Method and system for assessing presence of traumatic brain injury

Abstract

Methods and systems for assessing brain health of a patient are described. In certain forms, the assessment of the presence of traumatic brain injury (TBI), such as mild traumatic brain injury (mTBI), of the patient is performed by analyzing data representative of movement of the patient's eye.

Description

Statement corresponding to the application

[0001] This application is based on Australian Patent Application No. 2023902472, the entire contents of which are incorporated herein by reference. Technical Field

[0002] This technology relates to the field of brain health assessment, and in particular to the assessment of the presence of traumatic brain injury (TBI), such as mild TBI (mTBI) (commonly known as concussion). Background Technology

[0003] The human eye is a complex and sophisticated organ used to perceive the world around us. The eye captures light and forms an image on the retina. This generates electrical signals that are transmitted to the brain and interpreted as visual information. Scientists and medical professionals have recognized the importance of tracking eye movements in the diagnosis and treatment of various medical conditions. In particular, eye movements have been found to be a useful indicator of brain lesions, including the presence of certain medical conditions.

[0004] An example of a medical condition that can be assessed using eye movement is traumatic brain injury, including mild traumatic brain injury (mTBI). MTBI is a complex neurobehavioral phenomenon caused by mechanical forces from a direct impact to the skull or by indirect forces such as acceleration / deceleration, resulting in deformation of brain tissue. It can cause a range of symptoms, including headache, dizziness, fatigue, depression, anxiety, irritability, loss of consciousness, and cognitive impairment, which can last from days to years due to damage to axonal microstructures and changes in neurometabolism leading to disruption of brain networks. The effects of a concussion can affect the brain's ability to control eye movement, leading to symptoms such as double vision, blurred vision, and coordination problems.

[0005] Despite increased understanding of the biomechanics and pathophysiology of concussion, standardized biomarkers (neither clinical nor serological) still exist. Diagnosis of concussion can be based on a combination of self-reported symptoms and physical and neurological examinations. Self-reported symptoms may not be fully disclosed to the physician and are subjective. A classic examination method involves the physician having the patient look at their fingers as they move them around and observing how the patient's eyes track the movement. Such methods can be subjective and error-prone, and may ideally require a clinical setting suitable for careful testing.

[0006] In many situations, it may be necessary to quickly and accurately assess the likelihood of a concussion away from the clinical setting. Such situations include during contact or competitive sports such as football (NFL, English football, Australian Rules Football), rugby, boxing, and martial arts, or at the scene of an injury such as a road vehicle accident.

[0007] Brain imaging methods (such as CT and MRI scans) require expensive and bulky equipment that is not portable and rarely has sufficient sensitivity to diagnose mTBI, and are therefore unsuitable for in situ diagnosis as required in the above cases.

[0008] One existing system is Oculogica's EyeBox. This system relies on convergence (a measure of how well the eyes work in sync) to determine the likelihood of a concussion, producing a sensitivity of 80.4% and a specificity of 66.1% in detecting mTBI (Samadani, U., Spinner, RJ, Dynkowski, G., Kirelik, S., Schaaf, T., Wall, SP, & Huang, P. (2022), Eyetracking for classification of concussion in adults and pediatrics, Frontiers in neurology, 13. doi:10.3389 / fneur.2022.1039955). This sensitivity may be insufficient as a diagnostic tool to immediately determine whether a patient can continue high-risk activities for mTBI, such as physical contact sports. Furthermore, the operating protocol (convergence) used by EyeBox requires a large desktop computer with a stable platform. Inaccurate readings and the physical characteristics of the EyeBox make it unsuitable for use in sports fields and field environments to provide concussion indications in the event of an injury.

[0009] The system from Neuroalign uses convergence, saccades, and reaction time, with a frame rate of only 100Hz. A frame rate of only 100Hz may miss many subtle fixational eye events (such as microsaccades and other fixational metrics, which will be described later in the context of this technique).

[0010] Smooth tracking analysis is an existing area of ​​research in mTBI. Michael Kelly (without a commercially available device) developed a portable device that displays a 10-second smooth tracking figure-eight protocol, which has been used in 849 athletes (12–18 years old) and 98 patients with mTBI. Patients with mTBI showed significantly skewed tracking movements based on z-scores from canonical data (Kelly, M. (2017), Technical Report of the Use of a Novel Eye Tracking System to Measure Impairment Associated with Mild Traumatic Brain Injury, Cureus, 9(5), e1251.doi:10.7759 / cureus.1251). Similarly, both Neuroflex (Saccade Analytics) and Righteye VisionSystem utilize smooth tracking (sensitivity to mTBI detection is not reported), but only the magnitude of the error is measured. RightEye does not account for calibration error, which is a significant confounding factor affecting the accuracy of smooth tracking.Maruta et al. from EyeSync (NeuroSync) also relied on smoothing the tracking error in their protocol (Maruta, J., Heaton, KJ, Kryskow, EM, Maule, AL, & Ghajar, J. (2013), Dynamic visuomotor synchronization: quantification of predictivetiming, Behav Res Methods, 45(1), 289-300. doi:10.3758 / s13428-012-0248-3; Maruta, J., Heaton, KJ, Maule, AL, & Ghajar, J. (2014), Predictive visual tracking: specificity in mild traumatic brain injury and sleep deprivation, Mil Med, 179(6). 619-625. doi:10.7205 / MILMED-D-13-00420;Maruta, J., Spielman, LA, Rajashekar, U., & Ghajar, J. (2018), Association of Visual Tracking Metrics With Post-concussion Symptomatology, Frontiers in Neurology, 9. doi:10.3389 / fneur.2018.00611;Maruta, J., Suh, M., Niogi, SN, Mukherjee, P., & Ghajar, J. (2010), Visual tracking synchronization as a metric for concussionscreening, J Head Trauma Rehabil, 25(4), 293-305. doi:10.1097 / HTR.0b013e3181e67936). The inventors believe that this metric can be improved using the metric described in this specification.

[0011] The RightEye Vision System uses a gaze stability estimate that employs a metric called the area of ​​the bivariate isoline ellipse (BCEA), which measures the dispersion of eye-tracking coordinates within an elliptical region on a graph, specifically how many fall within the 68th percentile of the distribution (Snegireva, N., Derman, W., Patricios, J., & Welman, K. (2021), Eye tracking toassess concussions: an intra-rater reliability study with healthy youth and adult athletes of selected contact and collision team sports, Experimental Brain Research, 239(11), 3289-3302. doi:10.1007 / s00221-021-06205-6). Their gaze protocol presents stimuli for only up to two seconds and requires participants to move their heads while making an effort to gaze. Such short stimulus presentations do not elicit many eye movements that could be beneficial for analysis. In another study, RightEye's protocol showed a target duration of 7 seconds (BCEA as the outcome measure), demonstrating a significant difference between subjects who had suffered concussions and those who had not, which was improved when combined with a statistical model that included eye convergence and divergence (Hunfalvay, M., Murray, NP, & Carrick, FR (2021), Fixation stability as a biomarker for differentiating mild traumatic brain injury from age matched controls in pediatrics, Brain Inj, 35(2), 209-214, doi:10.1080 / 02699052.2020.1865566). However, this still yielded only 65% ​​sensitivity and 70% specificity in the diagnosis of mTBI. Leonard and colleagues also specifically studied fixational eye movements in mTBI.The team used a 30Hz scanning laser ophthalmoscope (a large desktop machine used in clinics) to track gaze events at 480Hz. Their gaze metrics were evaluated in orientation histograms, specifically examining BCEA, gaze saccade velocity, acceleration, amplitude, and drift. During their test protocol, only their measurements of gaze saccade peak velocity, acceleration, and amplitude during these gaze tasks were shown to be significant between patients who had recently suffered a concussion and healthy patients (Leonard, BT, Kontos, AP, Marchetti, GF, Zhang, M., Eagle, SR, Reecher, HM, … Rossi, EA (2021), Fixational eye movements following concussion, Journal of Vision, 21(13), 11. doi:10.1167 / jov.21.13.11).

[0012] Cifu and colleagues used another large desktop application (EyeLink II at 500 frames / second) in a group of military personnel with a history of TBI (at least 8.5 months post-injury) (Cifu, DX, Wares, JR, Hoke, KW, Wetzel, PA, Gitchel, G., & Carne, W. (2015), Differential eye movements in mild traumatic brain injury versus normal controls, Journal of Head Trauma Rehabilitation, 30(1), 21-28. doi:10.1097 / htr.0000000000000036). Their fixation metrics included localization variance, root mean square of eye velocity, and mean and absolute mean eye velocity during fixation. They also included BCEA as an additional metric for the geographic distribution of eye-tracking coordinates. No significant results were obtained using this approach.

[0013] In a pilot study of nine patients with acute concussion and nine healthy controls, researchers used a Nintendo® Wii device and a 120Hz head-mounted eye-tracking system to measure the percentage of time spent gazing fixed on the center of the game screen during play (percentage of time at center) and the number of times the gaze deviated (eye movement) away from the center of the screen (gaze deviation), showing significant differences between groups in both measures (Murray, NG, Ambati, VN, Contreras, MM, Salvatore, AP, & Reed-Jones, RJ (2014), Assessment of oculomotor control and balance post-concussion: a preliminary study for a novel approach to concussion management, Brain Injury, 28(4), 496-503. doi:10.3109 / 02699052.2014.887144).

[0014] Another recent study used an integrated virtual reality headset and eye tracker (HTC Vive with a 250Hz eye tracker) to evaluate 86 concussion patients within 50 days of injury, measuring "the number of saccades generated, the size and speed of microsaccadic movements, the area covered, and the ratio of vertical to horizontal components of fixational eye movements" (Mortazavi, M., Thirunagari, P., Sarva, S., & Pita, M. (2022), Microsaccadic Fixational Eye Movements as an Oculomotor Marker for Concussion, Neurology, 98(1 Supplement 1), S5-S5. doi:10.1212 / 01.wnl.0000801776.06317.1f). Their team found a higher mean value for microsaccadic movements in their concussion patient cohort. Additionally, during fixation, microsaccades and drifts cover the more vertical area.

[0015] There is a need for more tools that allow for the assessment of medical conditions through eye movement in a more convenient, objective, and / or accurate manner than some existing diagnostic tools.

[0016] Technical Purpose

[0017] The purpose of this technology is to provide an improved method, system, and / or apparatus for assessing a patient's brain health (e.g., assessing the presence of traumatic brain injury (TBI)). Alternatively, the purpose of this technology is to at least provide the public with a useful option. Summary of the Invention

[0018] According to one aspect of the present technology, a method for assessing a patient's brain health is provided. In some forms, the method may include assessing the presence of traumatic brain injury (TBI) in the patient (e.g., mild traumatic brain injury (mTBI)). The method may include analyzing data representing the patient's eye movements.

[0019] According to one aspect of the present technology, a computer-implemented method for assessing the presence of TBI in a patient is provided, the method comprising:

[0020] Receive eye data, which represents the patient's eye movements;

[0021] Analyze the eye data to determine indications of the presence of TBI in the patient; and

[0022] Output this instruction.

[0023] In some forms, eye data representing eye movement can include eye data representing fixational eye movement (i.e., eye movement during a fixation task). More specifically, eye data can represent eye movement when a patient gazes at a fixed target. The fixed target can be displayed at an off-center location relative to the patient. In some forms, eye data can represent eye movement when a patient gazes at multiple fixed targets that appear sequentially (e.g., at random time intervals and / or for random durations). The multiple fixed targets can be displayed at different locations within the eye's visual field.

