Method and apparatus for objectively determining optimal correction of a subject's ocular refraction - Patents.com

Abstract

1. A method for objectively determining optimal photorefractive correction for a subject, comprising: a) providing successive discrete lens powers to the subject's eye until a recorded neural signal corresponding to one of the lens powers exhibits maximum neural activity compared to all other recorded neural signals, and recording corresponding successive neural signals of the subject while the subject's eye is receiving visual stimulation through each lens power (Blocks E2-E5); b) determining that the optimal correction of the subject's ocular refraction corresponds to a lens power for which the subject exhibits a neural signal indicative of a given reduced neural activity compared to the maximum neural activity obtained in step a) (blocks E6 and E7).

Description

[Technical field]

[0001] The present invention relates to the field of ocular refraction in a subject.

[0002] More specifically, the present invention relates to an apparatus and method for determining the optimal correction of a subject's ocular refraction. [Background technology]

[0003] Refraction is the phenomenon whereby a ray of light changes direction when it hits the interface between two different media. Refraction specifically occurs when a ray of light hits the eye of a subject, in which case it is referred to as "ocular refraction". Such ocular refraction will be referred to as "refraction" in the remainder of this specification. Such refraction can be abnormal, causing the image seen by the eye to be formed outside the retina. Being able to measure the refraction of each of a subject's eyes is an important aspect of ophthalmic examinations, especially when vision correction is required. Today, this measurement can be done from two different methods: objective refraction and subjective refraction.

[0004] Objective refraction testing, which does not depend on the subject's response, can be performed using equipment that is currently widely available to prescribers, such as retinoscopes, autorefractometers, and aberrometers. Objective refraction testing is performed quickly and reliably. However, objective refraction testing only provides information about the refractive error of the optical system of the eye at a given moment, and such information is insufficient to determine the visual needs of the subject. Therefore, objective refraction testing only provides an approximation of the optimal correction that the subject needs.

[0005] Subjective refraction tests attempt to assess visual function beyond just the optics of the eye by also taking into account the brain processes involved. The test relies on the subject's response to determine the optimal correction that gives the best possible visual acuity in distance vision with minimal accommodation effort. Subjective refraction tests are therefore more effective at capturing the actual visual needs of a subject than objective refraction tests, but it is recommended to use the results of objective refraction tests (i.e., use an estimate of the optimal correction) as a starting point for subjective refraction tests.

[0006] However, there are several factors that make subjective refraction testing less than an ideal measurement method. Indeed, given the nature of the measurement, which is based only on the subject's response, its results can be quite noisy, since it involves information simultaneously processed by several cortical areas, i.e. areas important for assessing the quality of vision (or visual acuity), but also many associated areas that are not directly involved in visual perception but intervene in different ways depending on the subject (in terms of the type of associated area activated and the level of activation), such as areas related to intention judgment. Such a complex and noisy process often results in the subject himself not being able to confidently determine the correction that gives the clearest image perception. Moreover, subjects are well aware of the essential role played by the visual acuity test and pay particular attention to those factors that may lead to erroneous results or make the visual acuity test a stressful experience.

[0007] All these factors inevitably lead to subjective refraction being a longer and more complicated procedure. Furthermore, there is a large inter-examiner variability in subjective refraction, i.e. the optimal correction determined in subjective refraction is highly dependent on the skill of the eye care professional performing the test.

[0008] Thus, a need exists for improved determination of the optimal correction of a subject's refraction. Summary of the Invention [Means for solving the problem]

[0009] One object of the present invention is to provide a method for objectively determining the optimal correction of a subject's ocular refraction, which leads to identifying the lens power that produces the best perception by the subject, which best perception is as comfortable as possible for the subject in the long term.

[0010] The above object is achieved by providing a method as claimed in claim 1.

[0011] In the present invention, "ocular refraction" is considered to be the refraction of one eye of a subject. The "optimal correction" of a subject's refraction is the correction applied to the subject's eye that allows the subject to have clear vision at distances without or with as little accommodation as possible of the subject's eye. The optimal correction that allows this clear vision at distances without or with as little accommodation as possible of the eye is also the one that is comfortable in the long term.

[0012] The method of the present invention relies on neural activity to identify which lens power will produce the best refraction for the subject's eye and allow the subject to feel comfortable in the long term.

[0013] Surprisingly, the subject's neural activity in brain regions related to the quality of visual perception (also called visual acuity) seems to be greatest when the subject starts to accommodate, not when the subject does not need any accommodation (or as little accommodation as possible) to see the visual stimulus clearly. By "starting to accommodate" we mean that the subject accommodates in a way that is measurable by the phoropter used in the test, for example more than 0.25D, or more than 0.50D, or more than 1D, depending on the accuracy of the phoropter used in the test. In fact, contrary to preconceptions, the neural activity reaches a maximum due to neural artifacts appearing in the subject's brain, which are associated with the initiation of an accommodation response of the eye receiving the visual stimulus. Such accommodation gives the subject the impression of improved visual acuity, but in reality is not necessary for the subject's eyes to see with a satisfactory quality. Moreover, such accommodation may be uncomfortable for the subject in the long term. Accommodations that induce neural artifacts are linked, for example, to miosis (constriction of the pupils of the eye), convergence (simultaneous inward movement of both eyes towards each other), or power changes in the crystalline lens.

[0014] Notably, to determine the optimal correction, the method of the present invention finds the subject's neural signals that exhibit a given reduced neural activity in the brain regions related to vision compared to the subject's maximum neural activity recorded during the test. By doing so, the method of the present invention takes into account the presence of neural artifacts induced by the onset of accommodation when determining the optimal correction of refraction.

[0015] Thanks to the method of the invention, the determination of the optimal correction of the subject's refraction is improved while still remaining objective: the subject does not have to answer any questions to determine the optimal correction, which is determined solely on the basis of the respective nerve signals recorded when at least one eye of the subject receives a visual stimulus through the respective lens power.

[0016] According to an advantageous embodiment of the method of the invention, in step b), the given reduced neural activity is determined by correcting the maximum neural activity based on a predefined coefficient k related to the onset of the eye's accommodation response.

[0017] More precisely, in this advantageous embodiment, the factor k takes into account and removes neural artifacts induced by the onset of accommodation that may appear while the optimal correction is being searched for. Thanks to the factor k, the determination of the optimal correction of the subject's refraction is improved while still remaining objective.

[0018] According to another embodiment, in step b), the given reduced neural activity is determined by machine learning taking into account a set of neural signals recorded for many subjects whose eyes have been provided with a succession of individual lens powers, the optimal correction of which is already precisely known.

[0019] Further advantageous properties of the method according to the invention are given, jointly or separately, in claims 3 to 12.

[0020] It is a further object of the present invention to provide an apparatus for objectively determining the optimal correction of a subject's eye refraction, which leads to identifying the lens correction that produces the subject's best perception, which best perception is as comfortable as possible for the subject in the long term.

[0021] The above object is achieved by providing an apparatus as claimed in claim 13.

