all subjects and all patented stimulus sequence variants. Focusing on mean gain and fixed response estimation, the methods did not show any advantage in assuming different dynamic gain characteristics for individuals, or in tailoring the response estimation method to the individual or pupil. In fact, the patents provide evidence that such dynamic effects do not occur, and, as mentioned above, U.S. Patent No. 9,848,771 goes so far as to state that the system that modulates responsiveness in the EWN is too slow to react to a random volley of stimuli presented at 0.25 second intervals.

[0012] All response estimation methods used in multifocal testing methods used so far assume that, if there are nonlinearities in the system under test, they are static; i.e., their nonlinearities do not change during the course of the test. As mentioned before, classical response estimation methods involve cross-correlation between the stimulus history, i.e., the set of test sequences that control the presentation in each area, and the response of the nervous system, including a version of the sum of all responses, the so-called pooled response. This method was first presented in 1965 by Y. W. Lee and M. Schätzen in the paper "Measuring the Wiener Kernel of Nonlinear Systems by Cross-Correlation", which was published in the International Journal of Control, Vol. 2, pp. 237-254. This method is still commonly used today. The Wiener kernel mentioned in the title of this paper is also called linear and nonlinear weighting functions. The nonlinear kernel captures the effect of additive nonlinear interactions between stimuli. The linear weighting function is also known as the temporal impulse response to repeated short-duration impulse stimuli. We describe methods for estimating responses in systems with dynamic split gain control where there are rapid changes in gain depending on stimulus history, and where these dynamic changes may vary from person to person and even from pupil to pupil. These nonlinearities are not captured by standard extensions such as the Wiener kernel.

[0013] Later, more flexible response estimation methods based on multiple linear regression were presented in 2005 in the paper "Effects of Temporal Sparsity and Binocular Display on Multifocal Visual Evoked Potentials" by A.C. James, R. Lussekite, and T. Mades (Visual Neuroscience, vol. 22, pp. 45-54). Although these methods were more flexible, they were only suitable for systems in which the Wiener nonlinearity was unchanged. Unlike cross-correlation, regression response estimation allows for adding features to the response estimation model to take into account aspects of the recording or response therein. For example, when recording evoked electrical responses, it is not uncommon for electrodes to pick up a signal (hum) from the AC mains power supply. Thus, a term can be added to the regression model to split the variance in the recording into separate components for the physiological response and the mains power frequency, and thus the neural response can be estimated simultaneously with the hum from the mains power supply. Biological responses often contain some degree of autocorrelation. Response estimation based on cross-correlation to multifocal stimulation cannot take into account the autocorrelation. As a result, the variance due to the autocorrelation is misassigned to the estimate and appears as noise. The responses to multifocal stimuli and an autoregressive model can be estimated simultaneously as a kind of noise term, leading to improved fitted estimates.

[0014] Another advantage of the regression framework, as demonstrated by James et al. 2005 (above), is that it allows the determination of standard errors (SEs) in each of the estimated coefficients. Thus, if one has 88 stimulus sequences that control the presentation of brief stimuli at 44 locations in the visual field of each eye, in response estimation one extracts the average size of the response to presentation at the 88 locations, one for each response, and also extracts the 88 SEs. In contrast, classical cross-correlation methods do not provide an estimate of the error. In principle, the regression frame can be extended to an iterative form of response estimation that can fit nonlinear coefficients.

[0015] Mathematically, estimating more coefficients from the same amount of data means fewer data points in the pupillography per estimated coefficient, and therefore a larger standard error in each estimated coefficient, i.e., a worse estimate. This process is like estimating the mean of a data set, i.e., a single coefficient and its standard error (SE). By definition, the SE is the standard deviation divided by the square root of the number of data points averaged. Thus, the fewer the number of points averaged, the larger the SE. Thus, as with any statistical estimation process, it is undesirable to try to estimate too much for the number of available data points in the pupillography. In principle, a multifocal test could be made longer to collect more data, but this may result in the test being longer than is acceptable or desirable.

[0016] In a more typical test, 44 regions of the visual field of each eye are tested, and both eyes are tested simultaneously. The mean response amplitude for each region and the time to peak response to stimulus presentation in each region are estimated. This means that, at a minimum, two coefficients for the 88 regions / pupillograms, or 176 coefficients, have been estimated. If two more coefficients per region were added to account for changes in the system nonlinearity in that region, the total would be 2 × 176 = 352. Even if this were possible, the standard error would increase by about the square root of 2, i.e., more than 40%. Any strategy for estimating dynamic nonlinearities must therefore minimize the number of extra coefficients to be estimated, and the measures such as variance that are taken into account (model R 2 Strategies are needed to achieve improved performance based on model improvements through the use of the fit statistic (often called goodness-of-fit statistics).

[0017] As mentioned above, the regression framework can be extended to an iterative form of response estimation that can fit nonlinear coefficients. We have introduced a family of such iterative response estimation methods. The objectives of these were to (1) include the possibility of rapid fluctuations in responsiveness, i.e., dynamic gain, (2) determine whether such fluctuations are due to recent changes in the stimulation history and the estimation of coefficients for such changes in the form of so-called gain kernels, and (3) determine whether response estimation from pupil responses to multifocal stimuli can be improved by efficiently estimating only a small subset of the many possible gain kernel coefficients, thus reducing the need for large amounts of data. Note that the normal temporal impulse response for each stimulus region is estimated in the same process, such that the estimation of the gain kernel improves the estimation of the regional temporal impulse response. Another question is whether it is possible or desirable to estimate individual gain kernels for each subject or pupil, or whether it is possible or desirable to fix the value of the gain kernel for each person, which may depend on factors such as gender and age.

[0018] Using these new methods, a series of discoveries have been made that improve the accuracy of local temporal impulse response estimation: these discoveries demonstrate for the first time that rapid stimulus history-dependent changes in responsiveness should be taken into account for the purposes of response estimation, and that these can be reduced to a surprisingly small set of coefficients when including the rapid changes associated with interocular stimulus switching.