[0024] Additionally or alternatively, eye data representing eye movement may include eye data representing smooth eye following of eye movement (i.e., eye movement during a smooth following task). More specifically, eye data may represent eye movement as the patient gazes at a moving target. The moving target may outline a shape, for example, by repetitively outlining that shape.

[0025] In some forms, the method may include determining the direction of gaze by detecting the localization of the iris of the eye. Detecting the localization of the iris may include applying an object detection algorithm to eye data representing eye movement (e.g., a one-off object detector, such as the YOLO (You Only Look Once) v7 real-time object detector). From this, the curvature of the iris can be calculated, and thus, pupil localization can be accurately estimated even when the pupil is obscured by the eyelid.

[0026] In some forms, the method may include classifying eye data representing eye movement using a transformer model (e.g., a bidirectional transformer model).

[0027] In some forms, the method further includes receiving target data representing the location of a target while eye data representing eye movement is captured. The method may further include analyzing the target data to determine an indication of the presence of TBI in the patient.

[0028] In some forms, the steps of analyzing eye data and optionally target data may include: determining one or more measures of eye movement, and determining an indication of the presence of TBI in the patient based on the one or more measures.

[0029] In some forms, the one or more measures may include:

[0030] a) A measure of the average error between the localization of a fixed target and the localization of the patient's gaze during the fixation task. For example, a measure of the average error could be the root mean square error (RMSE).

[0031] b) A measure of the frequency of microsaccades during fixation tasks;

[0032] c) A measure of one or more characteristics of one or more square wave jumps (SWJs) during the fixation task; and

[0033] d) A measure of smooth following interruptions when following a moving target. For example, a measure of a smooth following interruption could be the time before the interruption occurs in a smooth following task. An interruption may occur when the accuracy of the eye following a moving target falls below a certain threshold.

[0034] In some forms, the steps of analyzing data may include: determining any combination of one or more of the following measures: a), b), c), and d).

[0035] In some forms, the steps of analyzing data may include: determining any combination of two or more measures among measures a), b), c), and d), and the step of determining an indication may include combining two or more measures. For example, two or more measures may be combined into a weighted average.

[0036] In some forms, determining the indicator may include comparing one or more measures, and / or a combination of two or more measures, to one or more predetermined thresholds. In some forms, one or more predetermined thresholds may be determined from similar measures determined from other patients. In other forms, one or more predetermined thresholds may be determined from similar measures determined from the patient at one or more earlier times.

[0037] In some forms, one or more measures may include a measure of the smoothness of eye movement when following a moving target. In some forms, the measure of smoothness may be combined with measure d) to determine it.

[0038] In some forms, the method includes: outputting an indication as a quantitative and / or qualitative assessment of a patient's risk of having TBI.

[0039] In some forms of this technology, the method further includes controlling a display screen to display a target to the patient. In some forms, the display screen can be controlled to display a fixed target. The fixed target can be displayed at an off-center location relative to the patient. In some forms, the display screen can be controlled to display multiple fixed targets that appear sequentially (e.g., at random time intervals and / or for random durations). The multiple fixed targets can be displayed at different locations within the eye's field of vision. In some forms, the display screen can be controlled to display a moving target. The moving target can outline a shape, for example, repeatedly outlining that shape. The method can further include generating data representing the location of the target.

[0040] In some forms, the method further includes controlling a camera to capture an image of the eye while the eye is gazing at a target. The method may further include generating data representing eye movement from the image captured by the camera.

[0041] According to one aspect of the present technology, a computer-implemented method is provided for assessing the presence of TBI in a patient. The method may include: receiving eye data representing fixational eye movements of a patient's eyes while the patient gazes at a fixed target during a fixation task. The method may further include: determining one or more measures of the eye movements from the eye data. The one or more measures may include measures of one or more characteristics of one or more square wave jerks (SWJs). The method may further include: determining an indication of the presence of TBI in the patient based on the one or more measures. The method may further include outputting the indication.

[0042] In some forms, the one or more features may include atypical features of the one or more SWJs. The inventors have identified that SWJs in patients undergoing TBI can be morphologically diverse in the sense of inducing atypical movements that differ in shape from those expected in healthy patients. For example, this morphological diversity may manifest as the occurrence of different types or sequences of movements, or as measures such as duration, shape, amplitude, and / or velocity. SWJs exhibiting these atypical characteristics can be considered malformed compared to typical SWJs and may be referred to herein as “malformed” SWJs.

[0043] In some forms, at least one of the one or more SWJs may include two or more phases. In the example, at least one SWJ may include three or more phases. An SWJ with two phases may be referred to as biphase, and an SWJ with more than one phase may be referred to as multiphase.

[0044] The reference to phase in a SWJ should be understood as referring to a unique event in the SWJ that can be distinguished from other events in the SWJ, including a sequence of movements or shifts. For example, a typical SWJ in a healthy patient includes a primary saccade deflection away from the target, a gliding portion, and an accurate saccade recovery back to the target—all of which are collectively considered a single phase (i.e., a monophasic SWJ). Events deviating from this form can be considered additional phases, whether they are additional events (e.g., additional deflections or recoverys) or combinations thereof.

[0045] In some forms, determining the one or more metrics may include classifying each of two or more phases. For example, a phase may be classified into two or more of the following: primary saccade deflection, subsequent saccade deflection, saccade recovery, progressive recovery, saccade peak, and glide phase.

[0046] In some forms, determining the one or more metrics may include quantizing one or more features of each of the two or more phases.

[0047] In some forms, one or more features of primary saccade deflection and / or subsequent saccade deflection may include one or more of the following: saccade deflection count, saccade velocity, saccade amplitude, and peak deflection.

[0048] In some forms, one or more features of saccadic recovery and / or progressive recovery may include one or more of the following: saccadic or progressive recovery count, saccadic or progressive speed, saccadic or progressive amplitude, peak deflection, and classification (e.g., accurate, excessive, or insufficient).

[0049] In some forms, one or more characteristics of saccade peaks may include one or more of the following: saccade peak count and saccade amplitude.

[0050] In some forms, one or more characteristics of the gliding phase may include one or more of the following: duration, fibrillation, and slope (e.g., the slope is a flat angle, positive toward recovery, or negative away from recovery).

[0051] In some forms, indicators of the variability of gaze stability between SWJ events can be determined. In some forms, one or more characteristics of the variability of gaze stability between SWJ events may include one or more of the following: fibrillation and drift (e.g., continuous movement away from the target).

[0052] In some forms, the indication for determining the presence of TBI in a patient can be based at least in part on the cumulative sum of SWJ occurrences. In one example, biphasic or multiphasic SWJs can be counted as a single SWJ occurrence. In other examples, each phase of a biphasic or multiphasic SWJ can be counted as an SWJ occurrence.

[0053] In some forms, a higher number of phases in a SWJ can be assigned a higher weight. In some forms, hierarchical weighting can be applied at least in part based on the complexity of the SWJ. In some forms, weighting can be at least in part based on the number of phases in the SWJ. For example, a higher weight can be assigned to an SWJ with a higher number of phases.

[0054] For example, a hierarchical weighted approach can be applied, where:

[0055] 1. A typical SWJ is given a weighted W1;

[0056] 2. Abnormal SWJs are given a weighted W2;

[0057] 3. Multiphase SWJ[2] (i.e., determined to have two phases) is given a weighted W3;

[0058] 4. Multiphase SWJ[3] is weighted by W4;

[0059] 5. The multiphase SWJ[4] is weighted by W5; and

[0060] 6. The multiphase SWJ[n] is given a weighted W n+1 ,

[0061] Where W1 < W2 < W3 … < W n+1 .

[0062] In some forms, weighting can be adjusted based on one or more measures of one or more features of the associated phase. For example, weighting can be biased based on features such as the magnitude of peak deflection.

[0063] For example, algorithms that implement this weighting could include:

[0064] Indication of the presence of TBI = (typical SWJ) 总和 * W1) + (Deformed SWJ) 总和 * W2) + (multiphase SWJ[2]) 总和 * W3) + … + (polyphase SWJ[n]) 总和 * W n+1 )

[0065] In some forms, each phase of a biphasic or multiphasic SWJ can be counted as an SWJ occurrence.

[0066] For example, algorithms that implement this weighting could include:

[0067] Indication of the presence of TBI = (typical SWJ) 总和 * W1) + (Deformed SWJ) 总和 * W2) + (multiphase SWJ[2] phase) 总和 * W3) + … + (multiphase SWJ[n] phase) 总和 * W n+1 )

[0068] In some forms, additional eye data representing fixational eye movements of the patient's eyes during a second fixation task can be received. The determination of a second indication of the presence of TBI in the patient can be based on this additional eye data. The progress of neuronal recovery in the patient can be determined by comparing the second indication of the presence of TBI with a previous indication of the presence of TBI.

[0069] According to one aspect of the present technology, a system for assessing the presence of TBI in a patient is provided, the system including a processor configured to perform a computer-implemented method according to another aspect of the present technology.

[0070] According to one aspect of the present invention, a computer-readable medium is provided having instructions stored thereon for performing a computer-implemented method for assessing the presence of a patient's TBI according to another aspect of the present invention.

[0071] Further aspects of the present technology (which should be considered in all its novel aspects) will become apparent to those skilled in the art after reading the following description which provides at least one example of a practical application of the present technology. Attached Figure Description

[0072] Referring to the following figures, one or more embodiments of the present technology will be described below by way of example only and not as a limitation, in which:

[0073] Figure 1A This is a frontal view diagram of the human eye;

[0074] Figure 1B yes Figure 1A The cross-section of eye 101 in the sagittal plane;

[0075] Figure 2A It is an illustration showing typical eye behavior of microsaccades;

[0076] Figure 2B This is an illustration of typical eye behavior exhibiting microtremors;

[0077] Figure 2C It is an illustration demonstrating typical eye behavior that exhibits drift;

[0078] Figure 2D It is a diagram illustrating typical eye behavior of a square wave jumping rapidly;

[0079] Figure 2E It is a graph illustrating exemplary eye behavior of a square wave jump;

[0080] Figure 2F It is a graph illustrating exemplary eye behavior of abnormal square wave jumps;

[0081] Figure 2G It is a graph illustrating exemplary eye behavior of biphasic square wave jumps;

[0082] Figure 2H It is a graph illustrating exemplary eye behavior of polyphasic abnormal square wave jumps;

[0083] Figure 2I It is a graph illustrating exemplary eye behavior of abnormal square wave jumps;

[0084] Figure 2J It is a graph illustrating exemplary eye behavior of polyphasic abnormal square wave jumps;

[0085] Figure 2K It is a graph illustrating exemplary eye behavior of abnormal square wave jumps;

[0086] Figure 2MIt is a graph illustrating exemplary eye behavior, showing the saccadic peak in a distorted square wave jump;

[0087] Figure 3 This is a schematic illustration of an apparatus and / or system for analyzing eye movements, according to an exemplary form of the present technology;

[0088] Figure 4 This is a schematic illustration of an eye-tracking system according to an exemplary form of the present technology;

[0089] Figure 5 This is a schematic diagram of an exemplary data analysis system according to one form of the present technology;

[0090] Figure 6 This is a flowchart outlining an exemplary method for assessing the presence of TBI in a patient, according to one form of the present technology;

[0091] Figure 7A It is a diagram of a display screen showing the movement path of a target, according to one form of the present technology;

[0092] Figure 7B yes Figure 7A Another illustration of the display screen also shows the projection of the target's motion path over time on the X and Y axes;

[0093] Figures 8A to 8D This is an illustration of radially transformed eye-tracking data in a smoothing follow-up task according to this technique; and

[0094] Figures 9A to 9D This is an illustration of eye-tracking data in another form of smooth following task according to this technique. Detailed Implementation

[0095] 7.1 Eyes

[0096] This technology relates to apparatus, systems, and methods for analyzing data representing eye movement (e.g., the eye's ability to follow a target). Some relevant aspects of eye anatomy and movement will now be described. This technology primarily relates to analyzing human eye movement, but can be applied to analyzing the eyes of other animals in other forms.