[0022] The device of the invention more precisely comprises at least a neural sensor, for example an electrode, capable of detecting neural signals originating from at least one area of ​​the subject's brain linked to visual acuity, i.e. the quality of the visual perception of the subject's eye, and a control unit which records the respective neural signals obtained when the subject's eye receives a visual stimulus through the respective lens powers, finds the recorded neural signals associated with the maximum neural activity, and determines which neural signals will show a given reduced neural activity compared to the maximum neural activity obtained in step a).The control unit then finds which lens power will give such a neural signal of reduced neural activity and concludes that this lens power is the optimal correction of the subject's refraction.

[0023] Advantageously, the device of the present invention allows for the objective determination of the optimal correction of a subject's refraction, independent of the operator of the device and without questioning the subject.

[0024] Further advantageous features of the device according to the invention are given, together or separately, in claims 13 and 14.

[0025] Advantageously, the device comprises at least one active power lens, the power of which is driven by a control unit to provide a distinct lens power at which the subject's eye receives visual stimulation depending on an analysis of the neural activity of previously recorded neural signals.

[0026] According to one embodiment, at least one active power lens is incorporated into a pair of glasses or contact lenses.

[0027] Detailed Description of the Preferred Embodiments The following description, referring to the accompanying drawings, will make clear what constitutes the present invention and how it can be achieved. The present invention is not limited to the embodiments shown in the drawings. Thus, when features mentioned in the claims are followed by reference signs, it will be understood that such signs are included only for the purpose of improving the comprehension of the claims and do not limit the scope of the claims. [Brief description of the drawings]

[0028] [Figure 1] 1 is a flow chart summarizing the main steps of the method of the present invention; [Diagram 2] 1 is a schematic diagram of an apparatus of the present invention; [Diagram 3] FIG. 1 is a schematic diagram of a first embodiment of a neural sensor of a device according to the invention. [Figure 4] FIG. 2 is a schematic diagram of an additional electrode included in the neural sensor of the device of the present invention. [Diagram 5] FIG. 1 is a graph showing eight spectral signals S1-S8 derived from eight neural signals recorded for a given subject for eight distinct lens powers ranging from +1 diopter to -0.75 diopter, where the x-axis gives the frequency in Hertz (Hz) and the y-axis gives the exponent of the Fourier transform called "amplitude" in (mV)2.Hz-1. [Figure 6] 6 is a graph showing the amplitude of the main peak of each of the spectral signals of FIG. 5 as a function of lens power (in diopters); [Figure 7] 13 is a graph showing the amplitude of the main peak of each spectral signal obtained from various nerve signals recorded on another test subject for lens powers ranging from +1D to -0.75D. [Figure 8A] 1A-1C show examples of eyeglasses including active power lenses according to the present disclosure. [Figure 8B] 1A-1C show examples of eyeglasses including active power lenses according to the present disclosure. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0029] The present invention provides an apparatus 10 and method for objectively determining the optimal correction of a subject's ocular refraction.

[0030] As explained in the introduction, ocular refraction is the refraction that occurs when a ray of light strikes one of the eyes of a subject. Such ocular refraction will be referred to as "refraction" in the remainder of this document. If the image caused by such refraction is formed outside the retina, the refraction is considered to be defective. In the case of defective refraction, it is necessary to know how to correct the refraction in order to improve the subject's vision. The optimal correction of refraction is considered to be the correction that allows the subject to have clear vision at far distances (also called infinite distances) without accommodation of the corresponding eye or with as little accommodation as possible and in as comfortable a state as possible. More precisely, the device 10 and method of the present invention determine the optimal correction of the (defective) refraction of a subject (called test subject) by: - an analysis of each of the test subjects' neural signals recorded while at least one of the test subjects' eyes 1 receives a given visual stimulus through the respective lens power; and - knowledge of the presence of neural artifacts that may be induced by the initiation of an accommodative response in the eye receiving a visual stimulus, The judgment is made objectively based on the above.

[0031] All features described in relation to the apparatus of the invention, in particular the features described in relation to the processor of the apparatus, also apply to the method of the invention, and vice versa.

[0032] As shown in FIG. 2, the device 10 of the present invention includes: at least one neural sensor 11 for detecting neural signals originating from at least one region of the subject's brain, the neural signals being linked to the subject's visual acuity; a control unit 15, a) recording successive nerve signals of the test subject while eye 1 of the test subject receives visual stimulation through successive distinct lens powers until one of the recorded nerve signals exhibits a maximum neural activity compared to all other recorded nerve signals; b) determining that the optimal correction of the subject's eye refraction corresponds to a lens power at which the subject exhibits a neural signal indicative of a given reduced neural activity compared to the maximum neural activity obtained in step a); and a control unit 15 adapted to

[0033] Thus, the device 10 can record and analyze neural signals while presenting visual stimuli to the test subject through successive lenses L1, L2, L3, L4, L5 of different powers that make the test subject's vision more or less blurred. Based on this analysis, and knowing that part of the neural activity can be induced by accommodation, the device 10 then finds out what lens power will allow the subject to see the visual stimuli most clearly at a distance, without accommodation or with as little accommodation as possible, and with as much comfort as possible in the long term. Note that in this specification, the term "blurred" is opposed to the term "clear".

[0034] Typically, visual stimuli are displayed on a screen 12 and the test subject is asked to view with one eye (monocular testing) or both eyes 1 (binocular testing). Preferably, the test is monocular, with the first eye 1 being covered while the other eye is being tested. If binocular testing is selected, both of the subject's eyes do not necessarily view the visual stimuli through the same lens power.

[0035] Classically, the screen 12 is placed at a distance of about 6 meters (m) from the test subject to simulate vision at far distances. The screen 12 may be included in the device 10 of the present invention, in which case the visual stimuli shown thereon may be controlled by the control unit 15. Alternatively, the screen may be a separate element excluded from the device 10.

[0036] The visual stimuli that the test subject is asked to view can be any type of image. For example, the visual stimuli are blinking Gabor patches. Such specific visual stimuli make it possible to improve the signal-to-noise ratio of the neural signals recorded by the neural sensors 11 of the device 10, thus facilitating further analysis of each recorded neural signal.

[0037] A refractometer 13 (also called a "phoropter" or "refractor") is used to facilitate changing the lens power at which the test subject views the visual stimuli.

[0038] More precisely, the refractometer 13 can place lenses L1, L2, L3, L4, L5 of selected power in front of the examined eye 1 of the test subject. The power of each lens L1, L2, L3, L4, L5 blurs the vision of the eye 1 of the test subject to a greater or lesser extent. The powers of the lenses L1, L2, L3, L4, L5 are generally given in diopters. Herein, the power of each test lens is a relative power compared to the power of a reference lens LR. The power of the reference lens LR is preferably a rough estimate of the optimum correction of the test subject. Such a rough estimate is obtained, for example, by an objective refraction test as described in the introduction. Alternatively, the power of the reference lens can be 0 diopters (D). Alternatively, the reference power can be a previously known correction value of the subject.

[0039] Thus, the power of the lens tested in front of the test subject's eye 1 indicates a positive or negative diopter compared to the power of the reference lens: the more positive the power, the more blurred the vision of the corresponding eye 1 of the test subject.

[0040] Preferably, the refractometer 13 is included in the device 10 of the present invention. More preferably, the refractometer 13 is controlled by a control unit 15.