[0019] The overall objective of these methods is to efficiently assess the extent and function of the visual field of a person or animal, in other words to generate data about those visual fields. An example of an animal is a race horse. Horses are subject to livestock farming, and as such, certain horses may have more or less suitable visual field extents to be successful race horses. If the subject is a human, the visual field data may be used in combination with other data to determine the person's suitability for a particular occupation, or to determine whether they should be allowed to drive a particular vehicle. The other data may be metabolic or physical data of the person. The visual field data may also be used to monitor a person's condition. For example, a doctor may diagnose a person as having diabetes using a blood glucose test. It may be useful for the doctor to monitor a patient's visual field data over time for any changes due to diabetes or complications such as cataracts that are not associated with diabetes. Finally, a medical professional may use the combination of visual field data and other data to diagnose a person as having a particular eye or brain disease, which may be diagnosed in the opinion of the doctor or other medical professional. In clinical practice, monitoring the stability of treatment outcomes is likely to be a more common use than as an aid to diagnosis, given that once an individual has been diagnosed, they may be monitored multiple times throughout their life. Summary of the Invention

[0020] An embodiment of the present invention reduces a rich set of multifocal stimulus variations to a much smaller set of symbolic stimulus groups, and then fits an individual gain kernel for each of these groups that summarizes the dynamic changes in the pupil response gain in parallel with the response measurement system estimating (i.e., measuring) the temporal impulse response for each multifocal stimulus region. That is, the term "gain kernel" should be interpreted as measuring the dynamic changes in the pupil response gain. A set of symbolic stimulus groups with specific characteristics has been shown to be more useful for assessing visual fields. The gain kernel describes how the last few time steps of the stimulus history affect the responsiveness of the pupil system to a current stimulus selected from a particular symbolic stimulus group. In characterizing aspects of the pupil nervous system, these stimulus group-dependent gain kernels are an improvement of the response estimation (i.e., measurement) method for multifocal visual field testing using pupil recordings. To illustrate these steps, a non-limiting example of a set of cluster volley stimuli is decomposed into symbolic stimulus groups and incorporated into the response estimation (i.e., measurement) is shown to provide a surprising improvement over the original cluster volley method with the addition of surprisingly few additional coefficients. Stimulation methods other than the cluster volley method may also be chosen. [Brief description of the drawings]

[0021] Method, apparatus and system configurations are illustrated, by way of example only, with reference to the accompanying drawings, in which:

[0022] TIFF2025507979000002.tif237164TIFF2025507979000003.tif237164TIFF2025507979000004.tif237164TIFF2025507979000005.tif209164 DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0023] Figure 1 shows an array of 44 multifocal stimuli presented to the human or animal eye. Each stimulus (i.e., stimulus element) is a diamond. Here, only the outer edge of the stimulus (i.e., stimulus element) is shown as a black outline. In this particular, non-limiting demonstration, all parts of the diamond-shaped stimulus area are bright yellow and are briefly displayed on an unchanging dimmer background. During testing, the subject stares straight into the center of the stimulus array and fixates their eyes on a small red cross located there. In the case of animals, they may be anesthetized. At the start of the test, no stimuli are displayed, only the background. Once the test begins, the stimulus sequence determines when a particular stimulus will appear in its place. Each yellow stimulus is displayed for 33 ms whenever the stimulus sequence calls for its display. Over the course of the test, a particular region has a low probability of appearing in a pseudo-random manner. This randomness is necessary for the multifocal method to work, as it keeps the presentation of a stimulus in one region statistically independent from the stimuli displayed in other regions. That is, at any time step, each individual stimulus element has approximately a 50% probability of appearing. The stimuli are arranged in five rings, shown in Figure 1A and 1B. The overlapping regions are not displayed together, but the way in which they overlap when displayed simultaneously is shown in Figure 1C. Examples of left and right hemifield sets, or families, of either the three rings in Figure 1A, or the two rings in Figure 1B, are shown in Figure 1D. As the axis labels indicate, the array of 44 stimuli spans 30 degrees of the visual field from a central fixation point.

[0024] In practice, two arrays of stimuli may be displayed simultaneously, i.e., 44 for each eye. Figure 2 shows that these are displayed simultaneously to two eyes (200A, 200B) using a non-limiting stereo configuration. Figure 2 shows a computer system 201 with a computer 203. A light source 205, for example an infrared light source, is controlled by the computer 203. The subject's eyes are illuminated with infrared light from the light source 205 via a pair of cold mirrors (207A, 207B), which are transparent to infrared light. The cold mirrors function as normal for visible light, and using them together with objective lenses (209A, 209B) allow each eye to see one of two display systems (211A, 211B), each display system displaying a multifocal array of stimuli at the pupil of each eye, for example as shown in Figure 1. An opaque wall (213) separates the left and right halves of the device so that the eyes cannot see the opposite side. The display systems (211A, 211B) also include visual cues that allow the subject to automatically binocularly fuse the images displayed on the two display systems. Thus, the subject is not aware that he is looking at two images, one displayed on each of the two display systems, but instead perceives that he is looking at one image. Two sensors (215A, 215B), for example in the form of infrared video cameras, capture moving images of the pupil, and the computer 203 of the computer system 201 continuously measures the pupil diameter to provide a temporal record of the pupil response to the multifocal stimuli. The computer 203 of the computer system 201 uses the 88 stimulus histories, i.e., the display sequences used, and the pupil responses, to implement a response estimation (or calculation) method to estimate (or measure) the temporal impulse response for each stimulus sequence and set of gain kernels.

[0025] FIG. 3A shows a set of 176 estimated temporal impulse response waveforms, 44 from each eye and pupil. Downward deflection of the waveforms indicates a decrease in pupil diameter. FIG. 3B shows the peak amplitude of the constriction and the time to that peak constriction, i.e., the time to peak, estimated for each impulse response. The time to peak provides a measure of the lag of the response. In traditional visual field testing, sensitivity is estimated by scaling the stimulus on a decibel scale, and the μm constrictions shown here are later converted to decibel amplitude to obtain sensitivity.

[0026] FIG. 4 shows how 88 stimulus sequences were combined to display stimuli in a random cluster volley in the specific, non-limiting example that produced the response in FIG. 3. Time in seconds proceeds from top to bottom in each of columns 1-3. There is a pair of panels in each row of columns. Each pair of panels displays a stimulus that will be displayed to either the left or right eye (column headings) in a manner as shown in FIG. 2. At 0.25 s intervals, some of the stimuli in a hemifield family are shown in bright yellow for 33 ms. The display stimulus area is shown here in gray. This figure shows that approximately half of the possible areas of the hemifield family are not shown, as shown in FIG. 1D, with the dotted borders indicating where other stimuli from each hemifield family appeared. Thus, at the start of the test, at time 0 (A), six members of the left hemifield stimulus family are potentially activated for the left eye. At 0.25 s, approximately half of the possible areas in the right hemifield family are active (three in this case) and will be displayed to the left eye. At 0.5 s, a new volley of left hemifield stimuli is presented to the right eye. This continues over time until 1.75 s, when one cycle of stimuli is completed. The sequence of hemifield families is then repeated every 2 s in a round-robin fashion, as shown in rows I-X of cycles 2 and 3, but at each time step, there is a 50% chance that any member region of the family will be active, i.e., displayed. This randomized presentation of individual regions keeps the stimulus sequences statistically independent, allowing multifocal response estimation methods to extract responses from pupillography. In this example, the cycle is repeated until all stimulus regions have been presented 96 times, requiring a total of 6 min. In some cases, other sets of families are used (see below), with different numbers of iterations per region. Different numbers or colors of stimuli can also be used.