[0097] 7.1.1 Anatomical Structure of the Eye

[0098] Figure 1A This is a front view illustration of a human eye 101, which includes an eyeball 104 and a pupil 106. The movement of the eye 101 can be characterized by the movement of the eyeball 104 and / or the pupil 106 along two mutually perpendicular axes, for example, along an axis in the lateral direction relative to the body (i.e., the horizontal direction when the body is upright). Figure 1A The example is illustrated by the x-axis 107, and the axis in the vertical direction relative to the body (i.e., the vertical direction when the body is upright). Figure 1A In the example, the y-axis is 108. These axes are also... Figure 1B The above example shows that the diagram is Figure 1A The cross-section of eye 101 in the sagittal plane (vertical when the body is upright).

[0099] 7.1.2 Eye Movement

[0100] The body has muscles that control the movement of the eyeball 104. In the human eye 101, these are the extraocular muscles and the intraocular muscles. There are six extraocular muscles that control the movement and alignment of the eye within the orbit, and a seventh muscle that controls a portion of the upper eyelid. The intraocular muscles control the movement of the lens and the dilation / contraction of the pupil, which allows the eye to focus on near objects and control how much light enters the eye.

[0101] Several different types of eye movements were characterized, including:

[0102] • Salivation – is a rapid, ballistic movement of the eye that abruptly changes the point of fixation. They range in amplitude from small movements, such as those made while reading, to much larger movements, such as those made while gazing around a room. Salivation can be voluntarily initiated, but it occurs reflexively whenever the eyes are open (even when fixated on a target).

[0103] Microsalivation – is a type of fixational eye movement. These are small, involuntary micro-eye movements similar to jumps, miniature forms of voluntary salivation. They typically occur during prolonged visual fixation to prevent decay.

[0104] • Forward saccades – scans directed toward the target, usually produced by reflex;

[0105] • Reverse saccade – A saccade that moves away from the target, usually generated by willpower;

[0106] • Drift – The brain mechanisms behind eye drift are not fully understood, but these are slow, gradual movements that occur between microsaccades during fixation;

[0107] Tremor – a small, high-frequency perturbation that occurs between microsaccades;

[0108] • Fixational eye movements: for example, microsalivation, square wave jerks, tremor, and drift;

[0109] • Fixation – Fixation consists of slow and minute movements (fixational eye movements) that help align the eyes with a target and avoid sensory decay. The duration can vary, for example, between 50-600 ms; and

[0110] • Smooth tracking – These are movements that are much slower tracking movements of the eye, designed to remain in the fovea of ​​the eye. Such movements are voluntary in the sense that the observer can choose whether to track the moving stimulus, and occur between saccades.

[0111] • Square jump – A form of fixational eye movement that occurs within 150-500 ms, moving away from and back to the fixation point with approximately equal magnitudes. Further discussion of square jumps is provided below in relation to eye movement dysfunction.

[0112] Figures 2A to 2D This is a diagram illustrating some typical eye behaviors of eye movements as explained above. Figure 2A A graph illustrating the change of eye gaze coordinates over time is shown. The two small gaze direction changes illustrated can be characterized as microsaccades. Figure 2B A graph illustrating the change of the horizontal (x) coordinate of the eye's gaze over time is shown. The small, high-frequency perturbations illustrated can be characterized as microtremors. Figure 2C The graph illustrates the horizontal (x) and vertical (y) coordinates of eye gaze at different points in time, with the lines representing successive points on the graph. The gradual shift in gaze over time illustrated in this graph can be characterized as a drift. Figures 2A to 2C The movement illustrated is a typical fixational eye movement, that is, a small involuntary eye movement that may occur when a person attempts to fix their gaze on a target. Figure 2D A graph illustrating the change of eye gaze coordinates over time is shown. The abrupt change in gaze direction, exemplified by a short period of time (e.g., several hundred milliseconds) away from the target and then returning to it, can be characterized as a square wave jump (SWJ). At least for some examples in this application, SWJ typically occurs in the horizontal (lateral) direction, and therefore, the graph plotted on... Figure 2D The gaze location on the vertical axis can be the horizontal (x) coordinate of the eye's gaze. For completeness, it should be understood that the discussion of movement in the horizontal direction is not intended to exclude the possibility that SWJ occurs in other directions (e.g., the vertical direction).

[0113] This technique can be used to analyze any or more of the eye movement types described above.

[0114] Eye movement dysfunction occurs when there are abnormalities or impairments in normal eye movements (e.g., saccades and smooth following may be inaccurate relative to the target, or may be interrupted during movement or irregular in timing). Poor saccadic distance is a motor error that results in over- or under-aiming of the eye at a target, accompanied by corrective saccades. Several measures can be used to quantify eye movement dysfunction; for example, saccadic gain is the ratio of eye movement to target position, and stimulus delay is the delay in response to a stimulus in the form of a target presented in the visual plane before the start of a motor command.

[0115] 7.1.2.1 Square wave jump

[0116] Figures 2E to 2M A theoretical graph illustrating the change over time of one-dimensional eye positioning 1000 (i.e., the horizontal coordinate of eye gaze) relative to fixation target positioning 1002 is shown to illustrate some more detailed characteristics of the square wave jump mentioned in this specification. Figures 2F to 2M Examples of abnormal biphasic and polyphasic SWJs exhibiting the various eye movement behaviors described in this paper are presented.

[0117] As explained above and as Figure 2D As illustrated simply, square wave jerks (SWJ) are forms of fixational eye movements that occur within 150–500 ms, moving away from and back to the fixation point with approximately equal magnitudes. Fixational eye movements (including microsalivations, drifts, and tremors such as these) are believed to improve visibility by preventing neural adaptation to invariant stimuli. Given the importance of detecting and characterizing square wave jerks for some technical forms, the characteristics of square wave jerks in healthy and unhealthy patients will now be described in more detail.

[0118] refer to Figure 2E The diagram shows a theoretical curve of the one-dimensional eye localization 1000 (i.e., the horizontal coordinate of the eye gaze) relative to the fixation target localization 1002 over time. In healthy patients, the SWJ exhibits a clear and accurate single deflection of less than three degrees in the horizontal direction away from the fixation target 1002. This initial deflection may be generally referred to herein as the primary saccadic deflection (“PSD”) 1010.

[0119] Following PSD 1010 is glide phase 1020, in which eye positioning 1000 is maintained. A peak deflection (“PD”) 1014 may occur between PSD 1010 and glide phase 1020. Figure 2E Not shown in the image, but can be found in [reference]. Figure 2FThe reference to peak deflection 1014 should be understood as referring to the initial overshoot of the sacral movement before returning to the glide phase 1020. In healthy patients, PD 1014 is small or negligible. In contrast, Figure 2F The abnormal SWJ was shown, which had a significant PD 1014 after a slow PSD 1010.

[0120] In healthy patients, the gliding phase 1020 can last between approximately 70 ms and 700 ms (average approximately 200 ms), during which only small fibrillation occurs. In contrast, Figure 2F A deformed SWJ is shown with a gliding phase 1020 having a negative slope. After the gliding phase 1020, saccade recovery (“SR”) 1030 returns eye positioning 1000 to fixation target 1002. In healthy patients, SR 1030 is accurate.

[0121] In individuals with mTBI (including those with persistent symptom burden and those with moderate to severe phenotypes), the brain-eye system that manages square wave jerks does not function reliably, and some features of square wave jerks may be observed to have changed.

[0122] It has been established that square wave jumps in some patients with mTBI can be biphasic, meaning that they return to the fixation target with an over-aimed saccade and are recorrected by a third corrective saccade (unlike the classic 'desktop' appearance of square wave jumps) – see [link to relevant documentation]. Figure 2G .

[0123] The inventors have identified that movement can be morphologically diverse in other respects and, in severe cases, may involve a series of multiphasic saccades. For example, some mTBI cases may exhibit multiple deflections away from the target with a larger peak deflection than healthy patients, and failed attempts to return to the target, which can be excessive (overshooting) or insufficient (undershooting). Fibrotic fibrillation and drift may also be present in the glide phase. In describing the characteristics of square wave jumps in detail, the inventors have introduced new terminology to further simplify the description of square wave jumps as biphasic. In fact, complex multiphasic behaviors have been identified, which include categorizable and quantifiable multi-stage interruptions of gaze movements.

[0124] As will be explained later, in some forms of this technique, these features can be measured and used to determine the severity of the condition. For example, saccades can be quantified along with their subsequent gliding phase. When measuring trauma, the complexity of saccade recovery failure and saccade amplitude can be considered in addition to, or as an alternative to, the frequency and / or amplitude of SWJ.

[0125] exist Figures 2F to 2K The image shows some illustrative examples of square wave rapid beats in patients with mTBI.

[0126] In this example of the technique, SWJ can be defined as including primary saccadic deflection (“PSD”) 1010. The measurement of PSD 1010 may include one or more of the following: saccadic velocity (“SV”), saccadic amplitude (“SA”) 1012 (note that microsacca is described in the literature as an eye movement of less than three degrees, while saccadic is greater than three degrees) and peak deflection (“PD”) 1014.

[0127] In this example of the technique, SWJ can be defined as one or more additional saccade deflections following PSD 1010, referred to herein as saccade deflection n (“SDn”) 1016 – that is, additional 'n' deflections away from the target following PSD 1010. The metric of SDn 1016 may include one or more of the following: saccade deflection count (n) (“SDC”), saccade velocity (“SV”), saccade amplitude (“SA”) 1012, and peak deflection (“PD”) 1014.

[0128] In an example of this technique, SWJ can be defined as including a glide phase 1020 between movements. The measure of glide phase 1020 can include one or more of the following: duration (e.g., expected to be approximately 100 ms to 400 ms) and slope (e.g., a flat angle, positive towards recovery, or negative away from recovery). In the example, fibrillation can be measured using the RMS error from the line, a value exceeding a standard deviation threshold (e.g., 1) indicating the presence of anomaly.

[0129] In examples of this technique, SWJ can be defined as including one or more attempts at recovery, such as saccade recovery (“SRn”) 1030 and progressive recovery (“GRn”) 1040, where 'n’ is the number of attempts at recovery. Progressive recovery (“GRn”) 1040 can be distinguished from saccade recovery (“SRn”) 1030 by the rate at which recovery occurs. For example, normal microsaccades occur in less than 10 ms, where most of the time is spent in a 200 ms glide phase (making them appear substantially vertical on the trajectory) before an equally fast recovery. Progressive recovery can occur in the 50-60 ms time range and can be more tilted or curved relative to vertical. The metrics for saccadic recovery (“SRn”) 1030 and / or progressive recovery (“GRn”) 1040 may include one or more of the following: saccadic / progressive recovery count (n) (“SRC” / “GRC”), saccadic / progressive speed (“SV” / “GV”), saccadic amplitude (“SA”), peak deflection (“PD”), and classification of recovery (e.g., accurate, excessive, or insufficient).