[0041] Furthermore, advantageously, in the device 10, the refractometer 13 is an automatic refractometer 13 controlled by a control unit 15 of the device 10. The control unit 15 controls the refractometer 10 to automatically change the power of the lenses L1, L2, L3, L4, L5 to which the eye 1 of the test subject receives a visual stimulus. Such an automatic change of the lenses L1, L2, L3, L4, L5 placed in front of the eye 1 of the test subject is preferably based on an analysis of the neural activity of neural signals previously recorded by the neural sensor 11. The change of power can be performed by an operator.

[0042] Most preferably, the change in lens power is performed by the brain-computer interface based on the analysis of previously recorded neural signals for previously tested lens powers. The test subject is therefore not asked any questions and is not subject to any operator control of the change in lens power by the refractive index 13. Furthermore, the change in lens power occurs at the optimal moment for the subject. In this case, to make the test efficient, the next lens power is selected based on previously directly recorded and analyzed neural signals.

[0043] In particular, if the neural activity of the neural signal recorded for the frequency tested at time t is higher than the neural activity of the neural signal recorded for the frequency tested at time t-1, then the next frequency tested at time t+1 should be lower than the frequency tested at time t. Conversely, if the neural activity of the neural signal recorded for the frequency tested at time t is lower than the neural activity of the neural signal recorded for the frequency tested at time t-1, then the next frequency tested at time t+1 should be higher than the frequency tested at time t.

[0044] The neural sensor 11 of the device 10 is intended to detect neural signals occurring in the brain of the test subject while the test subject tries to see as clearly as possible the visual stimuli presented to him through the lenses L1, L2, L3, L4, L5.

[0045] In this embodiment, as shown in Fig. 2, the neural sensor 11 comprises at least three electrodes 110, preferably five electrodes 110, capable of detecting the electrical activity of a group of specific nerves that are stimulated when at least one eye of the test subject tries to clearly see a visual stimulus received by that eye. These specific groups of nerves are preferably located at the back of the subject's brain. The neural signals detected by these electrodes 110 are linked to the visual acuity of the test subject's eye, i.e., to the quality of the subject's visual perception.

[0046] The electrodes 110 are designed to be placed on the back of the test subject's head in order to record electroencephalograms (or neural signals) originating from the back of the test subject's brain. In this embodiment, the electrodes 110 are advantageously integrated onto a headset of an armchair 2 (see FIG. 3 ) in which the test subject is supposed to sit and proceed to the refraction test.

[0047] As shown in Figures 2 and 4, the neural sensor 11 may additionally comprise at least three additional electrodes 111 capable of detecting the electrical activity of specific groups of nerves that are stimulated when the test subject is trying to accommodate his / her eye 1 to clearly see the visual stimulus and that may also be involved in image quality processing, judgment and analysis. These specific groups of nerves are located, for example, in the prefrontal cortex of the subject's brain.

[0048] The additional electrode 111 is a frontal electrode 111 designed to be placed on the test subject's forehead to record electroencephalograms (or neural signals) originating from the prefrontal cortex of the test subject's brain.

[0049] As shown in FIG. 4, in this embodiment an additional electrode 111 is integrated, for example, on the chin rest or on the front support 3 of the refractometer 13 .

[0050] In this embodiment, electrode 110 and a final additional electrode 111 continuously detect neural activity from the scalp and transmit the neural activity to an amplifier (not shown), which increases the signal to facilitate its analysis.

[0051] Recording of each nerve signal by the nerve sensor 11 is preferably performed immediately before, during and following reception of a visual stimulus by the test subject's eye 1 in order to detect activation of specific nerves and their relaxation.

[0052] The control unit 15 of the device 10 of the present invention is arranged to communicate with the neuro-sensor 11 and, where appropriate, the refractometer 13 and / or the screen 12. This communication may be established using either wireless communication means or line-based communication means.

[0053] The control unit 15 may be located remotely from the test subject. In an alternative embodiment, the control unit can be worn by the test subject. In the example shown in the figures, the control unit 15 is located remotely from the subject. In the example shown in Figures 8A and 8B, the control unit 15 is worn by the subject. In this last example, the control unit 15 is integrated into glasses 50.

[0054] 2, the control unit 15 here comprises, like such a neuro-sensor 11, a memory 150 and a processor 151 in communication with each other, the memory 150 and the processor 151 also in communication with the refractometer 13 and / or the screen 12, as appropriate. The control unit 15 is, for example, integrated in a computer.

[0055] In particular, the memory 150 may be configured to receive each neural signal detected by the neural sensor 11 and record (or store) each neural signal so that the processor 151 can analyze each recorded neural signal.

[0056] The memory 150 may also be configured to store, corresponding to each nerve signal, which lens power elicited that nerve signal. The memory 150 may also store baseline lens powers for the test subject.

[0057] The analysis of each nerve signal recorded in the memory 150 is carried out by the processor 151 of the control unit 15. Such analysis mainly includes: - processing each neural signal to infer the corresponding neural activity (step a), - an estimate of the maximum neural activity detected among all recorded neural signals (step a), and - Determination of a given reduced neural activity, which is compared with a maximum neural activity, in order to find the lens power that results in such reduced neural activity and thus corresponds to the optimal correction of the refraction of the test subject (step b), Includes.

[0058] In the present invention, two elements are involved: a determination of the maximum neural activity and a determination of the reduced neural activity given; is particularly important.

[0059] In practice, the reduced neural activity corresponds to a maximum neural activity from which accommodation-induced neural artifacts can be inferred or eliminated. The ocular accommodation response that induces neural artifacts is linked, for example, to pupil constriction, convergence, or power changes of the crystalline lens.

[0060] The processing of each neural signal aims to infer the neural activity of the test subject related to the lens power being tested that leads to that neural signal. It is noted that areas of the subject's brain related to visual acuity all have increased neural activity when the subject perceives the visual stimulus better (clearer).

[0061] In this embodiment, the test subject's neural activity associated with one lens power is determined by extracting at least one feature of the neural signal recorded for that lens power.

[0062] Here, the feature of the neural signal that gives an index of neural activity is the amplitude in the spectral signal derived from the recorded neural signal. Higher neural activity corresponds to higher amplitude in the spectral signal. More specifically, neural activity is inferred from the value of the amplitude of the main peak shown in the spectral signal.

[0063] Thus, the analysis of the nerve signal carried out by the processor 151 of the control unit 15 mainly comprises the washing out of the nerve signal from noise and its transformation by a Fourier transform to obtain a spectral signal. In this case, the observed feature is the exponent of the Fourier transform given in millivolts per Hertz squared (mV2.Hz-1), as shown in Fig. 5. The exponent of the Fourier transform is hereafter called "amplitude" for simplicity. From the spectral signal the amplitude of the main peak can be inferred, and such amplitude is therefore considered as the nerve activity of the test subject. In practice, methods of washing out the nerve signal from noise and transforming the nerve signal into a spectral signal are well known.

[0064] In a further aspect of the invention, the device 10 comprises at least one active power lens 51, the power of which is driven by the control unit 15. The lens power of the at least one active power lens 51 is arranged and / or configured to be modified when the subject is exposed to a visual stimulus during the determining step. Additionally, the lens power of the at least one active power lens 51 is arranged and / or configured to be equal to the optimal correction once the control unit 15 determines this value.