[0027] However, continuing with the current example stimuli, we see that the set of hemifield families can be assigned to different kinds of symbolic stimulus groups. The purpose of these groups is to further reduce the number of estimated gain kernels from the complete set of possible gain kernels, one for each hemifield family. Each family is assigned to a specific group, which have a common gain kernel. Thus, the number of coefficients that need to be estimated is reduced. However, there is a risk that too many or the wrong families are assigned to a specific symbolic analysis group, which results in a poor overall estimate. That is, a wrong group assignment may not capture important features of the dynamics of the pupillary nervous system. Efficient and advantageous symbolic groups have been discovered by experimentation.

[0028] Figure 5 shows three possible symbolic group assignment methods. In this figure, each column represents a different group assignment strategy that is repeated for each cycle of the full test sequence. In the left column 1 (including rows A-H), all hemifield families are black, indicating that they are all treated as a single group for response estimation. That is, a single gain kernel (i.e., set of gain kernel coefficients) is estimated for all of them, and thus the method is referred to as the common gain. In the center column 2, each hemifield family is black, and one of six grayscales from dark to light or white indicates that each hemifield family is treated as one of eight groups for response estimation. That is, a different gain kernel is estimated for each of the eight symbolic groups, and thus the method is referred to as the full hemifield method. In column 3, the hemifield families are assigned to one of two groups in the order ABBA, ABBA, etc. for the duration of the stimulus. In this figure and in Figure 6, the vertical axis labels of each row reflect the group assignments: Group 1, Group 2, etc. These assignments are maintained for all cycles of stimulation.

[0029] Figure 6 shows three alternative assignments of symbolic groups. In column 4 (A-H), there are four symbolic stimulus groups, assigned 1-4 for the three-ring stimuli in columns A-D, and then repeated for the similar two-ring stimuli in columns E-H. The assignment is therefore called the half-hemifield method, regardless of whether the hemifield family has two or three rings. The middle column 5 shows the assignment to two groups (dark gray and white, and vertical axis labels) corresponding to the stimulated eye. The assignment is therefore called the within-eye method. In column 6 on the right, the hemifield family is assigned to two groups (shades of gray or white, and vertical axis labels). The assignment always switches to the similar hemifield in the contralateral eye, and therefore this assignment to symbolic stimulus groups is called the switch-eye method.

[0030] All the previous examples had families that were collections of stimuli from either the left or right hemifield. As shown in Figure 7, other kinds of families can also be used. Each column in Figure 7 shows a different kind of family that is repeated in a round-robin fashion as in Figure 4. As in Figure 4, different stimuli within each family are randomly activated (gray areas). The leftmost column, variant 1 (rows A-H), uses families placed in the lower and upper hemifields instead of the left and right. The middle column, variant 2 (rows I-P), shows families with alternating pairs of quadrants. At this point, the family consists of a regular subset of the stimulus areas. The third column in Figure 7, variant 3 (rows Q-X), is similar to the middle variant 2, but in variant 3, sporadic extra areas are displayed. To clarify which are the sporadic extra areas, they are displayed in light gray. In practice, they may be the same or similar brightness and color as the other stimuli. Variant 3 shows that irregular families can exist, where various degrees of deviation from the regular pattern of the family are allowed. It will be apparent that the rows of FIG. 7 can be assigned to symbolic stimulus groups such as those of FIGS. 5 and 6 or other groups.

[0031] The examples show that the described method (and associated system) reduces the number of gain kernels that need to be estimated while at the same time improving the accuracy of the impulse responses obtained from each stimulation region that are estimated along with the gain kernel.

[0032] The embodiments of the present invention have been developed primarily for use as a method and system for quantifying visual field by estimating the response of the visual nervous system from a recording of pupil size over time obtained in response to multifocal stimuli. The embodiments include a method and system for quantifying stimulus-dependent dynamic changes to the pupil system, i.e., dynamic changes in pupil response gain. The dynamic changes in pupil response gain, measured by a small set of coefficients, are referred to in this document as "gain kernels". This is a departure from estimating only linear and nonlinear weighting functions to characterize systems that include nonlinearities. While in principle there is one gain kernel for all stimulus sequences, they are instead estimated for small symbolic groups of stimulus sequences. Thus, if 88 stimulus sequences are divided into 8 groups, only 8 gain kernels are estimated instead of 88. If the stimuli are grouped together by two symbolic stimulus groups, only two gain kernels need to be estimated. The estimation of gain kernels for a limited number of symbolic stimulus groups has never been considered or attempted before. Apart from the estimation of gain kernels, the embodiments are based on the discovery that certain symbolic stimulus groups are surprisingly advantageous.

[0033] Above, some characteristics of certain non-limiting types of multifocal stimuli are described. These displays display stimuli displayed pseudo-randomly in 44 regions of the visual field of each of the two eyes, as shown in Figures 1 and 2. As shown in Figure 3, the stimuli are presented in a so-called cluster volley. In this manner, one of eight groups of stimuli distributed in the left or right hemifield of the left or right eye is selected for display at each 0.25 second interval. The sequence of eight hemifield families is repeated in a round-robin fashion. Figure 3 shows three cycles of this iterative process. In any one 0.25 second time step, there is a 50% chance that each stimulus region of the hemifield family is active, i.e., that the stimulus region appears as bright yellow within its diamond-shaped boundary against a darker background. The total stimulus duration is 383 seconds, over which period each of the 88 stimulus regions is activated 96 times. Once displayed, each stimulus remains bright (active) for 1 / 30th of a second (33 ms). The activation of each of the 88 regions is controlled by 88 stimulus sequences, also called stimulus sequences or stimulus trains. These consist of 0s and 1s, as shown in Figure 8A. When the sequence changes from 0 to 1, a stimulus is activated, i.e. it is displayed at 33 ms. The interval between the time steps of the stimulus sequences, i.e. their temporal resolution, is 1 / 30 s. A volley of stimuli appears on average every 0.25 s.