[0130] In the examples of this technology and in particular, refer to Figure 2M A SWJ can be defined as including one or more saccade peaks (“SS”) 1050. The reference to saccade peaks should be understood as referring to anomalies in the SWJ where saccade recovery is initiated but unsuccessful and returns to a deflected state, then attempts a more typical recovery are made. Measurements of SS 1050 may include one or more of the following: saccade peak count (n) (“SSC”) and saccade amplitude (“SA”) 1012.

[0131] 7.1.3 Eye Tracking

[0132] This technology relates to apparatus, systems, and methods for analyzing data representing the movement of eye 101. Such data can be obtained by “tracking” the movement of eye 101. Unless the context explicitly states otherwise, the term “tracking” is intended to mean the action of identifying the manner in which eye 101 moves over a period of time. By identifying the movement of eye 101, it is possible to characterize and analyze these movements. In some forms, the movement of eye 101 is tracked by observing the movement of pupil 106. Pupil 106 is an opening through which light enters the internal portion of eye 101, and therefore its positioning indicates the visual direction of the eye.

[0133] The eye can track a target, which can be a stationary target or a moving target. A stationary target can be an object or a representation on a display screen that is held in a fixed position relative to the eye's field of vision for a certain period of time. A moving target can be an object or a representation on a display screen that moves relative to the eye's field of vision; for example, the representation can move on a display screen presented to the eye. The target can be represented by a dot-like object or image (e.g., a small image, such as a dot) on the display screen. Alternatively, the target can be a larger or more complex object or image. The target can be alternatively referred to as a visual stimulus.

[0134] Eye tracking may include measuring and optionally characterizing any errors in the eye's ability to track a target. For example, the error may be the difference in the eye's gaze direction compared to the target's localization. This difference can be represented by any suitable parameter, such as physical distance, a distance equivalent metric (e.g., pixels on a display screen), or an angle representing the angular difference between the eye's gaze direction and the target's direction relative to the eye.

[0135] 7.2 Eye Analysis Device / System

[0136] Figure 3 The diagram illustrates schematic illustrations of certain forms of devices and / or systems 200 for analyzing eye movements according to the present technology. Such devices / systems may also be referred to as eye analysis devices / systems 200.

[0137] The eye analysis system 200 may include an eye tracking system 300 and a data analysis system 400. The eye tracking system 300 may be configured to track the movement of an eye 101 and output data representing the movement of the eye 101. This data may be provided to the data analysis system 400, which may analyze the data representing the movement of the eye 101. The data analysis system 400 may output certain information obtained from the analysis process. Although described below as independent functional systems, in some forms, the eye tracking system 300 and the data analysis system 400 may be implemented in the same one or more physical systems (e.g., a computer or computing network). In other forms, different physical systems may implement the functionality provided by each of the eye tracking system 300 and the data analysis system 400. In some forms, each functional system may be implemented by multiple physical systems.

[0138] 7.2.1 Eye Tracking System

[0139] In some forms of this technology, the eye-tracking system 300 can be any component assembly configured to track the movement of the eye 101 and output data representing the movement of the eye 101. Any suitable eye-tracking system 300 can be used, with reference to... Figure 4 To describe an exemplary form, the figure is a schematic illustration of an eye-tracking system 300 according to an exemplary form of the present technology.

[0140] 7.2.1.1 Display Screen

[0141] exist Figure 4 In the exemplary form shown, the eye-tracking system 300 includes a display screen 310. The display screen 310 may include any means configured to visually present information to a viewer. For example, the information may be in the form of an image. The display screen 310 may be controllable to change the information displayed to the viewer. In particular, the display screen 310 may display a target 303 to the patient. For example, a stationary target or a moving target may be displayed to the viewer. In the case of a moving target, the target 303 may move along a motion path 305. The range of eye movement can be important in detecting certain medical conditions, so in some forms, the display screen is positioned to occupy a large portion of the field of vision of the eye 101, for example, more than 100°. In other forms, the system may explore eye movements closer to the center of the field of vision, in which case the display screen may be smaller. The target 303 may move on the display, so the viewer must move their eyes a significant distance to follow the movement of the target (e.g., up, down, left, and right).

[0142] In some exemplary forms, display screen 310 is an electronic display, such as an LCD, LED, or OLED screen. In some forms, display screen 310 can be displayed to a viewer via a virtual reality (VR) or augmented reality (AR) display system, while in other forms, reflections of display screen 310 can be displayed to a viewer. The information displayed on display screen 310 can be controlled by a processor, which may be included as part of display screen 310 or configured to control display screen 310 via a physical or wireless connection. In some forms, display screen 310 is included as part of an electronic device (e.g., a portable electronic device 350, such as a smartphone, tablet, laptop, etc.).

[0143] In some forms, the display screen 310 may be self-illuminating; for example, the display screen 310 may include light-emitting elements (such as LEDs). In other forms, the display screen 310 may be non-self-illuminating; for example, the display screen may use electronic ink (e-ink) to display information. In such forms, a separate light source can be used to illuminate the display screen. It should be understood that the level of illumination should be appropriate for camera settings (e.g., frame rate, exposure, and resolution) to ensure that the image is sharp and not blurry.

[0144] In some forms, the eye-tracking system 300 may include multiple displays.

[0145] 7.2.1.2 Camera

[0146] In some forms of this technology, the eye-tracking system 300 may include a camera 320. The camera 320 may include any optical device configured to capture and record visual images. A suitable frame rate for the camera 320 may be guided by the Nyquist-Shannon sampling theorem, which states that the sampling frequency should be at least twice the frequency of the recorded movement. It is believed that frequencies below approximately 100 Hz carry almost no information about fixational eye movements; therefore, in some forms, the eye-tracking system 300 uses a camera 320 with a minimum frame rate of approximately 200 Hz. For example, in one form, the camera 320 has a frame rate of 240 Hz. This provides a 4.2 ms time interval between frames and thus allows the capture of eye movements with time frames longer than this timescale, including, for example, subtle fixational movements (such as SWJ) and polyphase SWJ components, whose duration can be as short as approximately 5 ms.

[0147] Furthermore, the resolution of camera 320 can be high enough to capture eye movements detected and analyzed in the analysis method. For example, microsaccades are typically less than 3 degrees of visual angle. In some forms, for example, the camera can have an accuracy of approximately 0.01 degrees of visual angle. In some forms, for example, camera 320 can have sufficient resolution, and camera 320 can be positioned relative to eye 101 such that the size of the image capturing eye 101 is at least 500x500 pixels.

[0148] The captured images can be displayed on a screen, which in some forms may be a screen 310 included as part of the eye-tracking device 200, while in other forms, the camera may be configured to transmit images to another device, which itself may include a screen for displaying the images or memory for storing the images for display elsewhere. Images can be transmitted via wired or wireless connections, for example, to a screen remote from the eye-tracking system 300. Images can be displayed on the screen in real-time, near real-time, or at a time later than when the image was captured by the camera. In some forms, camera 320 includes memory configured to store visual images. It should be understood that in some forms, camera 320 is a digital camera, and a reference to an image may mean data recorded by the camera to represent the image.

[0149] In some exemplary forms, camera 320 may be included as part of an electronic device (e.g., portable electronic device 350, such as a smartphone, tablet, laptop, etc.). In such a form, camera 320 includes a display screen 310, which may be configured to display images captured by camera 320.

[0150] 7.2.1.3 Processor

[0151] In some forms of this technology, the eye-tracking system 300 may include one or more processors. Although multiple different processors may be used, the processors may operate together or may be considered as operating together as functional units. For the purposes of the following discussion, and for convenience, the description will refer to a single processor configured to perform any of the functions described, but it should be understood that multiple processors may be used in some forms.

[0152] The processor can be configured to generate data representing the movement of the eye 101 from image data captured by the camera 320. The processor can be included as a device similar to the camera (e.g., Figure 4 The processor may be part of a portable electronic device 350 in the form shown, or it may be located away from the camera 320 and receive image data from the camera 320, for example, via a wired or wireless communication link.

[0153] The processor can also control the display screen 310 to present information to the patient, such as a fixed or moving target for the patient to gaze at during eye tracking. The processor can be configured to generate data representing the positioning of the target 303. A suitable motion protocol can be provided to the processor, for example, from memory or via a suitable communication link.

[0154] The processor can be further configured to output data representing the movement of eye 101. The data can be output by the processor via a suitable communication network or via an output device (e.g., display screen 310). Alternatively, the information can be stored in memory (e.g., the memory of portable electronic device 350) for later output. The output data representing the movement of eye 101 can be a series of data representing the positioning and / or movement of eye 101 (e.g., pupil 106) at multiple times. The data can be output in any suitable format (e.g., a CSV file).

[0155] The processor can be further configured to output data representing the location of target 303 (if the target is moving, this data may include data representing the movement of target 303). This data can occur simultaneously with data representing the movement of eye 101, i.e., such that the target data represents the movement of target 303 while eye 101 is gazing at it and the eye movement is captured in eye movement data. The processor can output the data by sending information via a suitable communication network or by outputting information via an output device (e.g., display screen 310). Alternatively, the information can be stored in memory (e.g., the memory of portable electronic device 350) for later output. The output data representing the movement of target 303 can be a series of data representing the location and / or movement of target 303 at multiple times. The data can be output in any suitable format (e.g., a CSV file).

[0156] In some forms, the processor can output data representing the movement of eye 101 together with data representing the movement of target 303 (e.g., in the same data file). The output data can be timestamped so that the position of eye 101 is recorded relative to target 303 at each time point.

[0157] 7.2.2 Data Analysis System

[0158] Figure 5This is a schematic illustration of an exemplary data analysis system 400 according to one form of the present technology. The data analysis system 400 may include a hardware platform 402 that manages the collection and processing of data from an eye-tracking system 300, such as data representing movement of eye 101 and movement of target 303. The hardware platform 402 may include a processor 404, a memory 406, and other components typically present in such computing devices. The hardware platform 402 may be located locally on the eye-tracking system 300, or it may be located remotely from the eye-tracking system 300 and receive data via a suitable communication link, such as a network 416. In the exemplary form of the illustrated technology, the memory 406 stores information accessible by the processor 404, including instructions 408 executable by the processor 404 and data 410 that can be retrieved, manipulated, or stored by the processor 404. The memory 406 may be any suitable device known in the art capable of storing information in a manner accessible by the processor 404, including computer-readable media or other media storing data readable by electronic devices.

[0159] Processor 404 may be any suitable device known to those skilled in the art. Although processor 404 and memory 406 are illustrated as being within a single unit, it should be understood that this is not intended to be limiting, and the respective functions described herein may be performed by multiple processors and memories, which may be remote from each other or the processing system 400, or may not be remote from each other or the processing system. Instructions 408 may include any set of instructions suitable for execution by processor 404. For example, instructions 408 may be stored as computer code on a computer-readable medium. Instructions may be stored in any suitable computer language or format. Data 410 may be retrieved, stored, or modified by processor 404 according to instructions 410. Data 410 may also be formatted in any suitable computer-readable format. Again, although data is illustrated as being contained in a single location, it should be understood that this is not intended to be limiting – data may be stored in multiple memories or locations. Data 410 may also include a record 412 of control routines for various aspects of system 400.