[0065] In the example of Figures 8A and 8B, at least one active power lens 51 is incorporated into a pair of glasses 50 intended to be worn by a subject.

[0066] In this example, such glasses 50 comprise two active power lenses 51 each including a lens power arranged and / or configured to be modified by the control unit 15 .

[0067] The glasses 50 further comprise a frame 52 holding two active power lenses 51 .

[0068] The active power lenses 51 are each incorporated into the glass of the eyeglasses 50 .

[0069] According to this embodiment, the neural activity sensor 11 may be included in the eyewear temple of the frame 52 or may be in direct contact with the subject's skull. To this end, the device 10 may comprise an electrode 110.

[0070] 8A-8B, the electrodes 110 of the device 10 are deployed from the eyeglasses 50 to extend to a region of the head that will provide access to neural signals originating from at least one region of the subject's brain. For example, the electrodes 110 of the device 10 contained within the eyeglasses 50 are positioned and / or configured to record neural signals originating from an occipital region of the subject's brain. The device 10 is, for example, incorporated within the frame 52 of the eyeglasses 50 or near a portion of the frame 52 that terminates behind the subject's ear.

[0071] According to this embodiment, the device 10 does not have to be combined with a refractometer 13. The refractometer 13 is directly replaced by a pair of spectacles 50, the lens power of each active power lens 51 of which can be changed via a control unit 15 in a similar manner to the refractometer 13.

[0072] The control unit 15 of the device 10 is configured to be in communication with the screen 12. For example, the screen 12 displaying the visual stimuli can be a computer screen or a smartphone screen located at a distance of 2 meters or 4 meters from the glasses 50.

[0073] The lens power of each power active lens 51 of the spectacles 50 can be modified. For this purpose, the control unit 15, the memory 150 and the processor of the device 10 integrated in the spectacles 50 work as in the example of FIG. 2 using the refractometer 13.

[0074] The control unit 15 is configured to modify the lens power of each lens 51 according to the method of the present disclosure. In practice, the control unit 15 of the device 10 controls the lens power of each active power lens 51 through which the test subject's eye 1 receives visual stimulation. Such modification of the power lenses is preferably based on an analysis of the neural activity of the signals recorded by the neural sensor 11.

[0075] Analysis of each nerve signal is performed by a processor in a control unit 15 integrated into the glasses 50. Once the optimum correction for the user has been determined, the glasses 50 store the result in memory 150 and use it to correct the lens power of each active power lens 51 accordingly.

[0076] Therefore, there is no need for an operator to change the lens power of the active power lens 51. Also, the refraction measurement can be performed directly by the subject without any assistance from an operator, so this device is more autonomous than the device 10 disclosed in FIG. 2, which requires the assistance of an operator.

[0077] 8B, the glasses 50 comprise only one control unit 15 configured and / or arranged to control the change in lens power and record neural signals for each active power lens 51. In another embodiment, the glasses 50 comprise a control unit 12 associated with each active power lens 51.

[0078] The method performed by the device 10 can be programmed to run on a particular day, for example every month, or every year, and / or can be based on a change in the subject's correction, for example in the case of an evolution of myopia or in the case of presbyopia of the subject. Once the optimal correction is determined by the control unit 15, the lens power of each lens 51 is changed to be equal to the previously determined optimal correction.

[0079] The eyeglasses 50 include different types of active power lenses 51, e.g. Alvarez lenses as known in the art; fluid lenses, such as those disclosed in EP 2 149 537; Liquid crystal based active lenses, such as those disclosed in EP 3115436 using a blue phase state, It may comprise:

[0080] Each power active lens 51 may be an active lens on eyewear.

[0081] In another embodiment, the active power lens 51 is integrated into a contact lens, as disclosed, for example, in the document US2020064658. Thus, in this last embodiment, the eyeglasses 50 do not comprise a frame 52. The active power lens 51 is held directly by the contact lens on the subject's eye. The neural activity sensor 11 may be included in the contact lens. In addition, the sensor may include an electrode 110 that may be wirelessly connected to the active power lens 51. The control unit 15 is preferably included directly in each contact lens.

[0082] In another embodiment, the active power lens 51 can be an active intraocular lens optic (IOL).

[0083] Because each active power lens 51 is associated with a subject's eye, the lens power of each active power lens 51 can likewise be varied independently.

[0084] On Fig. 5, eight spectral signals are shown that derive from the neural signals recorded by the electrodes 110 when the test subject views the flickering Gabor patch through respective lenses of various powers (a total of eight different lenses). Spectral signal S1 is obtained for a lens power of +1 diopter, spectral signal S2 is obtained for a lens power of +0.75 diopter, spectral signal S3 is obtained for a lens power of +0.5 diopter, spectral signal S4 is obtained for a lens power of +0.25 diopter, spectral signal S5 is obtained for a lens power of 0 diopter, spectral signal S6 is obtained for a lens power of -0.25 diopter, spectral signal S7 is obtained for a lens power of -0.5 diopter, and spectral signal S8 is obtained for a lens power of -0.75 diopter. In this example, the lens power of 0 diopter indicates that the test subject views the visual stimulus through a lens of a power corresponding to the optimal correction of his or her refraction obtained by subjective refraction testing. All other powers given are relative powers compared to the lens power of optimum correction.

[0085] In FIG. 5, the shape of each spectral signal is similar, i.e. each spectral signal contains a main peak centered at a frequency comprised within the range [5 Hz; 25 Hz], preferably [14 Hz; 15 Hz], which frequency ultimately differs for each spectral signal.

[0086] The control unit 15 can compare all the spectral signals and find out which spectral signal among S1 to S8 shows a peak with the maximum amplitude A.

[0087] Here, the lens power that gives the smallest amplitude of the peak of the spectral signal is the lens at which the test subject perceives the visual stimulus most blurry.

[0088] As shown in Fig. 5, the smaller the lens power, the better the subject perceives the visual stimulus, and therefore the greater the amplitude of the corresponding peak of the spectral signal, up to a certain lens power that provides the highest (or maximum) amplitude of the peak of the spectral signal. Such a maximum amplitude corresponds to the clearest perception of the visual stimulus to the brain of the test subject. However, such a clearest perception is not comfortable for the subject in the long term when it comes to accommodation.

[0089] As shown in Fig. 5, when the lens power is reduced below the lens power that provides the maximum amplitude, the amplitude of the peak of the corresponding spectral signal decreases because the test subject's perception of the visual stimulus becomes less and less blurred again, such a phenomenon being due, for example, to the fact that the test subject cannot accommodate sufficiently at this low lens power, or due to other ophthalmic constraints.

[0090] In the example of Fig. 5, the spectral signal showing the main peak with the smallest amplitude is the spectral signal S1 obtained for a lens power of +1 diopter, and the spectral signal showing the main peak with the largest amplitude is the spectral signal S7 obtained for a lens power of -0.5 diopter. The spectral signals S2 to S6 obtained by gradually decreasing the lens power between +0.75D and -0.25D show a gradually increasing amplitude of the respective main peaks, while the spectral signal S8 obtained for a lens power of -0.75D shows a main peak with an amplitude smaller than the maximum amplitude.

[0091] As shown in FIG. 5, the amplitude of the spectral signal S5 obtained from the nerve signal recorded for optimal correction (0D) is not the maximum amplitude obtained for the spectral signal S7.