[0034] The pupil response is a continuous record of pupil diameter captured over time in synchrony with the stimulus presentation. The pupil diameter responds to the sum of all activity generated by the stimuli presented to the two eyes. The pupil diameter contains this information as it is transmitted to the iris from the pretectal olivary nucleus (PON), which receives input from the extrastriate visual cortex and the retina. The two PONs each provide information to both Edinger-Westphal nuclei, each of which innervates one iris, from both eyes, and therefore each iris responds to stimuli on both retinas. The stimulus sequence and pupil light response recordings are then submitted to a regression response estimation method or system to extract the response of the visual system to each of the 88 stimulus regions. Because there are two pupil recordings, one from each eye, the response estimation process is repeated twice to yield a total of 176 estimated responses. Figure 3A shows the set of 176 linear weighting functions, or average temporal impulse responses, that are the result of a standard regression response estimation process.

[0035] However, recent studies have provided evidence that the pupillary system does not remain linear, and therefore incorporated into the response estimation process is a method for capturing the nonlinear dynamics in the form of a gain kernel. Figure 8 shows steps in an exemplary process. Figure 8A shows the first 50 seconds of the 88 stimulus sequences used to generate the responses of Figure 3. There is one sequence per stimulus region, and 44 per eye as in Figures 1 and 4. The horizontal lines in Figure 8A are sequences of 0, and the occasional upward-pointing checkmark is a 1 indicating that a stimulus should be displayed. In this non-limiting example, the stimulus sequences are displayed in blocks of eight different grey tones. These represent the eight symbolic stimulus groups in column 2 of Figure 5. Strictly speaking, groups 1-8 in column 2 of Figure 5 are at the top of the page in Figure 8, with 12 regions per hemifield family at the bottom and 10 regions per hemifield family at the top.

[0036] TIFF2025507979000006.tif117162

[0037] TIFF2025507979000007.tif117162

[0038] TIFF2025507979000008.tif105162

[0039] TIFF2025507979000009.tif30162

[0040] TIFF2025507979000010.tif49162

[0041] TIFF2025507979000011.tif51162

[0042] TIFF2025507979000012.tif50162

[0043] TIFF2025507979000013.tif67162

[0044] TIFF2025507979000014.tif66162

[0045] TIFF2025507979000015.tif108162

[0046] TIFF2025507979000016.tif77162

[0047] TIFF2025507979000017.tif41162

[0048] TIFF2025507979000018.tif39162

[0049] The regression framework can be summarized as follows: The class of standard methods for measuring, or in statistical terms "estimating", the impulse response, i.e., basis function, for each stimulated visual field region can be summarized as follows. Basis function estimate = function(pupillary response record, stimulus record)(a)

[0050] Equation (a) can be summarized as follows: The estimate of the basis function characterizing the average response to the stimulus and its scale are a function of the pupillography and the stimulus recordings. Typically, one or two time-varying pupillary response recordings from one subject are obtained in one recording session, resulting in several stimulus recordings describing the onset and duration of the stimulus for each tested portion of the visual field in one or both eyes, as shown in Figure 3A. There may be more than one basis function per tested visual field region, in which case the average response per region will be a linear or nonlinear combination of the basis functions per region. The measurement of the basis functions can be done by regression methods or cross-correlation. In the regression framework, the basis function estimation process can be improved by including other terms in the equation. Basis function estimate = function(pupillary response record, stimulus record, other terms)(b)

[0051] TIFF2025507979000019.tif87163

[0052] TIFF2025507979000020.tif77163

[0053] TIFF2025507979000021.tif50163

[0054] If a single temporal component with 61 time steps is chosen, then in total there are 61 (time steps) + 88 (area) = 149 parameters, fewer than the log-normal parametric case with 177. For the two components there are 298 basis function parameters. Then, as shown in Inverso, S.A., Goh, X.L., Henriksson, L., Vanni, S., and James, A.C., 2016, "From Evoked Potentials to Cortical Currents: Decomposition of V1 and V2 Components with Retinotopy-Constrained Source Estimation Without fMRI. Human Brain Mapping," Vol. 37, pp. 1696-1709, the bilinear, gain, noise and autoregressive parameters can be estimated simultaneously using the Gauss-Newton method.

[0055] TIFF2025507979000022.tif78163

[0056] TIFF2025507979000023.tif47163

[0057] [Evidence 1] To provide a data set to demonstrate the steps described herein, 94 normal controls (43 males) aged 49.6±19.5 years (mean±standard deviation), ranging from 18 to 91 years, and 40 subjects with glaucoma in one or both eyes aged 67.2±8.60 years, ranging from 50 to 83 years, were tested. Glaucoma causes localized damage to patches of the retina, resulting in changes in sensitivity and lag that can be characterized by creating a map of visual function provided by the present disclosure.

[0058] Each individual was tested twice, approximately 2 weeks apart, with three variants of cluster volley multifocal stimuli (Figure 3). For simplicity, we refer to the three test variants as P129, P130, and P131. The stimuli followed the 44 areas / eye format of Figure 1. P131 differed in that the array of stimuli was isomorphically reduced by a factor of two. Thus, there are 44 scale stimuli per eye, although the P131 stimuli sample a 4 times smaller area of ​​the visual field. The P129 and P131 stimuli were generated using a dim 10 cdm -2 The P130 stimulus was bright yellow on a background of 10 cd / m 2 The P129 stimulus was bright green on a red background. The stimulus intensity was varied to achieve luminance balance. The P129 stimulus ranged from 73 to 150 cd / m 2 , P130 is 65~150cd / m 2 , P131 is 134~288cd / m 2 The individual test areas were presented pseudorandomly such that each area presented a 33 ms stimulus with an average interval of 4 seconds. The test duration was just over 6.3 minutes, and each area was presented 96 times. In all tests, both eyes were tested simultaneously using an apparatus similar to that shown in Figure 2.

[0059] During the test, pupil diameter was recorded every 1 / 60th of a second, synchronized with the presentation of multifocal stimuli on two displays, one for each eye. The data were then downsampled to a time step of 1 / 30th of a second. To make the test manageable for subjects, each 6.3-minute test was divided into nine segments of approximately 42 seconds each (Equation 6). Subjects were presented with segments separated by a break of at least 7 seconds, during which they were free to blink. Parts of pupil recordings involving blinking were removed. If more than 15% of a segment was lost due to blinking, the segment was repeated. This only occurred in approximately 1 segment out of 200 segments.

