METHOD AND DEVICE FOR CALCULATING AN INTRAOCULAR LENS

Abstract

A computer-implemented method for calculating an intraocular lens (IOL) is provided. The method involves calculating the IOL based on a biometric parameter of the eye to be treated. The IOL calculation takes into account the expected change in intraocular pressure (IOP) resulting from treatment with the IOL, compared to the IOP before treatment, the expected IOP after treatment, and / or the IOP before treatment.

Description

[0001] The present disclosure relates to a computer-implemented method for calculating an intraocular lens (IOL).

[0002] The present disclosure relates to a computer-implemented method for training an artificial intelligence-based model which, after training, is configured to calculate an IOL.

[0003] The present disclosure relates to a computer-implemented method for training an artificial intelligence-based model which, after training, is configured to determine a post-operative intraocular pressure and / or a change in intraocular pressure from pre- to post-operative intraocular pressure.

[0004] A data processing device can be provided which is designed to at least partially carry out one, several or all of the aforementioned methods.

[0005] A computer program and / or a computer-readable medium may be provided, comprising instructions which, when the computer program or instructions are executed by a computer, cause the computer to at least partially carry out one, several or all of the aforementioned methods.

[0006] An intraocular lens, or IOL for short, is an artificial lens implanted into the eye. It can be used to treat, and optionally correct, cataracts and / or refractive errors. A special case for the exclusive correction of refractive errors is the so-called 'phakic' IOL, which is inserted anterior to a still intact, possibly still accommodating, crystalline lens.

[0007] IOLs differ in particular in terms of whether they are placed in the anterior or posterior chamber of the eye, the required incision size (<2 mm (micro-incisional cataract surgery, MICS for short) or larger), the diameter of the optical zone (for example, 5 to 7 mm or <1.4 mm in the case of small-aperture IOLs for focus extension), the type of mount (for example, capsular bag-supported, sulcus-supported, scleral-fixed, iris-fixed), the lens material (rigid or foldable; PMMA, acrylic, silicone, hydrogel; hydrophobic, hydrophilic; clear, yellow, possibly transparent) and the type of IOL used.or specially absorbent, for light-based postoperative corrections) and the haptic material (nitinol or polymer), haptic type (e.g. C-shape, plate shape), the realized focus type (e.g. monofocal, astigmatism-correcting, EDOF or multifocal) and the lens shape (especially spherical, toric, but also with regard to the lens edge design, for example to reduce glare (“glistening”) or the “after-cataract” risk (cell overgrowth)).

[0008] The term cataract refers to the eye disease “cataract,” in which the lens of the eye becomes cloudy, hardened, and sometimes thickened, so that vision gradually deteriorates, starting with reduced perception of details or contrasts, and continuing with increased perception of scattering (halos) around high-contrast light sources, and eventually leading to complete blindness in the case of an untreated cataract.

[0009] The term ametropia can be understood in this case to mean, among other things, an eye in which the length of the eye is not exactly in proportion to the refractive power (ability to focus light rays) of the eye and / or the components of the eye are not coordinated in such a way that normal visual acuity is developed (as is the case with astigmatism, for example).

[0010] If only the term "ametropia" is used below, this should be understood to mean, where technically appropriate, ametropia and / or cataracts. If only the term "cataract" is used below, this should be understood to mean, where technically appropriate, ametropia and / or cataracts.

[0011] Traditionally, IOLs are selected or fitted based on measured and / or estimated measurements of the eye to be treated, known as biometric parameters. Even accommodating IOLs under development may require their average refractive power to be adjusted to the patient's biometric parameters due to their limited range of accommodation. IOLs are typically offered with refractive powers ranging from 5 to 30 D, in increments of 0.5 D, but higher powers are occasionally produced for the treatment of special cases. Multiple IOLs can also be implanted in combination.

[0012] The selection of the appropriate IOL for a patient is the responsibility of the surgeon or treating physician. The surgeon must consider many factors. Firstly, the appropriate method for calculating the IOL refractive power should be selected depending on the individual biometric parameters of the eye. For this purpose, various more or less suitable formulas are generally used for exceptionally long, normal, or exceptionally short eyes. In the simplest case, the input parameters are based on keratometry and the axial length of the eye. Due to their simplified model assumptions, the formulas usually also contain an empirically determined correction factor, such as the so-called A-constant. The axial length is usually understood to be the distance from the corneal vertex to the fovea on the visual axis of an eye, i.e. on the axis from a fixated object point to the center of sharpest vision on the retina.