[0160] Hardware platform 402 can communicate with display device 414 to display the results of the analyzed data. In some forms, display device 414 may be a display screen 320 included as part of eye-tracking system 300. Hardware platform 402 can communicate via network 416 with one or more other devices (e.g., user devices, such as tablet computer 418a, personal computer 418b, or smartphone 418c, or other devices including sensors) or one or more server devices 420 having associated memory 422 for storing and processing data collected by local hardware platform 402. It should be understood that server 420 and memory 422 can take any suitable form known in the art, such as a “cloud-based” distributed server architecture. Network 416 may include various configurations and protocols, including the Internet, intranet, virtual private network, wide area network, local area network, private network (whether wired or wireless) using one or more company-proprietary communication protocols, or combinations thereof.

[0161] After analyzing the data representing the movement of eye 101, the data analysis system 400 can be configured to output certain information obtained from the analysis process, examples of which will be described in more detail below. The information can be output by the data analysis system 400 via network 416 or via an output device (e.g., display device 414, tablet computer 418a, personal computer 418b, or smartphone 418c). Alternatively, the information can be stored in a memory (e.g., one or both of memory 406 or 422) for later output from the data analysis system 400.

[0162] In some forms, the hardware platform 402 of the data analysis system 400 may include computing devices, such as laptops or PCs. In other forms, the hardware platform 402 may include multiple computing devices configured to operate together to perform data analysis / processing.

[0163] 7.3 Methods for analyzing eye-tracking data

[0164] In some forms of this technology, one or more methods are provided for analyzing eye-tracking data (i.e., data representing the movement of eye 101). Unless otherwise stated, it should be understood that these analytical methods can be derived from, as described above and relative to, [the methods described above]. Figure 5 The data analysis system 400 described is implemented.

[0165] 7.3.1 Methods for assessing the presence of TBI

[0166] Some forms of this technology relate to methods, systems, and apparatus for assessing the presence of traumatic brain injury (TBI) in a patient. In some forms, the assessment is for the presence of mild traumatic brain injury (mTBI) (which may alternatively be referred to as concussion).

[0167] Figure 6 This is a flowchart outlining an exemplary method 600 for assessing the presence of TBI in a patient. Method 600 can be performed by a data analysis system 400 (such as...) Figure 5 The method is performed as illustrated in the diagram, but in some forms, some steps of the method (e.g., preprocessing steps) may be performed by the processor of the eye-tracking system 300. The method may include: receiving eye data representing movement of the patient's eye 101, analyzing the eye data to determine an indication of the presence of the patient's TBI, and outputting that indication. Each of these steps will be explained in more detail below.

[0168] 7.3.1.1 Target Display

[0169] In the first step 601 of an exemplary method according to the present technology, the display screen 310 is controlled to display a target 303 to the patient for viewing by the patient's eyes 101. In some forms, one or both of two types of targets 303 may be displayed to the eyes 101: a stationary target and / or a moving target.

[0170] The case where target 303 is a fixed target can be referred to as a fixation task. That is, a fixed target 303 is displayed on display screen 310, and the patient is instructed to gaze at the fixed target. Display screen 310 can be controlled to display the fixed target at an off-center location relative to the patient. That is, a location not directly in front of the patient within the line of sight when the eyes are looking straight ahead. In some forms, off-center positioning can be considered to include an angle of up to 120° opposite the central fixation point. It should be understood that, considering the distance of display screen 310 from the eye 101, the display screen needs to be large enough so that target 303 can be displayed at this positioning.

[0171] In some forms, the display screen 310 can be controlled to display multiple fixed targets sequentially (i.e., one after another). Each fixed target can be displayed for a random duration, and / or the time interval between each displayed fixed target can be random. The selection of the random duration and / or random time interval can be limited to a certain maximum and minimum time period. Alternatively, multiple fixed targets can be displayed at different locations within the eye's field of vision. The selection of the display position of each fixed target on the display screen 310 can also be randomly selected from a trajectory on the display screen 310 having eccentric coordinates relative to the eye 101.

[0172] The case where target 303 is a moving target can be referred to as a smooth following task. That is, the moving target 303 is displayed on display screen 310, and the patient is asked to gaze at the moving target and maintain this gaze as the patient moves around on display screen 310. Display screen 310 can be controlled such that the moving target follows a predetermined path on the screen, and in some forms, the moving target 303 can outline a specific shape (e.g., a circle or an ellipse), and the moving target 303 can repeatedly outline the same shape. Figure 7A and Figure 7B In the example of the display screen 310 shown, the target 303 can be any icon (e.g., a dot), and the target follows a motion path 305 on the display screen 310. The motion path 305 can be elliptical, circular, sinusoidal, zigzag, or any other motion path deemed suitable for testing a patient's eye tracking. The target 303 can move in a clockwise or counterclockwise direction. Figure 7B In the exemplary display 210, the positioning of target 303 along its motion path 305 over time is shown in successive positions 303a, 303b, 303c, and 303d. The figure also shows the projection of this path over time onto the X-axis and Y-axis. In this example, the elliptical motion path 305 of target 303 allows the eye to move in both the X-direction (corresponding to points 303a and 303c) and the Y-direction (corresponding to points 303b and 303d) to train the range of motion of the eye. In some examples, for some test scenarios, the motion path 305 can be changed to be flat or off-axis. The repetitive movement of target 303 along the motion path 305 smoothly trains the human eye / brain interface. In the example shown, the movement on the X and Y axes is inherently sinusoidal independently, but in other examples it can be modified to a sawtooth or square wave.

[0173] In some formats, the speed and amplitude of the movement of the target 303 on display 210 can be altered over time. This increases the cognitive stress and physiological demands on the subject during the smooth follow-up task. Increased cognitive stress will increase the severity of any symptoms, and accelerated movement can help identify breakpoints, as explained later. In some formats, the subject can be tested in a series of trials, and the speed of movement of the target 303 or the rate of acceleration of the target 303 can be gradually increased in each successive trial.

[0174] 7.3.1.2 Eye Tracking

[0175] At step 602, which can occur simultaneously with step 601, the movement of eye 101 is tracked while performing a gaze and / or smooth tracking task. An eye-tracking system 300, such as that described above, can be used for this step, and camera 320 can capture images of eye 101 during the task while eye 101 is gazing at target 303.

[0176] The eye-tracking system 300 can generate data representing eye movement from images captured by the camera 320. This data can be provided to the data analysis system 400 using any suitable data communication protocol. The data representing eye movement from images captured by the camera 320 can include eye data representing eye movement and target data representing target positioning.

[0177] In some forms, a calibration step can be performed in which the patient gazes at a target at one or more locations on display screen 310. Data generated from this step can determine a transformation used to change the detected eye positioning coordinates to the actual gaze coordinates. This transformation can be applied to eye movement data, as explained later.

[0178] 7.3.1.3 Preprocessing

[0179] At step 603, the data analysis system 400 receives eye data representing the movement of eye 101 during the task. The data analysis system 400 may additionally receive target data representing the location of target 303 during the task (if the data analysis system 400 does not already have this data), for example, this target data may be received from the processor of the eye tracking system 300.

[0180] In some forms, eye data representing the movement of eye 101 during the task and / or target data representing the location of target 303 during the task may need to be preprocessed at step 604 before further analysis can be performed. An example of such preprocessing will now be described. It should be understood that while the data analysis system 400 is described as performing these preprocessing steps, in other forms, the eye-tracking system 300 may perform some of these steps before providing the data to the data analysis system 400.

[0181] In one exemplary step, two datasets (i.e., representing eye movement and representing target localization / movement) are processed so that they can be easily compared and analyzed together. For example, a primary extraction function can be applied to load the datasets, align them, and organize them. This function can also facilitate further analysis of each task in the protocol. More precisely, data representing the localization of target 303 during the eye-tracking protocol (which may consist of the target position and timing on display 310) is saved as annotations, which are loaded and sampled at a frame rate consistent with the frame rate of the data representing eye movement.

[0182] Regarding eye data representing eye movement, pupil-tracking coordinates can be loaded and resampled before being merged with concurrent data representing target localization. In some exemplary forms of this technique, the gaze direction can be determined by detecting the localization of the iris relative to the sclera. Detecting the localization of the iris better handles upper eyelid occlusion (as may occur when tracking the pupil) and any reflections compared to detecting the localization of the pupil. The radius of the iris is essentially fixed compared to the pupil, which changes size as it dilates. Furthermore, detecting the external curvature of the iris (which is greater than the curvature of the pupil) provides excellent sampling. Therefore, detecting the curvature of the iris under perspective can be used to estimate the orientation of the eyeball in the orbit with good accuracy. For example, in some forms using this method, the orientation of the eye can be determined to a subpixel estimate of approximately 0.25% of a pixel, which is roughly equivalent to using a 2000x2000 pixel camera. For a typical adult eye with a width of 24 mm, this allows for a tracking accuracy of approximately 0.06 degrees for the eye center. This is below the minimum amplitude of typical eye movement and thus improves the accuracy of the results. For example, the minimum amplitude of SWJ can be approximately 0.2° to 1.5°. The components of multiphase SWJ can be as small as 0.1°, and useful information can exist even below this scale in terms of quantification.

[0183] In some forms, this can be achieved by applying pre-filtering to the iris to normalize its color and texture, and applying an object detection algorithm to the eye data to find the boundary location of the iris (e.g., a one-off object detector). An example of a one-off object detector that can be applied to find the iris boundary location is the YOLO (YouOnly Look Once) v7 real-time object detector. The one-off object detector can first be trained using an appropriate eye dataset where all existing irises are accurately labeled. Subsequently, a partial curve detector can be applied to estimate the full elliptical shape of the iris, and thus estimate the center of the iris and pupil (even if the pupil is obscured by the eyelid). Determining the center of the iris and pupil can use existing techniques known to those skilled in the art to provide the gaze direction. In some forms, the method may include applying a secondary tracker that observes the curvature of the pupil to estimate its diameter, which can change size as the pupil dilates.

[0184] In some forms, eye data can be processed to determine the location of one or more corners of the eye 101. This can be achieved using existing corner detection techniques. If a change in corner location is detected, the determined pupil location, and therefore the determined gaze direction, can be appropriately adjusted.

[0185] In some forms, blink detection can be performed. In some forms, blink detection is performed to remove eye data captured during blinking from further analysis. This avoids any anomalous influence of eye movement that may occur during blinking on the results. In other forms, blink detection can be used to measure blinks, such as the time between blinks and / or blink frequency. In some forms, one or more of these can be useful metrics to help evaluate mTBI. In some forms, blink detection can include applying an object detection algorithm to an eye dataset (e.g., a one-off object detector). An example of a one-off object detector that can be applied to an eye dataset for blink detection in some forms is the YOLO (You Only Look Once) v7 detector. Similarly, a one-off object detector can first be trained using an appropriate eye dataset (where all present eyelids are accurately labeled). In some forms, the object detection algorithm can additionally be able to detect the eyelash portion of the eyelid in the image data of eye101. Because the eyelid positioning changes over time during blinking, the change in the vertical positioning of the eyelid can be used to measure eyelid velocity / blink velocity. In other forms, blinking can be detected by detecting the absence of the pupil and assuming that this absence is due to eyelid closure during blinking.

[0186] In another optional preprocessing step, the coordinates of eye positioning in the eye movement data can be transformed into gaze coordinates by applying any necessary transformations determined from the calibration step.