[0060] TIFF2025507979000024.tif66163

[0061] TIFF2025507979000025.tif134163

[0062] Table 1 shows the results of fitting an additive linear mixed-effects model to the gain kernel lag coefficients obtained from the response estimation, with four symbolic stimulus groups defined as in column 4 of Figure 6, with the gain kernel for each group having three lag coefficients, indicating the gain weights for each 0.25 s interstimulus interval. Thus, the response estimation model contained 12 gain kernel lag coefficients. The stimuli were P129 and the subjects were a group of 94 normal controls. Because there were two pupils and two repetitions, the data set contained 94 × 2 × 2 = 376 sets of 12 gain kernel lag coefficients. The linear mixed-effects model included factors for each of the 12 kernel lag coefficients and 12 factors for lag by female. Effects for age were also fitted, but these were not significant and therefore are not shown. Table 1 summarizes the results. The model F-statistic was 179, with an adjusted R 2 The value was 0.698.

[0063] The intercept, 0.350, is the average kernel lag coefficient across 12 recordings per pupil recording for males. Because this is an additive model, the next 11 rows, Gain Kernel Coef2 through Gain Kernel Coef12, show the differences from the intercept and the significance of those differences. Many of the p-values ​​are very small and are shown as 0.000. Of those that show p=0.000, the lowest value is p<8×10 -10 For these rows, the relative significance of the differences is best understood by examining the t-statistics (column t-Stat). The effect on women was less significant but could not be ruled out. The gain kernel coefficients across the 376 data sets were highly consistent for this four-group, three-lag gain kernel model, with a high model R of 0.698. 2 It was.

[0064] Table 2 shows the R of other symbolic group / lag coefficient response estimation models calculated for the 94 normal control subjects. 2 Here, since data for P129-P131 are included, there were 376 sets of gain coefficients for each of P129-P131. In fact, the data here differs in one respect from that which created Table 1. Here, pairs of repeated data for each pupil were subjected to sampling using a bootstrap process with replacement. In this process, 64 synthetic data sets were created from each pair by random sampling with replacement of 9 segments of each data set. Responses were estimated for these 64 data sets. The idea of ​​these bootstrap cross-validation methods in statistics is to obtain results that more accurately reflect the population average. Thus, estimates were made for responses for 376 x 64 = 24,064 data sets for each of P129-P131. Subsequently, the median of the 64 for each of the synthetic data sets was taken, and the more robust cross-validated data was submitted to the linear mixed-effects model for Table 1.

[0065] [Table 1] JPEG2025507979000027.jpg161166

[0066] The greyscale background of the cells in Table 2 indicates which response estimation method has a large R 2 As mentioned above, a large value means that the linear model of the average gain kernel coefficients in Table 1 accounts for a large proportion of the variance, and the R 2 When is high, the variation in coefficients is small across populations. For Groups 1, 2, and 4 methods, we used the symbolic stimulation groups in Figure 5-column 1 (1 group, common gain), and Figure 6-column 4 (4 groups, per hemifield), and column 6 (2 groups, switch-eye), respectively. Surprisingly, the switch-eye symbolic stimulation group combined with three lags, the Group 2, lag 3 model in Table 2, row 4 with its 6 gain kernel coefficients, showed nearly as good consistency (0.740 at P129) as the complex Group 4, lag 4 model with its 16 gain kernel lag coefficients (0.757 at P129).

[0067] [Table 2]

[0068] Table 3 compares the different group / lag response estimation methods with respect to diagnostic power and reproducibility metrics. Data are for normal controls and glaucoma patients for P129. Very high ROC values ​​were not expected, as the patients' glaucoma was generally mild and 18 of their eyes were estimated as normal by any diagnostic criteria. Thus, the %ROC values ​​for sensitivity and lag are intended to give a relative measure of diagnostic power. Moreover, diagnostic power is a surrogate for depicting true differences in visual fields, whether for diagnostic purposes or not. Here, R 2The values ​​are the average variance explained by the response estimation model, and as with %ROC, larger is better. In the grey-level colour coding scheme, a lighter cell background means better performance. Srep and Drep are measures of relative repeatability between replicate pairs, and for these, smaller values ​​are better. Unless otherwise specified, all response estimation models had three lag coefficients per group in the gain kernel. The type of symbolic stimulus group is indicated by the column in the group diagram. These refer to the columns in Figures 5 and 6, so column 2 refers to column 2 in Figure 5, and column 5 refers to column 5 in Figure 6. As shown in Table 2, the best performing model in terms of lowest complexity is the two-group, switch-eye symbolic stimulus group assignment. This was much better than the two-group, within-eye method with the same number of gain kernel coefficients.

[0069] [Table 3]

[0070] The bottom two rows of Table 3, labeled Gain Setting, are another type of response estimation model. Here, we took the cross-validated fits for the two promising cases from Table 2, fitted a mixed-effects model as in Table 1, and then fit all of the subject's data to the average gain kernel for those two cases when response estimation was performed. Performance was worse than letting each pupil find the best-fitting model. The gain setting method demonstrates another non-limiting variant of the basic demonstration. It shows that using a response estimation model that is tailored to the dynamic gain characteristics of each individual pupil outperforms traditional methods that use standard extensions such as the Wiener kernel.

[0071] [Table 4]

[0072] The question arises as to what is the efficiency of the switch eye set for the two groups, i.e. the symbolic stimulation group. We investigated the values ​​of the gain kernel lag coefficients for several cases. Here, we fit a model like that in Table 1 and investigated the average value of the predicted gain kernel coefficients for people aged 50 years. Table 4A is the eight gain kernels for the eight group assignments of Figure 5, column 2, with two lags. Lag 0 is the instantaneous gain, and lag 1 is the effect on gain relative to the time 0.25 seconds before. There is a clear pattern at lag 0: groups 2 and 6 have the largest gains. Table 4B,C are the group assignments of Figure 6, column 4, 4B is the female data, and 4C is the male data. They are three-lag fits. By inspection of columns 2 and 4, it is clear that they are similar, but in column 4, the similar hemifield families are equated by whether they are three-ring or two-ring families. As expected, group 2 has the largest lag 0 gain, corresponding to groups 2 and 6 of Table 4A. The gains for men are slightly smaller, but the patterns at lag 0 are similar, and the patterns at lag 2 are similar.