[0092] Figure 6 was obtained after analysis of the spectral signals of figure 5. It gives the value of the amplitude A of the main peak of each of the spectral signals of figure 5 as a function of the power (in diopters or D) of the lens which gives rise to such a spectral signal.

[0093] FIG. 7 was obtained in a similar manner to FIG. 6, but for another test subject for whom an optimal correction is sought. In other words, FIG. 7 was obtained after recording the neural signals of such other test subject while he received visual stimuli (e.g. Gabor patches) on his eyes through various lens powers ranging from -0.75D to +1D. In FIG. 7, a lens power of 0D indicates that the test subject is viewing the visual stimuli through his reference lens, which corresponds to a rough estimate of his optimal correction, obtained for example from an objective refraction test as described in the introduction. All other powers given are relative powers compared to the reference lens power.

[0094] 6 and 7 are each useful for graphically determining which lens power leads to the neural signal with the greatest neural activity for each test subject, where the greatest neural activity is taken as the maximum amplitude in all spectral signals.

[0095] In the present invention, "maximum" is defined as being higher than the two surrounding points. In other words, the neural signal recorded for a given lens power shows maximum neural activity only if it shows more neural activity than the neural signals recorded for both smaller and larger lens powers surrounding the given lens power.

[0096] Of course, the control unit 15 does not have to determine the maximum graphically, but can only determine it computationally. Figures 6 and 7 visually aid the reader in understanding the evolution of neural activity with changes in lens power and the determination of such maximum neural activity.

[0097] Once the maximum neural activity is found, the processor 151 of the control unit 15 determines a reduced neural activity that should correspond to the perception of the visual stimulus with as little accommodation as possible or no accommodation at all, in other words, the reduced neural activity that should be obtained without an accommodation response of the eye, in other words again, that corresponds to the maximum neural activity with the neural activity induced by neural artifacts associated with that accommodation response of the eye eliminated.

[0098] In a first advantageous embodiment of the device, the determination of the reduced neural activity is based on a correction of the maximum neural activity based on a predetermined coefficient k related to the accommodation response of the eye, where the coefficient k is associated with the accommodation inducing neural artifact of the test subject. In other words, the coefficient k corrects for the additional neural activity induced by the neural artifact linked to the accommodation.

[0099] The coefficient k is determined before carrying out the method of the present invention, for example, the coefficient k is stored in the memory 150 of the control unit 15.

[0100] The coefficient k is preferably a constant value less than or equal to 1. Preferably, the value of the coefficient k is strictly less than 1.

[0101] The value of the coefficient k depends on the feature of the neural signal on which the determination of the neural activity is based. In particular, when the feature of the neural signal indicative of the neural activity is the amplitude A of the peak of the spectral signal, the coefficient k is a constant value selected between 0.2 and 0.95, preferably between 0.6 and 0.95, more preferably between 0.7 and 0.9. The coefficient k may in particular be equal to 0.2, 0.25, 0.3, 0.35, 0.4, 0.45, 0.5, 0.55, 0.6, 0.65, 0.7, 0.75, 0.8, 0.85, 0.9 or 0.95.

[0102] In the first alternative, k is fixed across all individuals.

[0103] In a second alternative, the coefficient k may vary for different segments of the population. The segments of the population may for example be based on age, with the coefficient k for subjects who are children being different from the coefficient k for subjects who are young adults, which is also different from the coefficient k for subjects who are elderly. The segments of the population may also be complementary or alternatively based on different ocular defects of the subjects, for example refractive errors.

[0104] Since maximum neural activity occurs when accommodation is engaged, the value of the coefficient k should be closer to 1 as accommodation approaches zero. In other words, the coefficient k should be larger when a subject has little or no accommodation capacity remaining than when the subject exhibits a large capacity for accommodation.

[0105] In practice, to determine the value of the coefficient k that will compensate for the subject's neural artifacts associated with the onset of the ocular accommodation response, the following procedure can be performed. c1) a group of subjects for which the optimal correction is known is selected; c2) for each subject, - recording successive neural signals from the subject while at least one eye of the subject receives visual stimuli through a succession of distinct lens powers, including an optimally corrected lens power; - the neural activity of each recorded neural signal is analyzed and the recorded neural signal showing the greatest neural activity is inferred therefrom; the maximum neural activity is compared with the neural activity of the neural signal recorded for the optimally corrected lens power; c3) The coefficient k is estimated from a comparison step of neural activity carried out for each subject.

[0106] More precisely, in step c1), the group may comprise at least 10 subjects, preferably 20 or more. Of course, the group of subjects preferably does not include the test subject.

[0107] In step c1), the actual exact optimum correction for each eye of each subject of the group is already known, for example from the subjective refraction test described in the introduction.

[0108] In step c1), where the coefficient k is fixed across all individuals, the subjects of the group are preferably selected randomly.

[0109] Conversely, in step c1), when the coefficient k varies from one segment of the population to another, the subjects of the group are preferably selected based on the characteristics that the subjects have in common. In other words, the subjects of the group are selected to represent the segment of the population in question. For example, the subjects may all have the same age in a given range, i.e., they may all have ages between 5 and 15 years old, or between 15 and 25 years old, or between 25 and 40 years old, etc. In an alternative or supplementary example, the subjects are selected to be part of the same group if they have the same reference lens power, or the same optimal correction. In an alternative or supplementary example, the subjects are selected to be part of the same group if they all have the same ocular defect.

[0110] In step c2), each subject of the group is individually tested to record neural signals generated when visual stimuli are received through lenses of different powers, including the power of optimal correction for that subject.

[0111] The neural signal exhibiting the greatest neural activity is then determined.

[0112] In step c3), the maximum neural activity is then compared with the neural activity obtained for the optimally corrected lens power and a coefficient k is inferred therefrom.

[0113] More precisely, in a first alternative of steps c2) and c3), the determination of the maximum neural activity is performed in a manner similar to that described above with reference to figures 5 and 6. In other words, the analysis of the neural activity comprises a spectral analysis, a higher neural activity corresponding to a higher amplitude of the main peak of the spectral neural signal. In this alternative, the comparison between the maximum neural activity and the neural activity obtained for the optimal correction comprises a comparison of the amplitudes of the corresponding spectral signals.

[0114] An individualization factor k i is determined from this comparison to compensate for the accommodative response of a given subject.

[0115] More precisely, the individualization coefficient ki can be obtained from the following formula:

number

[0116] The coefficient k is deduced from a comparison of all the individualization coefficients k determined for each subject, in other words, the coefficient k is based on such a comparison of the amplitudes of the spectral signals carried out for each subject.

[0117] For example, the coefficient k can be the maximum individualization coefficient k obtained for a group of subjects. Alternatively, the coefficient k can be the average value of all the individualization coefficients k.