[0187] 7.3.1.4 Determination of Eye Movement Measurement

[0188] At step 604, the data analysis system 400 analyzes the data. As will be explained, this analysis can determine the presence of TBI in the patient (e.g., mTBI).

[0189] In some forms, the analysis may include: determining one or more measures of movement of eye 101, and determining an indication of the presence of TBI in the patient based on said one or more measures. In different forms of this technique, the determination may be based on each of the measures alone. Alternatively, in other forms, the determination may be based on a combination of two or more measures. Any combination of two or more measures may be used, and the form of this technique is not limited to any particular combination of measures. It should be understood that the accuracy of the determination may be improved if more measures are used in determining the condition.

[0190] Generally, the indication for the presence of a TBI can be based on comparing each metric in the metrics to one or more thresholds. If a corresponding metric is above or below its corresponding threshold, the presence of a TBI can be determined. It should be understood that a TBI is not a condition diagnosed in this binary manner, but thresholds can be selected to provide a specific level of confidence in the presence or absence of the condition. In some forms, multiple thresholds can be used for each metric, where each threshold provides a range of confidence levels regarding the presence or absence of the condition. Such information can be passed as output, as described in more detail below.

[0191] The following sections explain exemplary measures that can be determined from data, how they can be determined, and how they can be used to determine an indication of the presence of TBI. While exemplary methods for determining these measures have been explained, it should be understood that other methods may be used to determine these measures in other forms of the art. For example, in some forms, the data analysis system 400 may perform computer analysis of eye movement data based on algorithmic methods, such as applying one or more algorithms to: extract features from the eye movement data, classify the features of the eye movement data, derive values ​​for measures of the eye movement data, and derive and output an indication of the presence of TBI in the patient.

[0192] Furthermore, machine learning methods can be employed to analyze the large amounts of data generated during eye tracking and to determine metrics from that data. For example, in some forms, a machine learning model can be trained to determine an indication of the presence of TBI in a patient. Such a machine learning model can include one or more models of one or more machine learning algorithms (e.g., deep learning models using artificial neural networks). During training, the machine learning algorithm can be trained using learning techniques generally known in the field of machine learning by providing training data as input. In an example, the training data can include eye movement data as described herein, with examples from healthy patients and patients experiencing TBI. In this example, a machine learning model can be trained to identify metrics in the eye movement data described herein, from which an indication of the presence of TBI in the patient can be determined. In an alternative example, a machine learning model can be trained to infer an indication of the presence of TBI in the patient directly from the eye movement data.

[0193] 7.3.1.5 Micro-scan video rate

[0194] In an exemplary form, eye data representing eye movement during a fixation task is analyzed to determine a measure of the frequency of microsaccades during the fixation task.

[0195] In some forms, saccades are filtered from eye movement data using any suitable computational algorithm, such as the previously described velocity- and distance-based Hidden Markov Model (HMM) (Salvucci, DD, & Goldberg, J. H, 2000, Identifying fixations and saccades in eye-tracking protocols, paper presented at the Proceedings of the 2000 symposium on Eye tracking research, Palm Beach Gardens, Florida, USA, https: / / doi.org / 10.1145 / 355017.355028). This process clusters saccades into two groups based on the distance the eye moves between the beginning and end of each saccade: small saccades and large saccades. In some forms, the model may apply two states: one state for fixation and one state for any other eye movement. Alternatively, in other forms, the model can apply three states: one state for gazing, one state for the clipped value (after high-pass filtering), and one state for any other eye movement.

[0196] In some forms, this method is used to automatically label training data, which is then fed into a transformer model (i.e., a neural network that learns context and thus meaning by tracking relationships in successive data). The transformer model can be constructed by encoding eye movements for each frame as a series of linear vectors. These movements can be labeled as different categories as part of the training data. Different categories can be, for example, smooth following, fixation, saccades, microsaccades, tremors, and square wave jumps (or components thereof). While some training data can be automatically labeled, some manual overwriting can be applied to correct for any errors and fine-tune the labeling. The performance of the transformer model improves as more data is collected and labeled from patients. The transformer model can perform significantly better than traditional heuristics because it can be trained on large amounts of training data from which it can derive context from a series of movements preceding and following them. Furthermore, the model can be able to account for secondary influencing factors (such as current pupil positioning), which can be important because with sufficient data, certain parts of the eye movement range can affect performance.

[0197] The mean and standard deviation of the states can be initialized using percentiles of the input data. Information from annotations in the target localization data can be used to further classify saccades; for example, corrective saccades can be identified if they immediately precede the start of target display. Minimal saccades (e.g., saccades with amplitudes below a predetermined threshold) can be classified as microsaccades and / or microsaccade intrusions, and can occur along horizontal or vertical meridians. Saccades are voluntary actions of the eye toward a target based on human reaction time, while microsaccades are involuntary fixational movements that can typically have similar speeds but with amplitudes that may be an order of magnitude lower (e.g., from 0.2° to 1.5°) and are not limited by reaction time. In some forms, both measures can be used in the analysis.

[0198] In some forms, the frequency of microsaccades occurring during a fixation task can be determined when a target is presented to the patient. In cases where multiple fixed targets are presented to the patient for fixation, the microsaccade rate can be calculated as the average of the microsaccade rates for multiple individual targets.

[0199] In some forms, a positive diagnosis of TBI can be made if the frequency of microsaccades exceeds a certain threshold. In some forms, a diagnosis of mild TBI (i.e., mTBI or concussion) can be made if the frequency of microsaccades exceeds a first lower threshold but is below a second higher threshold. In such forms, a diagnosis of moderate or severe TBI can be made if the frequency of microsaccades exceeds a second higher threshold. In some forms, the threshold can be the absolute value of the microsaccade rate determined through experimental observation of previous patients. In other forms, the threshold can be calculated from a microsaccade rate control value determined experimentally to be "normal." For example, the threshold can be a certain percentage higher than the control value. This percentage can also be determined through experimental observation of previous patients.

[0200] 7.3.1.6 Error between gaze and target

[0201] In an exemplary form, eye data and target data representing eye movement during a fixation task are analyzed to determine a measure of the average error between the localization of the fixed target and the localization of the patient's gaze during the fixation task. For example, the measure of average error could be the root mean square error (RMSE) between the patient's gaze and the localization of the target, which may also be referred to as the root mean square deviation (RMSD). In other forms, any other statistical measure of the error between the localization of the fixed target and the localization of the patient's gaze during the fixation task can be used. When presenting the patient with multiple fixed targets for fixation, the measure of average error can be calculated as the average of the errors for multiple individual targets.

[0202] The error between the localization of a fixed target and the localization of the patient's gaze can be represented by any appropriate parameter, such as physical distance, a distance equivalent measure (e.g., pixels on a display screen), or an angle representing the angular difference between the direction of the eye's gaze and the direction of the target relative to the eye.

[0203] If calibration errors exist in the eye-tracking system 300 that measures the gaze direction of eye 101, the measurement of the average error may be skewed. To address this possibility, in some forms, a best-fit line can be identified on the participant's median gaze localization. It is assumed that the median gaze localization points towards the target, and the gaze data is thus adjusted. This method can correct for situations where the user and device may have moved relative to each other after calibration.

[0204] In some forms, a positive diagnosis of the presence of TBI can be made if the measure of the average error between the localization of the fixed target and the localization of the patient's gaze exceeds a certain threshold. In some forms, a diagnosis of mild TBI (i.e., mTBI or concussion) can be made if the measure of the average error exceeds a first lower threshold but is below a second higher threshold. In such forms, a diagnosis of moderate or severe TBI can be made if the measure of the average error exceeds a second higher threshold. In some forms, the threshold can be the absolute value of the average error determined through experimental observation of previous patients. It should be understood that such an absolute value can depend on the specific settings of the device used, such as the size of the display screen 310 and the distance between the display screen and the eyes 101. In other forms, the threshold can be calculated from an average error control value that has been experimentally determined to be "normal". For example, the threshold can be a certain percentage higher than the control value. This percentage can also be determined through experimental observation of previous patients.

[0205] 7.3.1.7 Square wave jump

[0206] Square wave jerks were selected as a strong candidate for eye movement metrics associated with mTBI due to their involuntary nature and association with various neuropathologies. They are thought to result from the failure of cortical cells to suppress saccades during fixation events.

[0207] In an exemplary form, eye data representing eye movement during a fixation task are analyzed to determine a measure of one or more characteristics of square wave jerks (SWJ) during the fixation task. For ease of understanding, references to SWJ analysis in this section should be understood to include all forms of SWJ described herein, including typical SWJ, aberrant SWJ, biphasic SWJ, and polyphasic SWJ.

[0208] In some forms, determining the one or more metrics may include classifying each of two or more phases. For example, a phase may be classified into two or more of the following: primary saccade deflection, subsequent saccade deflection, saccade recovery, progressive recovery, saccade peak, and glide phase.

[0209] In some forms, determining the one or more metrics may include quantizing one or more features of each of the two or more phases.

[0210] In some forms, the determination of the frequency and / or amplitude of SWJ can rely on identifying saccades using the methods explained above, and for each pair of saccades, identifying any one or more of the following: 1) the magnitude index, i.e., a measure comparing the magnitudes of the two saccades in the pair; 2) the angle between the two saccade vectors; and 3) the intersaccade interval (ISI), i.e., the time between saccades in the pair.

[0211] In eye movement data, the angle between saccades is between 0° and 180°, focusing on the horizontal meridian, and the saccade amplitude is taken into account.

[0212] In some forms, the model can be fitted to a dataset of paired or sequential saccades (identified as SWJ), for example, by clustering its features using a diagonal covariance matrix, thus avoiding capturing correlations between features. The model can be fitted using machine learning methods from suitable patient training data, and eye movements can be classified using a Bayesian Gaussian Mixture Model (GMM), an exponential Gaussian model, or a gamma distribution model of the inter-saccade intervals. In the example using a GMM, the GMM component with the mean closest to 180° can be considered the SWJ cluster. Certain features of this cluster (e.g., mean and variance) can be determined, and these features can be used later to identify SWJ. In some forms, other components may not be used.

[0213] Subsequently, during detection, a Hidden Markov Model (HMM) with two components can be used to assign paired saccades to either the SWJ state or the "other" state. The HMM can apply a Gaussian distribution to the observations (paired saccades). The SWJ state can be initialized with the mean and variance of the SWJ components from the GMM. The "other" state can be initialized with the mean and variance of the entire training dataset.

[0214] The transition matrix can be set to force an SWJ state to be followed by an "other" state. The "other" state can transition to an SWJ state or itself with a probability of 0.5. Thus, a saccade in a pair that is part of an SWJ cannot be identified as part of another saccade that immediately follows it.

[0215] In some forms, detecting SWJ may then include: using the HMM to run the Viterbi algorithm to obtain the most probable sequence of states.

[0216] In some forms, all eye movement classifications can be performed using a transducer model based on bidirectional attention. This provides a unified model for classifying eye movements over time. To train the transducer, sample eye data can be encoded into a series of movement vectors labeled with different categories, such as smooth following, fixation, saccades, microsaccades, tremors, and square-wave jumps. This labeling can utilize all the described automated processes. Subsequent processes can exist where experts can reclassify any label as needed. A transducer trained on this labeled data can leverage the context of a series of movements moving forward and backward over time. Furthermore, the model can consider different eye types and secondary influencing factors (such as current pupil positioning), which can be important because, with sufficient data, certain parts of the eye movement range can affect performance. The model can outperform human experts when it is computationally more efficient and robust than statistical models used to perform automated labeling on the training data. The model can also be retrained as new data is collected, further improving classification accuracy.