[0073] Table 4D shows the best performing two groups: the switch-eye symbolic stimulus group assignment, i.e., column 6 of Figure 6, fitted with three lag coefficients. It has the interesting property that hemifield families of the same eye are pooled, i.e., each step of group 1 includes a jump to the other eye after one 0.25 s time step delay, and the same is true for group 2, which is true over all cycles of the stimulus. This assignment was therefore called switch-eye. Fitting a gain kernel to these symbolic groups preserves the large lag 0 gain of group 2 seen in the previous complex model. Table 4E shows a two-lag version of the switch-eye group, with lag 0 results similar to Table 4D. Figure 4F is an example using the symbolic group assignment of Figure 6, column 5, the so-called within-eye assignment. Here, group 1 is all families presented to the left eye, and group 2 is the families presented to the right eye. The pattern of high lag 0 gain for group 2 seen in Table 4A-E seems to be lost. The results of comparing the three-lag versions of SwitchEye and WithinEye in Table 3 show that despite having the same number of groups and gain kernel lag coefficients, SwitchEye performs surprisingly better than WithinEye.

[0074] The fact that switch-eye works well indicates that there is an element of eye-dependent gain. This was unexpected because previous work reported in US Patent No. 9,848,771 showed that gain changes occur in the Edinger-Westphal nucleus, where inputs from both eyes are summed together, eliminating the possibility of eye-dependent gain changes. The switch-eye assignment groups stimuli from eyes that have not seen stimuli for nearly 0.5 s, capturing the resulting higher dynamic gain state. Taken together, these evidences suggest an unexpectedly strong modulation of responsiveness at the level of the pretectal olivary nucleus or even further down the pathway from the retina to the pupil.

[0075] According to one example, if there are multiple types of hemifields or quadrant complementary sets, e.g., if a multifocal array is composed of five rings of stimuli centered at a fixation point and half of the hemifields or quadrants are drawn from either rings 1,3,5 or rings 2,4, creating two types of hemifield or quadrant volleys, such that a cycle of volleys for many round robin cycles will have eight types of volleys, but again, only two symbolic stimulus groups and two gain kernels can be measured to capture the switching eye patterns of the symbolic stimulus group assignments.

[0076] Overall, the experimental objectives were achieved. Jointly estimating the gain kernel and the temporal impulse responses per region improves the value of the estimated responses. A specific parsimonious assignment of symbolic stimulus groups allows us to fit response estimation models that are as efficient and consistent as estimation models with gain kernel lag coefficients 2-4 times larger. These findings were entirely unexpected and were not anticipated by the prior art. It would also be advantageous to include the gain kernel in the response estimation model to capture dynamic changes in gain.

[0077] [Evidence 2] In the first demonstration, a basis function was used that was a parametrically defined log-normal function (Equation 5). Herein, we discuss the possibility of using bilinear basis functions rather than parametric basis functions. It is therefore shown that the advantages of estimating the gain kernel also arise when alternative basis functions are estimated. For this demonstration, data from a group of 170 normal controls were used. These subjects were tested with the previously described 44-area / eye P129 and P130 multifocal stimuli while both eyes were examined simultaneously and both pupils were recorded with an apparatus such as that in Figure 2. Each subject was tested twice, approximately 2 weeks apart. The hemifield assignments of the symbolic stimulus groups in column 4 of Figure 6 yielded response estimates without and with the gain kernel for the four groups. The resulting fitted gain coefficients are similar to those in Table 4, B, C.

[0078] Figure 9 shows the results of averaging the estimated bilinear basis functions across pupils, subjects, and iterations, demonstrating the effect of including a gain kernel in the response estimation for the P129 stimulation protocol. Thus, the waveforms are the average of 170 subjects × 2 pupils × 2 iterations = 680 estimated bilinear basis functions. The legend indicates that the black curve (gain in legend) is the average basis function when estimating the gain kernel as described above. The grey curve (no gain in legend) is the result when no gain kernel is fitted. The basis function waveforms are clearly different. The basis function obtained with the gain kernel is slower to peak and slower to recover to the baseline pupil size (0 on the vertical axis). Thus, fitting a gain kernel changes the timing of the estimated response.

[0079] TIFF2025507979000031.tif76163

[0080] Table 5 summarizes the difference in results between the two response estimation methods, with (Gain) and without (No Gain) gain kernel estimation. All table entries are medians of 680 estimates. Due to the large numbers, the mean differences are substantially different even when multiple comparisons are considered. The amplitude of the peak pupil constriction is much larger with the gain kernel (row 2, Gain) than without the gain kernel being fitted (row 1, No Gain). This was true for both P129 and 130. The t-statistic, i.e., the constriction amplitude per region divided by the SE per region, provides a measure of the signal-to-noise ratio per region, and is much larger for the gain kernel version. Similarly, the average R for the model including the gain kernel 2 values ​​are also larger. As noted in Figure 9, the time to peak is longer. Overall, the addition of a gain kernel to the response estimation process improves the fit and signal-to-noise ratio when a single bilinear basis function per pupil is used.

[0081] [Table 5]

[0082] TIFF2025507979000033.tif76163

[0083] Figure 10 illustrates the use of two orthonormal basis functions to estimate the response of a single pupil. In this example, the pupil is the left pupil, and the response shown in Figure 10A results from stimulating the left eye of a 43-year-old male using the 44-region P129 stimulus described above. The 44 estimated responses are shown in Figure 10A as 44 waveforms with different grey levels. It is noteworthy that the 44 waveforms estimated per region have different peak contraction amplitudes and times to peak. These are achieved by summing different mixtures of the two basis functions found by an iterative fitting process, as suggested by the text above relating to equation (10).

[0084] TIFF2025507979000034.tif95163

[0085] [Table 6]

[0086] TIFF2025507979000036.tif105163

[0087] The data for component 2 in Table 6 shows a potential drawback of the bilinear basis function method, with the t-statistics for component 2 being small. In fact, the absolute values ​​of the t-statistics could be of either sign, so they were taken before averaging (e.g., Figure 10C). Nevertheless, they are less than 2, indicating that on average they are not significantly different from 0. This may suggest that flexibility is gained by using a parametric basis function such as the lognormal function in Equation 5, but also including the time derivative in the response estimation process to add one additional parameter per region.