[0118] In a second alternative of steps c2) and c3), the operations of analysis, comparison and inference are carried out by machine learning. More precisely, a machine learning algorithm makes it possible to associate to each test lens power a corresponding neural activity, to determine the relationship between the maximum neural activity and the neural activity obtained with the optimal correction, and to infer the coefficient k from this correlation. Such an alternative is preferred if a large variability is found in the individualization coefficients ki obtained when implementing the first alternative. If this is the case, the recorded neural signals obtained for each subject and each lens power are provided to a machine learning algorithm that infers which features of the neural signals should be analyzed in order to find the maximum neural activity, and that determines the coefficient k that makes it possible to compare the neural signals obtained for the lens power corresponding to the optimal correction with the neural signals exhibiting the maximum neural activity. Such a machine learning algorithm may, for example, comprise an artificial neural network trained with a lot of data (neural signals with optimal correction and corresponding lens powers) and may be used to train the artificial neural network: 1. The characteristics of neural activity that should serve as the basis for determining and / or classifying the neural activity; and 2. Thanks to it, we can compare the maximum neural activity with the neural activity obtained with optimal correction, coefficient k, Help them find it.

[0119] Having now described how the coefficient k may be determined, it will be described how this coefficient k can be used to determine reduced neural activity and then infer the optimal correction of a subject's refraction.

[0120] Such a determination is graphically represented on FIG.

[0121] In this embodiment, where the neural activity is obtained through the amplitude of the main peak of the spectral signal, the processor 151 applies a coefficient k to the maximum amplitude Amax found from the analysis of all the spectral signals, thus finding a reduced amplitude A(reduced) of: A(reduced)=k*Amax

[0122] The optimal correction of the test subject's refraction corresponds to the lens power at which the subject records a neural signal of reduced neural activity.

[0123] It should be noted that the reduced neural activity can be the neural activity of one of the recorded neural signals or can be the neural activity of the extrapolated neural signal. In this embodiment, the reduced neural activity corresponds to the neural activity of one of the neural signals, as shown in FIG. 7.

[0124] Once processor 151 has determined the test subject's neural signal associated with maximum neural activity and calculated the reduced neural activity through coefficient k, processor 151 then finds the corresponding lens power that results in the neural signal exhibiting such reduced neural activity.

[0125] More precisely, in this embodiment, where the neural activity is obtained through the main amplitude of the spectral signal, the processor 151 infers which lens power should give the reduced amplitude of the spectral signal A(reduced). The lens power corresponding to the calculated reduced amplitude can be easily read, for example, on FIG. 7, whether the reduced neural activity is of the extrapolated or of the recorded neural signal.

[0126] The resulting lens power is the test subject's best correction.

[0127] As shown in Figure 7, the optimum correction for the other test subjects here corresponds to a lens that is +0.5D (compared to the reference lens power, here indicated as 0D). Note that the optimum correction is necessarily a lens with a power greater than the power of the lens that produces the maximum amplitude. On Figure 7, it is seen that the optimum correction is neither the power of the reference lens that corresponds to the approximation of the optimum correction, nor the power of the lens that produces the maximum amplitude in the spectral signal.

[0128] FIG. 1 shows the main steps of the method of the present invention for objectively determining the optimal correction of a test subject's refraction.

[0129] More precisely, the method is a) providing successive discrete lens powers to the test subject's eye 1 until a recorded neural signal corresponding to one of the lens powers exhibits maximum neural activity compared to all other recorded neural signals, and recording corresponding successive neural signals of the test subject while the test subject's eye 1 is receiving visual stimulation through each lens power (Blocks E2-E5); b) determining that the optimal correction of the subject's eye refraction is the lens power at which the subject exhibits a neural signal that indicates a given reduced neural activity compared to the maximum neural activity of step a) (blocks E6 and E7); Includes.

[0130] The method may be performed by an apparatus 10 of the present invention, together with a refractometer 13 and a screen 12 for displaying the visual stimulus, as described herein above.

[0131] More precisely, the processor 151 of the control unit 15 may be configured to carry out the calculation steps of the method of the invention described below.

[0132] For example, step a) is performed by the refractometer 13, which is ultimately controlled by the control unit 15, and the nerve signals are recorded in the memory 150 and analyzed by the processor 151. In step b) the nerve signals of reduced neural activity are determined by the processor 151 of the control unit 15.

[0133] Step a) Step a) is always carried out before step b).

[0134] In step a), the subject is asked to view visual stimuli through lenses of different powers, preferably in one eye only. The subject should test lenses of at least three different powers (block E2 in FIG. 1).

[0135] To do so, first, second and third lenses L1, L2, L3 of respective first, second and third powers are successively placed in front of the subject's eyes. While the test subject views the visual stimuli, respective neural signals are successively detected, for example by the neural sensor 11. Each detected neural signal is recorded, for example, in the memory 150.

[0136] As shown in block E3 of Fig. 1, the neural activity of each neural signal is then analyzed to find out whether it has reached a maximum value, taking into account that a point can only be considered a maximum if it is surrounded by two smaller values.For example, if three lens powers are used, with the power of the first lens L1 being smaller than the power of the second lens L2, which is itself smaller than the power of the third lens L3, then the neural activity has reached a maximum only if the neural activity obtained with the second lens L2 is greater than the neural activity obtained with the first lens L1 and greater than the neural activity obtained with the third lens L3.

[0137] As explained above, the subject's neural activity associated with a lens power is determined by extracting at least one feature of the neural signal recorded for that lens power.

[0138] Here, the feature of the neural signal that gives an index of neural activity is the amplitude in the spectral signal derived from the recorded neural signal. Higher neural activity corresponds to higher amplitude in the spectral signal. More specifically, neural activity is inferred from the value of the amplitude of the main peak shown in the spectral signal.

[0139] If the maximum value has been reached, the method proceeds to step b) (blocks E6 and E7 in FIG. 1).

[0140] If the maximum value has not yet been reached, a further lens is tested with a different power than the previously tested one (block E4 in FIG. 1).

[0141] Thus, a further lens is placed in front of the subject's eye and the corresponding neural signals are detected by the neural sensor 11 and recorded for further analysis.

[0142] Again, it is analyzed whether the neural activity has reached a maximum (block E5 in FIG. 1 ). If so, the method proceeds to step b) (blocks E6 and E7). If not, the method repeats blocks E4 and E5 to test further lenses exhibiting new powers. The repetition of blocks E4 and E5 is repeated until the maximum neural activity is reached.

[0143] Preferably, when providing the lens to the eye 1 of the test subject (blocks E2 and E4), the diopter step between two successive test lens powers is comprised within 0.1 to 0.5 D, for example 0.25 D. In the present invention, the expression "successive" is used to compare the powers of the lenses and not necessarily the moment when such lens powers are provided to the test subject, whereas the expression "consecutive" is used to compare the moment when such lens powers are provided to the test subject.

[0144] Preferably, each successive lens power provided to the eye 1 of the test subject is selected based on a reference lens power, which is a rough estimate of the test subject's optimal correction, as previously explained. Such a reference lens power can be obtained by an objective refraction test, as explained in the introduction. The reference lens power can be determined, for example, in a step prior to carrying out the method (block E1 in FIG. 1). The reference lens power can also be the test subject's previous correction value.

[0145] More precisely, the step in diopters (D) between two successive lens powers is preferably smaller when the lens power is closer to the reference lens power than when the lens power is further away from the reference lens power, for example the step between two successive lens powers is 0.5D when the power is more than ±1D compared to the reference lens power, and 0.25 or even 0.1D when the power is less than ±1D compared to the reference lens power.