[0217] The methods for identifying SWJ have been explained, and therefore, in some forms, the frequency of SWJ occurrence and / or measurements of the defined components of SWJ when a target is presented to a patient during a fixation task can be used as an inference of impairment. In the case of presenting a patient with a series of fixed targets for fixation, the SWJ frequency and / or amplitude can be calculated as an average of the SWJ frequency / amplitude for multiple individual targets, while the frequency of SWJ error is cumulative.

[0218] In some forms, a positive diagnosis of TBI can be made if the frequency and / or amplitude of SWJ exceeds a certain threshold. In some forms, a diagnosis of mild TBI (i.e., mTBI or concussion) can be made if the frequency and / or amplitude of SWJ exceeds a first lower threshold but is below a second higher threshold. In such forms, a diagnosis of moderate or severe TBI can be made if the frequency and / or amplitude of SWJ exceeds a second higher threshold. In some forms, the threshold can be the absolute value of the frequency and measurement of SWJ and failure determined through experimental observation of previous patients. In other forms, the threshold can be calculated from a control value of SWJ frequency determined experimentally to be “normal.” For example, the threshold can be a certain percentage above the control value. This percentage can also be determined through experimental observation of previous patients.

[0219] In some forms, the indication for determining the presence of TBI in a patient can be based at least in part on the cumulative sum of SWJ occurrences. In one example, biphasic or multiphasic SWJs can be counted as a single SWJ occurrence. In other examples, each phase of a biphasic or multiphasic SWJ can be counted as an SWJ occurrence.

[0220] In some forms, a higher number of phases in a SWJ can be assigned a higher weight. In some forms, hierarchical weighting can be applied at least in part based on the complexity of the SWJ. In some forms, weighting can be at least in part based on the number of phases in the SWJ. For example, a higher weight can be assigned to an SWJ with a higher number of phases.

[0221] For example, a hierarchical weighted approach can be applied, where:

[0222] 1. A typical SWJ is given a weighted W1;

[0223] 2. Abnormal SWJs are given a weighted W2;

[0224] 3. Multiphase SWJ[2] (i.e., determined to have two phases) is given a weighted W3;

[0225] 4. Multiphase SWJ[3] is weighted by W4;

[0226] 5. The multiphase SWJ[4] is weighted by W5; and

[0227] 6. The multiphase SWJ[n] is given a weighted W n+1 ,

[0228] Where W1 < W2 < W3 … < W n+1 .

[0229] In some forms, weighting can be adjusted based on one or more measures of one or more features of the associated phase. For example, weighting can be biased based on features such as the magnitude of peak deflection.

[0230] For example, algorithms that implement this weighting could include:

[0231] Indication of the presence of TBI = (typical SWJ) 总和 * W1) + (Deformed SWJ) 总和 * W2) + (multiphase SWJ[2]) 总和 * W3) + … + (polyphase SWJ[n]) 总和 * W n+1 )

[0232] In the example implementing this algorithm (where W1 = 1, W2 = 2, W3 = 3, W...),n+ = n+1), during which two healthy SWJs, four abnormal SWJs, two polyphasic[2] SWJs and one polyphasic[4] SWJ can be identified. Therefore, the score indicating the presence of TBI will be calculated as follows: (2*1) + (4*2) + (2*3) + (1*5) = 2 + 8 + 6 + 5 = 21.

[0233] As a further example, where each phase is counted as occurring, the algorithm for implementing this weighting could include:

[0234] Indication of the presence of TBI = (typical SWJ) 总和 * W1) + (Deformed SWJ) 总和 * W2) + (multiphase SWJ[2] phase) 总和 * W3) + … + (multiphase SWJ[n] phase) 总和 * W n+1 )

[0235] In this example, phase 总数 It is the number of phases determined in SWJ.

[0236] In the example (where W1 = 1, W2 = 2, W3 = 3, W...), n+ = n+1), during which two healthy SWJs, four abnormal SWJs, two polyphasic[2] SWJs and one polyphasic[4] SWJ can be identified. Therefore, the score indicating the presence of TBI will be calculated as follows: (2*1) + (4*2) + ([2*2]*3) + ([1*4]*5) = 2 + 8 + 12 + 20 = 42.

[0237] The higher the total score (i.e., the more pronounced the symptoms of TBI), the more severe the patient's condition. As noted above, the score can indicate relative severity based on comparison with experimentally determined data.

[0238] It should be understood that alternative ways of constituting metrics exist, and the examples provided herein are not intended to limit the scope to all forms of this technique.

[0239] 7.3.1.8 Interruption when following a moving target

[0240] In an exemplary form, eye data and target data representing eye movement during a smooth following task are analyzed to determine a measure of smooth following disruption when following a moving target.

[0241] In some forms, the detection of gaze dynamics during the acceleration phase of a smooth tracking task is considered a change-point detection (“breakpoint”) problem. The method may include analyzing data collected during the smooth tracking task and using rank statistics and dynamic programming (iteratively storing computed values ​​for subsequent analysis) to identify changes in the patient’s fixation accuracy during smooth tracking to search for changepoints, such as optimal unique changepoints. Exemplary suitable algorithms are mentioned below. Breakpoints can be determined via one or more of these algorithms (or any other suitable algorithm) that use rank statistics to find a unique breakpoint for each individual. In some forms, the algorithm essentially plots the overall error distribution as a function of time and then searches for points in the graph that are divided into two states. In some forms, the measure of the break may be the time prior to the break occurring in the smooth tracking task. A break may occur when the accuracy of the eye following a moving target falls below a certain threshold. More specifically, the method may include identifying changes in the error distribution between the detected localization of the eye’s gaze and the localization of the moving target on display 310. In some forms, these errors can be calculated from eye movement and target movement data as the Euclidean distance between the target's location on display 310 and the patient's gaze location on display 310 at each time point. In other forms, the errors can be calculated as some other distance measure between the target location and the gaze location, or as an angular error between these locations.

[0242] In some forms, the speed of eye movement during a smooth tracking task can also be calculated. Speed ​​can be calculated as a distance measurement per unit time or an angle measurement per unit time. Calculating eye movement speed allows for the calculation of a measure of the smoothness of the user's eye movement, i.e., how quickly the speed of eye movement is maintained and how uniformly the direction changes. This can be useful as an additional measure to be calculated besides the error between target localization and gaze localization, as the patient's gaze may be smoothly tracked or exhibit brief jumps, while still achieving low Euclidean errors (i.e., accuracy and hysteresis analysis). However, smoother movement with uniform speed (i.e., higher smoothness) indicates better performance. In some forms, one or more filters can be applied to the eye data to reduce the noise level of the eye data in a smooth tracking task, making it possible to assess the smoothness of the movement. In some forms, linear quadratic estimation, such as a Kalman filter, can be applied.

[0243] In some forms, algorithms can be applied to detect a decline in the eye's ability to follow a moving target. This algorithm can also be referred to as a change-point detection algorithm. In some forms, exemplary algorithms can rely on rank statistics, as described in Lung-Yut-Fong, A., Lévy-Leduc, C., & Cappé, O, 2015, "Homogeneity and change-point detection tests for multivariate data using rank statistics," Journal de la Société Françaisede Statistique, 156(4), 133-162. Exemplary algorithms can also use dynamic programming (i.e., iteratively storing computed values ​​for subsequent analysis) to search for the optimal unique change point. In some forms, change point detection algorithms can be implemented in the Ruptures Python package (Truong, C., Oudre, L., & Vayatis, N., 2020, Selective review of offline change point detection methods, Signal Processing, 167, 107299).

[0244] Figures 8A to 8D These are illustrations of radially transformed eye-tracking data in a smoothing follow-up task according to one form of the present technology. The figures illustrate eye-tracking data in a smoothing follow-up task where a target 303 moves along a circular path on a display screen 310. The graphs plot the radii of the target and gaze point relative to the angular positions of the target / gaze point, where the angular positions are radially transformed so that the target is always at a 0° angle. In these graphs, black dots 810 represent the positioning of target 303 on display screen 310. Blue dots 820 represent the eye gaze positioning on display screen 310. Due to the radial transformation, the coordinates of the blue dots 820 in the graphs represent the error relative to the target black dots 810. Figure 8A The image shows eye-tracking data during the gaze phase of the mission before target 303 begins to move. Figure 8B The image shows eye-tracking data during a slow phase of smooth following when target 303 moves at a constant speed in a circular path. Figure 8C The image shows eye-tracking data during the acceleration phase of smooth following when target 303 is moving at an accelerating speed within a circular path. Figure 8CThe data shown is from before the patient was determined to have reached the breakpoint. Figure 8D The image shows eye-tracking data during the acceleration phase of smooth tracking when the target 303 moves at an accelerated speed in a circular path and after the patient has been determined to have reached the breakpoint.

[0245] Figures 9A to 9D These are illustrations of eye-tracking data in another form of smoothing follow-up task according to the present technology. These figures illustrate eye-tracking data in a smoothing follow-up task where a target 303 moves along a circular path on a display screen 310. The graphs plot the horizontal (x) and vertical (y) coordinates of the target and gaze point on the display screen 310. In these graphs, black dots or lines 910 represent the localization of the target 303 on the display screen 310. Blue dots 920 represent the localization of the eye gaze on the display screen 310. Red dots / lines 930 represent regressions of the blue dots 920 indicating the time-averaged localization of the patient's gaze, which can be used, for example, when the eye tracker is not optimally calibrated. Figure 9A The image shows eye-tracking data during the gaze phase of the mission before target 303 begins to move. Figure 9B The image shows eye-tracking data during a slow phase of smooth following when target 303 moves at a constant speed in a circular path. Figure 9C The image shows eye-tracking data during the acceleration phase of smooth following when target 303 is moving at an accelerating speed within a circular path. Figure 9C The data shown is from before the patient was determined to have reached the breakpoint. Figure 9D The image shows eye-tracking data during the acceleration phase of smooth tracking when the target 303 moves at an accelerated speed in a circular path and after the patient has been determined to have reached the breakpoint.

[0246] In some forms, quantitative values ​​associated with the point of interruption can be used to determine the presence of TBI or even other neurological conditions (informally including neuromuscular junction disorders, diseases specifically affecting the extraocular muscles (such as orbital myositis or thyroid ophthalmopathy), and neurodegenerative disorders). For example, in some forms, the measure of interruption can be the time prior to the interruption in a smoothing follow-up task. If this time is below a certain threshold, a positive diagnosis of the presence of TBI can be made. In some forms, a diagnosis of mild TBI (i.e., mTBI or concussion) can be made if the time prior to the interruption is below a first higher threshold but above a second lower threshold. In such forms, a diagnosis of moderate or severe TBI can be made if the time prior to the interruption is also below the second lower threshold. In some forms, these thresholds can be absolute values ​​of the time prior to the interruption determined through experimental observation of previous patients. In other forms, the thresholds can be calculated from control values ​​of the time prior to the interruption that have been experimentally determined to be “normal.” For example, the threshold can be a certain percentage below the control value. This percentage can also be determined through experimental observation of previous patients.

[0247] 7.3.1.9 Combination of Measures

[0248] The preceding text has explained each of the multiple measures that can be used individually to determine the presence of a TBI. In some forms, step 605 of analyzing the data may include determining any combination of two or more measures described above. These two or more measures can be combined to determine an indication of the presence of a TBI.