[0088] TIFF2025507979000037.tif96163

[0089] Table 7 shows the resulting fitted gain kernel coefficients for the ON and OFF contrast for the four group assignments by hemifield for the symbolic stimulus groups in column 4 of Figure 6. As before, three lags are included. Thus, Table 7 follows the format for Tables 4, B, C, but the simultaneously estimated ON and OFF gain coefficients are entered separately. As in Tables 4, B, C, the largest gain was at lag 0 for group 2. On investigation, the OFF gain was indeed larger than ON, especially for lag 0, group 2, an unexpected favorable result. The t-statistics for the coefficients in the left column labeled Group 1 through Group 4 are shown in the four right columns similarly labeled. The lag 1 t-statistic for the OFF gain coefficients was surprisingly significant compared to the coefficients for the ON gains. This was also unexpected and revealed more about the pupillary system.

[0090] [Table 7]

Claims

1. presenting a set of multifocal stimuli in a portion of one or more visual fields of the subject, the set of multifocal stimuli being adapted to elicit a pupillary response from at least one pupil of the subject, the set of multifocal stimuli including a plurality of individual stimulus elements displayed in different components of the one or more visual fields, and presenting the individual stimulus elements simultaneously and in a statistically independent sequence; the set of multi-focal stimuli is displayed based on symbolic stimulus groups formed from a subset of statistically independent sequences; detecting responses to the displayed set of multi-focal stimuli using a sensor; Analyzing the detected responses; Measuring a dynamic change in pupil response gain based on the detected response; 11. A method for assessing the visual nervous system of a subject, comprising: measuring a temporal impulse response for each of the plurality of individual stimulation elements based on dynamic changes in measured pupil response gain.

2. The method of claim 1, wherein at any time step, the collection of multifocal stimuli is presented in families of a total number of the individual stimulus elements, which families are then assigned to an equal or smaller number of the symbolic stimulus groups, and dynamic changes in pupil response gain are measured only for the individual stimulus elements defined by the equal or smaller number of the symbolic stimulus groups.

3. 3. The method of claim 2, wherein at a particular time step, a set of multifocal stimuli is cyclically displayed in the visual fields of the two pupils, where a volley of the set of multifocal stimuli includes individual stimulus elements randomly selected from stimulus elements displayed in complementary left or right halves of the visual fields of the two pupils, where at successive time steps, a volley of the set of multifocal stimuli is displayed in the left hemifield of the left pupil, then in the right hemifield of the left pupil, then in the left hemifield of the right pupil, and then in the right half of the right pupil, with a probability of approximately 50% that an individual stimulus element from either the left or right hemifield of the left or right pupil is displayed, the sequence of volleys repeating cyclically a number of times, and the stimulus elements for the left and right hemifields of the left and right pupils being assigned to symbolic stimulus groups.

4. 4. The method of claim 3, wherein stimulus elements displayed in the left hemifield of the left pupil and the right pupil are assigned to a first symbolic stimulus group and stimulus elements displayed in the right hemifield of the left pupil and the right pupil are assigned to a second symbolic stimulus group, dynamic changes in pupil response gain are measured only for the first and second symbolic stimulus groups, and the impulse response is measured for each stimulus element based on the measured dynamic changes in pupil response gain.

5. 4. The method of claim 3, wherein the stimulus elements displayed in the left hemifield of the left pupil are assigned to a first symbolic stimulus group, the stimulus elements displayed in the right hemifield of the left pupil are assigned to a second symbolic stimulus group, the stimulus elements displayed in the left hemifield of the right pupil are assigned to a third symbolic stimulus group, and the stimulus elements displayed in the right hemifield of the right pupil are assigned to a fourth symbolic stimulus group, dynamic changes in pupil response gain are measured only for the first, second, third, and fourth symbolic stimulus groups, and the impulse response is measured for each stimulus element based on the dynamic changes in measured pupil response gain.

6. 4. The method of claim 3, wherein the stimulus elements displayed in the upper hemifield of the left pupil are assigned to a first symbolic stimulus group, the stimulus elements displayed in the lower hemifield of the left pupil are assigned to a second symbolic stimulus group, the stimulus elements displayed in the upper hemifield of the right pupil are assigned to a third symbolic stimulus group, and the stimulus elements displayed in the lower hemifield of the right pupil are assigned to a fourth symbolic stimulus group, dynamic changes in pupil response gain are measured only for the first, second, third, and fourth symbolic stimulus groups, and the impulse response is measured for each stimulus element based on the dynamic changes in measured pupil response gain.

7. 4. The method of claim 3, wherein the symbolic stimulus groups are arranged based on a first quadrant of stimulus elements for the left pupil and a second quadrant of stimulus elements for the right pupil, the first and second quadrants being diagonally opposed, the dynamic changes in pupil response gain are measured only for the symbolic stimulus groups, and the impulse response is measured for each stimulus element based on the dynamic changes in measured pupil response gain.

8. The method of claim 7 , wherein one or more of the symbolic stimulus groups further comprises one or more additional stimulus elements outside of the first and second quadrants.

9. The method of claim 3 , further comprising measuring two symbolic stimulus groups and two gain kernels to capture switched eye patterns of symbolic stimulus group assignments.

10. The method of claim 3, wherein if there are N different hemifields or quadrant volleys, N / 2 symbolic stimulus groups are assigned to capture the switch eye pattern of the symbolic stimulus group assignments.

11. 3. The method of claim 2, wherein the visual field of both pupils is stimulated by the display of a set of multifocal stimuli including four subsets of stimulus elements stimulating a first complementary portion and a second complementary portion of the visual field of the two pupils, the assignment to the two symbolic stimulus groups follows a sequence of the first complementary portion displayed to the first pupil assigned to a first group of the symbolic stimulus groups, the second complementary portion displayed to the first pupil assigned to a second group of the symbolic stimulus groups, the first complementary portion displayed to the second pupil assigned to the first group, and the second complementary portion displayed to the second pupil assigned to the first group, and the impulse response is measured for all stimulus elements based on dynamic changes in pupil response gain measured only for the first group and the second group.

12. The method of claim 11 , wherein the first and second complementary portions of the set of multifocal stimuli include sporadic deviations from a set pattern of the first and second complementary portions.

13. A method according to any one of the preceding claims, wherein the temporal impulse response is measured as a parametric function of a limited number of parameters for each part of the field of view using a log-normal distribution.

14. The method of claim 13 , wherein the time derivative of the parametric function is used in the response measurement.

15. The method according to any one of claims 1 to 12, wherein the temporal impulse response is measured as a set of one or more bilinear orthonormal basis functions.