[0146] In fact, neural signals obtained for lenses exhibiting powers close to the baseline lens power are more likely to exhibit maximal neural activity, and therefore testing lens powers that are close to each other improves accuracy in determining maximal neural activity.

[0147] According to an advantageous feature of the method of the present invention, the powers of each successive lens placed in front of the subject's eye in blocks E2 and E4 are not selected randomly, but in a specific order.

[0148] More specifically, the lens powers provided to eye 1 of the test subject are successively smaller and smaller, with the first lens power provided to the eye being one that blurs the subject's vision. In other words, the first lens power is preferably selected to be a diopter greater than the subject's baseline lens power, for example +2 diopters or +1 diopters compared to the baseline lens power.

[0149] To analyze whether a maximum neural activity has been reached by one of the neural signals and to know that successive lenses placed in front of the subject's eye 1 exhibit smaller and smaller powers, it is sufficient to compare the neural activity of the last recorded neural signal with the neural activity of the directly previously recorded neural signal. Thus, if the directly previously recorded neural signal exhibits more neural activity than the last recorded neural signal, it is considered that a maximum neural activity has been reached, and the directly previously recorded neural signal is the one that exhibits the maximum neural activity.

[0150] According to an advantageous feature of the method of the invention, the power of the next lens to be placed in front of the subject's eye is determined by the brain-computer interface, and the lens power is automatically changed based on the analysis of the neural activity of previously recorded neural signals. By "automatically" it is meant that no one operates the lens change apart from the machine, such operation being based on the analysis of previously recorded neural signals. Advantageously, the analysis and comparison of two consecutively recorded neural signals is fast enough to be performed in real time, i.e. in less than one minute, for example about 30 seconds. The comparison of two consecutive neural signals with their respective neural activity can be performed by plotting the neural activity (here the amplitude of the peak of the spectral signal) as a function of the lens power, as shown in FIG. 6. The comparison of two consecutive neural signals with their respective neural activity can also be performed by calculation.

[0151] Thus, refraction testing according to the method of the present invention, performed using an autorefractometer driven by a brain-computer interface, can be performed in approximately 10 minutes globally, much less time than subjective refraction testing, which takes approximately 15-20 minutes.

[0152] As soon as a maximum neuronal activity is found, step a) can be stopped and step b) starts (blocks E6 and E7 in FIG. 1).

[0153] Step b) Once the maximum neural activity is found, the reduced neural activity from such maximum neural activity is determined.

[0154] For example, as described above in connection with the apparatus of the present invention, a given reduced neural activity is determined by correcting the maximum neural activity based on a coefficient k related to the ocular accommodation response.

[0155] In other words, a correction factor k is applied to that maximum neural activity to find the reduced neural activity that would be obtained if the brain did not exhibit neural artifacts due to modulation (block E6 in FIG. 1).

[0156] For example, the coefficient k is provided to the memory 150 of the control unit 15 prior to execution of the method.

[0157] In practice, the coefficient k is applied to the feature of the neural signal on which the determination of the maximum neural activity is based, where the coefficient k is multiplied to the maximum amplitude Amax obtained from the analysis of the spectral signal to find the reduced amplitude A(reduced). A(reduced)=k*Amax

[0158] The coefficient k is previously determined in a step prior to step b), preferably before carrying out the method of the invention, in the manner described above in connection with the apparatus.

[0159] As explained in the device description, the coefficient k is a constant value that depends on the features of the neural signal on which the neural activity is determined.

[0160] For example, if the feature of the neural signal indicative of neural activity is the amplitude of a peak in the spectral signal, the coefficient k is a constant value chosen between 0.6 and 0.95, preferably between 0.7 and 0.9. The coefficient k may in particular be equal to 0.6, 0.65, 0.7, 0.75, 0.8, 0.85, 0.9 or 0.95.

[0161] Furthermore, the coefficient k may depend on the subject's age and / or the subject's ocular imperfections, such as refractive correction.

[0162] The reduced neural activity can be the neural activity of one of the recorded neural signals, or it can be the neural activity of an extrapolated neural signal.

[0163] In other words, the neural signal exhibiting reduced neural activity (here considered to be the neural signal exhibiting a reduced amplitude Areduced in the spectral signal) is one of the recorded neural signals or is extrapolated from the recorded neural signal. Extrapolated means that the neural signal was not recorded but is inferred from the recorded neural signal. Such extrapolation can be, for example, a graphical or mathematical extrapolation.

[0164] It can then be inferred, for example based on Figure 7, which lens power will give such a neural signal of reduced neural activity (block E7 of Figure 1), such lens power being the optimal correction of the refraction of the test subject.

[0165] Thanks to the method and device of the invention it is therefore possible to objectively identify the best correction for a test subject, even if the test subject is uncertain about his perception.

[0166] The present invention is not limited to what has been described herein above.

[0167] In particular, in a second advantageous embodiment of the method and device of the invention, the control unit 15 may be adapted to further provide a machine learning algorithm.

[0168] Such machine learning algorithms may be used, in particular, in step b) to provide both reduced neural activity and a lens power that should result in such reduced neural activity (blocks E6 and E7 of FIG. 1).

[0169] Such a second embodiment is particularly interesting when the coefficient k is expected to vary significantly across individuals.

[0170] Machine learning algorithms can process highly complex neural signals and relate such neural signals to optimal compensation and / or accommodation responses of the eye that induce neural artifacts in the subject's brain.

[0171] More precisely, machine learning algorithms take as input a training set of observed data points to "learn" a data structure, such as a formula, a set of rules, or other data structure. This learned data structure or statistical model can then be used to make generalizations about the training set or predictions about new data. As used herein, a "statistical model" refers to any learned and / or statistical data structure that establishes or predicts a relationship between two or more data parameters (e.g., inputs and outputs). The invention is described below with reference to neural networks, although other types of statistical models may be employed in accordance with the invention. For example, each data point in a training data set may include a set of values ​​that correlate with or predict another value at the data point.

[0172] Here, a machine learning algorithm is trained using a set of neural signals recorded for many subjects whose eyes were provided with a series of distinct lens powers and whose optimal correction is already precisely known because it has been determined, for example, by subjective refraction testing.

[0173] In the present invention, the machine learning algorithm can be configured to associate the neural activity of each recorded neural signal provided to the machine learning algorithm with the neural activity of the optimal correction of the subject's refraction. In other words, the input of the machine learning algorithm can be the recorded neural signals of the test subject, including the neural signal showing the maximum neural activity, and its output can be the neural signal showing the given neural activity that is characteristic of the optimal correction of the subject. Here, the given neural activity is determined by the machine learning algorithm as a reduced neural activity compared to the maximum neural activity. The reduced neural activity is characteristic of the subject clearly seeing the visual stimulus without accommodation (or with as little accommodation as possible).

[0174] Such machine learning algorithms of the control unit 15 can be based either on Long Short-Term Memory (LSTM) techniques or on Convolutional Neural Networks (CNN).

[0175] The LSTM technique is part of the Recurrent Neural Network (RNN). Classical RNN techniques include a network of neural nodes organized in successive layers. Each node (also called a neuron) in a given layer is connected in a one-way manner to each of the nodes in the next layer. This structure makes it possible to take previous moments into account in the neural network, since a first layer for a previous moment t-1 is connected to a second layer for the current moment t. This second layer is also connected to a third layer for the subsequent moment t+1, thus having multiple layers. Thus, each signal provided as input is processed in a temporal manner, taking into account the signals provided at previous moments.