[0249] For example, in some forms, the two or more measures can be combined into a weighted average. This allows for the application of larger weights to some of the measures. For instance, in one example, the weighted average could be calculated as the sum of any two or more of the measures described above, each weighted by a certain coefficient, where the sum of the coefficients is 1. In one example, the three measures, in order from most weighted to least weighted, could be: a measure of one or more features of SWJ during the fixation task, a measure of the microscan video rate during the fixation task, and a measure of smooth follow-up interruptions when following a moving target.

[0250] 7.3.1.10 Reference measures for individual patients

[0251] The foregoing has described how one or more measures can be determined from eye-tracking data, and these measures can be used to determine the presence of TBI in a patient. In some forms, the determination indicator may include comparing any measure, and / or any combination of two or more measures, to one or more predetermined thresholds. In some forms, the measures can provide an indication of the presence of TBI without requiring any prior testing on the patient under consideration. For example, one or more predetermined thresholds may be determined from measures determined from other patients. The measures may be absolute values ​​(which may indicate the presence or absence (or severity) of the patient's TBI), or the measures may be values ​​determined by comparing them to "normal" values ​​determined from experimental observations of previous patients.

[0252] In other forms, one or more predetermined thresholds can be determined from measures identified from the patient at one or more earlier times. For example, in some forms, an individual patient may have these measures calculated once or multiple times, and these measures are considered reference values ​​to establish a baseline for the measures applicable to that patient. These provide points of comparison for future analyses of that patient. When assessed at a future point in time, the same measures can be identified (using the methods described above) and compared to reference measures for the same patient. If any one or more of the identified measures (or a combination of measures, e.g., a weighted average) differs from an equivalent reference value by more than a certain threshold difference (this difference can be an absolute value or a percentage difference), this can indicate that the patient has TBI. Similarly, different thresholds can be used to determine whether the diagnosis is mild or severe TBI.

[0253] It should be noted that patients with chronic conditions may show elevated baseline levels (e.g., SWJ including deformities and potentially multiphasic) for a considerable period after injury (e.g., 6 months or longer). For example, healthy patients may have a score ranging from 0 to 5 using the following algorithm:

[0254] Indication of the presence of TBI = (typical SWJ) 总和 * W1) + (Deformed SWJ) 总和 * W2) + (multiphase SWJ[2]) 总和 * W3) + … + (polyphase SWJ[n]) 总和 * W n+1 )

[0255] Where W1 = 1, W2 = 2, W3 = 3, W n+ = n+1.

[0256] In contrast, scores of around 15 to 20 are not uncommon for patients still experiencing moderate symptoms.

[0257] In one exemplary form, repeated testing on the patient (i.e., subsequently collecting eye data representing eye movement at later time points during a fixation task) allows for the determination of changes in indicators of the presence of TBI over time. This change can be used as an indicator of neuronal recovery.

[0258] It is envisioned that continuous assessment of patients may be particularly useful for individuals at high risk of brain injury, such as those involved in contact sports.

[0259] 7.3.1.11 Indicator Output

[0260] Refer again Figure 6 An exemplary method according to this technology may include step 606, wherein an indication of the presence of TBI in a patient is output from a data analysis system 400. This indication may be output as a quantitative and / or qualitative assessment of the risk that the patient has TBI. For example, in some forms, the output indication may be a simple binary output of a positive or negative diagnosis of TBI or mTBI. In other forms, the output indication may be an indication of a definite risk of the presence of TBI or mTBI. Such a risk factor may be determined based on the amount by which one or more measures are above / below a corresponding threshold, as determined from earlier experimental observations. The risk factor may be expressed quantitatively (e.g., “the probability of the presence of TBI is 80%”) or qualitatively (e.g., “the patient is very likely to have TBI”). In other forms, the output indication may provide a qualitative description of the indicated severity, such as “TBI not detected,” “mild TBI detected,” “moderate TBI detected,” “severe TBI detected,” etc. The determination of these bands may be based on how the measures are compared to multiple thresholds, as determined from earlier experimental observations.

[0261] 7.3.1.12 Exemplary Application

[0262] The table below presents exemplary data from real-world tests of combined SWJ and deformed SWJ during a 20-second gaze period on a rugby player (and a civilian injured person). Many of these patients underwent the test before and after the injury. The score (i.e., an indication of the presence of TBI) was calculated using the following algorithm:

[0263] Indication of the existence of TBI = (count) 典型SWJ Typical SWJ * W1) + (count) 畸形SWJ (malformed SWJ * W2) + (count) 多相SWJ[2] Multiphase SWJ[2] * W3) + … + (count) 多相SWJ[n ]Multiphase SWJ[n] * W n+1 )

[0264] Where W1 = 1, W2 = 2, W3 = 3, W n+ = n+1.

[0265]

[0266] It should be noted that eye movement dysfunction may temporarily worsen in the first two weeks after injury due to the intrinsic pathophysiology of traumatic brain injury before improving; physical damage to brain tissue can trigger a neurometabolic cascade of 'secondary' damage (e.g., as described in Giza CC, Hovda DA. The new neurometabolic cascade of concussion. Neurosurgery. 2014 Oct;75 Supplement 4(0 4):S24-33. doi: 10.1227 / NEU.0000000000000505. PMID: 25232881; PMCID:PMC4479139).

[0267] It can also be observed that, in addition to the overall score returning to baseline over time, the complexity of potential SWJ events also decreases during the recovery process.

[0268] For the sake of completeness, while the examples above are provided relative to the identified algorithms and weighting schemes, it should be understood that they are not intended to limit the alternative methods of this disclosure described herein.

[0269] 7.4 Other Remarks

[0270] Unless the context clearly requires otherwise, throughout the specification and claims, the terms “comprise”, “comprising”, etc., shall be interpreted as having an inclusive meaning, the opposite of an exclusive or exhaustive meaning, namely, “including but not limited to”.

[0271] All disclosures of all applications, patents, and publications (if any) cited above and below are incorporated herein by reference.

[0272] Any reference to prior art in this specification is not and should not be construed as an admission or recommendation of any kind, as such prior art forms part of the common general knowledge in any country of the world.

[0273] This technology can also be broadly defined as any part, element, or feature that is individually or jointly mentioned or indicated in any or all combinations of two or more of the said parts, elements, or features in the specification of this application.

[0274] The preceding descriptions refer to the whole or parts having their known equivalents, which are incorporated into this text as if described separately.

[0275] It should be noted that various changes and modifications to the currently preferred embodiments described herein will be apparent to those skilled in the art. Such changes and modifications can be made without departing from the spirit and scope of the present invention and without diminishing its incidental advantages. Therefore, such changes and modifications are intended to be included within the scope of the present invention.

Claims

1. A computer-implemented method for assessing the presence of TBI in a patient, the method comprising: Receive eye data, which represents fixational eye movements of the patient's eyes during a fixation task as the patient gazes at a fixed target; Determine one or more measures of the movement of the eye from the eye data, wherein the one or more measures include measures of one or more features of one or more square wave jerks (SWJ); Indications for the presence of the TBI in the patient are determined based on one or more of the metrics; as well as Output the instruction.

2. The computer-implemented method according to claim 1, wherein the one or more features include the atypical features of the one or more SWJs.

3. The computer-implemented method according to any one of claims 1 to 2, wherein at least one of the one or more SWJs comprises two or more phases.

4. The computer-implemented method of claim 3, wherein determining the one or more metrics includes classifying each of the two or more phases.

5. The computer-implemented method of claim 4, wherein the potential classification of each of the phases includes two or more of the following: primary saccade deflection, subsequent saccade deflection, saccade recovery, progressive recovery, saccade peaking, and glide phase.

6. The computer-implemented method according to any one of claims 3 to 5, wherein determining the one or more metrics comprises quantizing one or more features of each of the two or more phases.

7. The computer-implemented method according to any one of claims 1 to 6, wherein the indication determining the presence of the TBI in the patient is based at least in part on the cumulative sum of SWJ instances.

8. The computer-implemented method of claim 7, wherein hierarchical weighting can be applied at least in part based on the complexity of each of the one or more SWJs.

9. The computer-implemented method of claim 8, wherein the weighting is based at least in part on the number of phases of each of the one or more SWJs.

10. The computer-implemented method of claim 9, wherein the weighting is adjusted based on one or more metrics of the associated phase or one or more features.

11. The computer-implemented method according to any one of claims 1 to 10, wherein the method further comprises: Receive target data, which represents the location of the fixed target when eye data representing fixational eye movement of the eye is captured; as well as The target data is used to determine one or more measures of the movement of the eye.

12. The computer-implemented method according to any one of claims 1 to 11, wherein the one or more measures further include a measure of the average error between the localization of the fixed target and the localization of the patient's gaze during the fixation task.

13. The computer-implemented method according to any one of claims 1 to 12, wherein the one or more measures further include a measure of the frequency of microsaccades during the gaze task.

14. The computer-implemented method according to any one of claims 1 to 13, wherein the method further comprises receiving eye data, the eye data representing smooth eye movement of the patient's eye as the patient gazes at a moving target during a smooth following task.

15. The computer-implemented method of claim 14, wherein the one or more metrics further include a metric for smooth follow-up interruptions when following the moving target.

16. The computer-implemented method of claim 15, wherein the metric for the smooth following interruption is the time prior to the interruption occurring in the smooth following task.

17. The computer-implemented method according to any one of claims 1 to 16, wherein the step of determining the indication comprises combining two or more of the metrics.

18. The computer-implemented method according to any one of claims 1 to 17, wherein determining the instruction comprises: The one or more metrics, and / or a combination of two or more of the metrics, are compared with one or more predetermined thresholds.

19. The computer-implemented method of claim 18, wherein the method comprises: The one or more predetermined thresholds are determined from similar metrics identified from other patients.

20. The computer-implemented method of claim 19, wherein the method comprises: The one or more predetermined thresholds are determined from similar metrics identified from the patient at one or more earlier times.

21. The computer-implemented method according to any one of claims 1 to 20, wherein the method further comprises: Control the display screen to show the fixed target to the patient.

22. The computer-implemented method of claim 21, wherein the method comprises: The display screen is controlled to show the fixed target at an off-center location relative to the patient.

23. The computer-implemented method according to any one of claims 1 to 22, wherein the method comprises: The direction of the gaze is determined by detecting the positioning of the iris of the eye.

24. The computer-implemented method of claim 23, wherein determining the direction of the gaze by detecting the positioning of the iris of the eye comprises: An object detection algorithm is applied to the eye data representing the movement of the eye.

25. The computer-implemented method according to any one of claims 1 to 24, wherein the method comprises: The eye data representing the movement of the eye is classified using a transformer model.

26. The computer-implemented method according to any one of claims 1 to 25, the method comprising: Receive additional eye data, which represents fixational eye movements of the patient's eyes during a second fixation task; A second indication for determining the presence of the TBI in the patient based on the additional eye data; as well as The progress of neuronal recovery in the patient is determined by comparing the second indication of the presence of the TBI in the patient with the previous indication of the presence of the TBI in the patient.

27. A system for assessing the presence of TBI in a patient, the system comprising a processor configured to perform a computer-implemented method according to any one of claims 1 to 26.

28. A computer-readable medium having instructions stored thereon for performing a computer-implemented method for assessing the presence of a patient’s TBI according to any one of claims 1 to 26.

Images