16. 1. A computer system for assessing the visual nervous system of a subject, the computer system comprising: A computer, a computer-controlled light source configured to display a set of multifocal stimuli in a portion of one or more visual fields of the subject, the set of multifocal stimuli being adapted to elicit a pupillary response from at least one pupil of the subject, the set of multifocal stimuli including a plurality of individual stimulus elements displayed in different portions of the one or more visual fields, the light source configured to simultaneously display the individual stimulus elements in a statistically independent sequence, and further configured to display, under computer control, the set of multifocal stimuli based on symbolic stimulus groups formed from a subset of the statistically independent sequences; at least two sensors positioned to detect responses to the set of multi-focal stimuli displayed; The computer includes: Analyzing the detected responses; Measuring a dynamic change in pupil response gain based on the detected response; A computer system configured to measure a temporal impulse response for each of the plurality of individual stimulation elements based on dynamic changes in measured pupil response gain.

17. The computer system of claim 16, wherein the light source is controlled by the computer to display, at any time step, a set of the multifocal stimuli within families of the total number of individual stimulus elements, which families are then assigned to the same or fewer number of symbolic stimulus groups, and the computer is configured to measure only dynamic changes in pupil response gain for the individual stimulus elements defined by the same or fewer number of symbolic stimulus groups.

18. 18. The computer system of claim 17, wherein the light source is controlled by the computer to periodically display a set of multifocal stimuli in the visual field of two pupils at a particular time step, the volley of the set of multifocal stimuli including individual stimulus elements randomly selected by the computer from stimulus elements displayed in complementary left or right halves of the visual field of the two pupils, and at successive time steps the light source is controlled by the computer to display a volley of the set of multifocal stimuli in the left hemifield of the left pupil, then the right hemifield of the left pupil, then the left hemifield of the right pupil, then the right half of the right pupil, with an approximately 50% probability that an individual stimulus element from either the left or right hemifield of the left or right pupil is displayed, the sequence of volleys repeating periodically a number of times, and the stimulus elements for the left and right hemifields of the left and right pupils being assigned to symbolic stimulus groups.

19. 20. The computer system of claim 18, wherein the stimulus elements displayed in the left hemifield of the left pupil and the right pupil are assigned by the computer to a first symbolic stimulus group and the stimulus elements displayed in the right hemifield of the left pupil and the right pupil are assigned by the computer to a second symbolic stimulus group, the computer being configured to measure dynamic changes in pupil response gain only for the first and second symbolic stimulus groups, and further configured to measure the impulse responses to all stimulus elements based on the measured dynamic changes in pupil response gain.

20. 20. The computer system of claim 18, wherein the stimulus elements displayed in the left hemifield of the left pupil are assigned by the computer to a first symbolic stimulus group, the stimulus elements displayed in the right hemifield of the left pupil are assigned by the computer to a second symbolic stimulus group, the stimulus elements displayed in the left hemifield of the right pupil are assigned by the computer to a third symbolic stimulus group, and the stimulus elements displayed in the right hemifield of the right pupil are assigned by the computer to a fourth symbolic stimulus group, the computer is configured to measure dynamic changes in pupil response gain only for the first, second, third, and fourth symbolic stimulus groups, and further configured to measure the impulse responses to all stimulus elements based on the measured dynamic changes in pupil response gain.

21. 20. The computer system of claim 18, wherein the stimulus elements displayed in the upper hemifield of the left pupil are assigned by the computer to a first symbolic stimulus group, the stimulus elements displayed in the lower hemifield of the left pupil are assigned by the computer to a second symbolic stimulus group, the stimulus elements displayed in the upper hemifield of the right pupil are assigned by the computer to a third symbolic stimulus group, and the stimulus elements displayed in the lower hemifield of the right pupil are assigned by the computer to a fourth symbolic stimulus group, the computer is configured to measure dynamic changes in pupil response gain only for the first, second, third, and fourth symbolic stimulus groups, and further configured to measure the impulse responses to all stimulus elements based on the measured dynamic changes in pupil response gain.

22. 20. The computer system of claim 18, wherein the symbolic stimulus groups are arranged by the computer based on a first quadrant of stimulus elements for the left pupil and a second quadrant of stimulus elements for the right pupil, the first quadrant and the second quadrant being diagonally opposed, the computer being arranged to measure dynamic changes in pupil response gain only for the symbolic stimulus groups, and further configured to measure the impulse responses for all stimulus elements based on the measured dynamic changes in pupil response gain.

23. 23. The computer system of claim 22, wherein one or more of the symbolic stimulus groups further comprises one or more additional stimulus elements outside of the first and second quadrants.

24. 20. The computer system of claim 18, wherein the computer is configured to measure two symbolic stimulus groups and two gain kernels to capture a switched eye pattern of a symbolic stimulus group assignment.

25. 20. The computer system of claim 18, wherein when there are N different hemifields or quadrant volleys, N / 2 symbolic stimulus groups are assigned by the computer to capture switch eye patterns of symbolic stimulus group assignments.

26. 18. The computer system of claim 17, wherein the light source is configured to stimulate the visual field of both pupils by displaying, under control of the computer, a set of multifocal stimuli including four subsets of stimulus elements stimulating first and second complementary portions of the visual field of the two pupils, and the assignment to the two symbolic stimulus groups follows a sequence of the first complementary portion displayed to the first pupil assigned by the computer to a first group of the symbolic stimulus groups, the second complementary portion displayed to the first pupil assigned by the computer to a second group of the symbolic stimulus groups, the first complementary portion displayed to the second pupil assigned by the computer to the first group, and the second complementary portion displayed to the second pupil assigned by the computer to the first group, and the computer is configured to measure the impulse response to all stimulus elements based on the measured dynamic change in pupil response gain for only the first group and the second group.

27. 27. The computer system of claim 26, wherein the first and second complementary portions of the set of multi-focal stimuli include sporadic deviations from the set pattern of the first and second complementary portions.

28. 28. A computer system according to any one of claims 16 to 27, wherein the computer is configured to measure the temporal impulse response as a parametric function of a limited number of parameters for each part of the field of view using a log-normal distribution.

29. 30. The computer system of claim 28, wherein the computer is configured to use a time derivative of the parametric function in measuring the response.

30. A computer system according to any one of claims 16 to 27, wherein the computer is configured to measure the temporal impulse response as a set of one or more bilinear orthonormal basis functions.

Images