[0176] CNN technology uses signals as images rather than a temporal method. The acquired signals are processed at once with all the data acquired during a given test period. Then, a mathematical image processing operation, such as convolution, is applied to the image obtained with the acquired signals to determine the output of the machine learning algorithm.

[0177] The machine learning algorithm may include a guide model that defines decision rules, and such guide model is configured to guide the predictions of the machine learning algorithm. These rules may include sub-correlation between the recorded neural signal showing the maximum neural activity and the recorded neural signal obtained for the optimal correction of the subject. For example, the guide model can specify that a given variation of a particular characteristic between the neural signal showing the maximum neural activity and the neural signal obtained with the optimal correction must correlate with a variation of accommodation (from a small accommodation response to no or almost no accommodation), and therefore must correlate with an optimal correction of refraction. In another example, the guide model can specify that a given combination of variations of a particular characteristic between the neural signal showing the maximum neural activity and the neural signal obtained with the optimal correction means a variation of accommodation (from a small accommodation response to no or almost no accommodation), and therefore means that an optimal correction of refraction has been found. This guide model allows the correlation by machine learning to be facilitated, thus reducing the time taken for correlation and improving its accuracy.

[0178] The control unit 15 can use a machine learning algorithm that is already trained, i.e. the neural network of the machine learning algorithm already contains a set of formulas or rules configured to provide a correlation between the recorded neural signal exhibiting the maximum neural activity and what should be the neural activity of the neural signal characteristic of the optimal correction, or the control unit 15 is configured to train a machine algorithm to determine this correlation.

[0179] The training of the machine learning algorithm is preferably carried out by providing the algorithm with a plurality of recorded neural signals related to a set of initial subjects whose optimal correction is known. By "initial subject" is meant a subject who participates in the learning of the machine learning algorithm. In other words, the initial subject provides data that allows the machine learning algorithm to correlate the recorded neural signals with each lens power provided to the subject, in particular to correlate the maximum neural activity among all recorded neural signals with the reduced neural activity of the neural signals associated with the optimal correction. Conversely, by "test subject" is meant a subject for whom a determination of the optimal correction is performed based on the machine learning algorithm, i.e. a subject for whom a prediction of his / her optimal correction can be performed based on the recorded neural signals.

[0180] This training is repeated many times to make the algorithm more accurate. As an example, the training of the algorithm may include at least 100 initial subjects, preferably 1000 initial subjects. [Explanation of symbols]

[0181] 1 eye 2 Armchairs 3 Front support part 10 equipment 11 Neural activity sensors 12 Control unit 13 Automatic refractometer 15 Control Unit 50 glasses 51 Active Power Lenses 52 Frames 110 electrodes 111 Additional electrodes 150 Memory 151 processor

Claims

1. 1. A method for objectively determining optimal correction of a subject's ocular refraction, comprising: a) providing, by a control unit, successive discrete lens powers to the subject's eye until a recorded neural signal corresponding to one of the lens powers exhibits a maximum neural activity compared to all other recorded neural signals, and recording corresponding successive neural signals of the subject while the eye of the subject is receiving visual stimulation through each lens power; b) determining by the control unit that the optimal correction of the subject's eye refraction is a lens power for which the subject exhibits a neural signal indicative of a given reduced neural activity compared to the maximum neural activity of step a); A method comprising:

2. 2. The method of claim 1, wherein in step b), the given reduced neural activity is determined by correcting the maximum neural activity based on a predetermined coefficient k related to an onset of an accommodation response of the subject's eye.

3. In step a), - the lens powers provided to the eye of the subject are successively smaller and smaller, and an initial lens power provided to the eye blurs the vision of the subject; - the neural activity of a last recorded neural signal is compared with the neural activity of a previously recorded neural signal, - step a) is stopped when the previously recorded neural signal exhibits more neural activity than the last recorded neural signal, the previously recorded neural signal being the one exhibiting the greatest neural activity; The method of claim 1.

4. 2. The method of claim 1, wherein in step a), each successive lens power provided to the eye of the subject is selected based on a reference lens power that is an approximation of the optimal correction for the subject.

5. 5. The method of claim 4, wherein each lens power is given in diopters (D) and the diopter step between two successive lens powers is smaller when the lens power is closer to the reference lens power than when the lens power is further from the reference lens power.

6. The method of claim 1 , wherein step a) is performed by a brain-computer interface, and the lens power is automatically changed based on an analysis of the neural activity of previously recorded neural signals.

7. 2. The method of claim 1, wherein in step b), the neural signal indicative of the reduced neural activity is one of the recorded neural signals or is extrapolated from the recorded neural signals.

8. The coefficient k is calculated by the following steps: c1) selecting a group of subjects for which the optimal correction is known; c2) for each subject, - recording successive neural signals of the subject while at least one eye of the subject receives visual stimuli through a succession of distinct lens powers including the lens power of the best correction; - analyzing the neural activity of each recorded neural signal and deducing the recorded neural signal that exhibits the greatest neural activity; - comparing said maximum neural activity with the neural activity of said neural signal recorded for said lens power of said best correction, c3) estimating the coefficient k from the step of comparing the neural activity performed for each subject; The method of claim 2 , wherein the concentration is determined by:

9. The method of claim 8 , wherein in steps c2) and c3) the operations of analyzing, comparing and inferring are performed by machine learning.

10. The method of claim 1 , wherein the neural activity of the subject associated with a lens power is determined by extracting at least one feature of the neural signal recorded for the lens power.

11. 11. The method of claim 10, wherein the feature is an amplitude in a spectral signal derived from the recorded neural signal, and higher neural activity corresponds to a higher amplitude in the spectral signal.

12. 1. An apparatus for objectively determining an optimal correction of a subject's ocular refraction, comprising: at least one neural sensor for detecting neural signals originating from at least one area of ​​the subject's brain, said neural signals being linked to the visual acuity of the subject's eye; a control unit, a) recording successive neural signals from the subject while at least one eye of the subject receives visual stimulation through successive distinct lens powers until one of the recorded neural signals exhibits a maximum neural activity compared to all other recorded neural signals; b) determining that the optimal correction of the subject's ocular refraction corresponds to a lens power for which the subject exhibits a neural signal indicative of a given reduced neural activity compared to the maximum neural activity obtained in step a); and a control unit adapted to:

13. 13. The device of claim 12, further comprising an autorefractometer controlled by the control unit to automatically change the power of the lenses (L1, L2, L3, L4, L5) through which the eye of the subject receives the visual stimulus in response to an analysis of neural activity of previously recorded neural signals.

14. 13. The apparatus of claim 12, wherein the neural sensor includes at least three electrodes positioned on the back of the subject's head and integrated into a headrest of a chair to record neural signals originating from the back of the subject's brain.

15. 13. The device of claim 12, wherein the device comprises at least one active power lens whose power is driven by the control unit to provide the distinct lens power at which the eye of the subject receives the visual stimulus in response to analysis of neural activity of a previously recorded neural signal.

Images