[0013] A widely used calculation method is the so-called IOL formula, e.g., according to Haigis, Holladay, Hoffer, Olsen, Shammas, or SRK. According to this formula, the patient's refraction D (initial / evaluation parameter) after IOL insertion is calculated as D=DIOL−f(K,AL,VKT,A) where f() is a classically known IOL formula, D IOL the refractive power of the IOL, K corneal refractive power or powers, AL is the measured axial length of the eye, VKT the measured anterior chamber depth, and A IOL type-dependent constant input variables are.

[0014] Advanced formulas, such as the Barrett formula or the Holladay 2, use additional fitting parameters, such as the crystalline lens thickness LT or the corneal diameter (WTW = white-to-white), the pre-operative refraction or the patient's age, which are primarily intended to help better take into account the effective lens position (ELP) of the IOL.

[0015] The various calculation methods (IOL calculation formulas) generally use different constants (i.e., IOL constants) depending on the IOL type. An A-constant is used, for example, in the SRK formula. In some cases, such as the Holladay formula, a so-called "surgeon factor" is also used. The surgeon factor is a value assumed by the surgeon based on their experience for the distance between the effective optical plane of the IOL and the postoperative iris plane.

[0016] To select the IOL, the physician specifies a target refraction (D = D Ziel ). For optimization, the physician calculates the refraction according to the formula described above for different IOLs by varying D IOL and A. In many cases, the physician uses IOLs of the same type, so there is no variation in A and the optimization is based on a formula calculation according to D IOL = D Ziel+ f(K, AL, VKT, A). For the case of emmetropia as the target, this results in the classic formula calculation of the IOL according to D IOL = f(K, AL, VKT, A). Often, a slight myopia is also desired, e.g., D ziel = -2 to -3 dpt, which allows the patient to do without glasses at home or when reading.

[0017] The constant A in the formulas can be determined empirically using a patient group to adjust the formula values ​​to the actual optimal refraction values. However, this adjustment only ensures that the mean of the refraction values ​​across the test group agrees with the formula.

[0018] An alternative method for calculating the IOL is ray tracing. Ray tracing is a process of ray tracing. Objects in the environment are only perceived by a person because they are illuminated by a light source and the objects reflect or backscatter these light rays, some of which ultimately reach the eye. The ray tracing process simulates this elementary natural phenomenon. If the optical system, i.e. the individual human eye with all its optical elements, is known, ray tracing can be used to calculate a "real" image created on the retina. The process is therefore based on a detailed eye model, possibly also using a complex corneal topography of the eye. This method does not use general correction factors (A constants), but certain assumptions are made regarding the effective (post-operative) lens position (ELP), if applicable.This is also necessary considering the tilt of the IOL to be implanted. This method is suitable for eyes with a wide variety of biometric parameters, such as long, normal, short, or after laser-based refractive correction (post-LVC), etc.

[0019] Ray tracing is then used to calculate the IOL's refractive power and residual refraction. To achieve a good correlation with subjective visual acuity, i.e., a result comparable to the patient's perception, various selection criteria or metrics can be used for the calculation. Although retinal image metrics have proven particularly suitable for this purpose, the following other selection criteria are also conceivable: - Evaluation of the image on the retina with regard to moments, entropy, compactness, shape or intensity distribution using point spread function (PSF), line spread function (LSF) or root mean square, - Evaluation of resolution by optical transfer function (OTF), such as modulation transfer function (MTF) or phase transfer function (PTF), - Evaluation of contrast by the contrast sensitivity function (CSF), - Evaluation of optical aberrations such as chromatic aberration, ray aberration, wavefront aberration, depth of field and binocular aberration of the image scale, - Evaluation of the classic refractive parameters diopter and astigmatism, usually on the visual axis through the fovea as the center of sharpest vision, but possibly also in the periphery of the eye.

[0020] The list provided here is merely exemplary, as other optical evaluation parameters known to those skilled in the art can also be used. In principle, any evaluation parameters or criteria that can be used to assess and quantify deviations from the ideal wavefront are applicable.

[0021] It is an object of the present disclosure to enrich the prior art outlined above.

[0022] According to the disclosure, this object is achieved by the subject matter of the independent claims, with optional further developments thereof being specified in the subordinate claims and the subclaims.

[0023] The problem is then solved by a computer-implemented method for calculating an intraocular lens (IOL), comprising calculating the IOL based on a biometric parameter of an eye to be treated with the IOL. When calculating the IOL, an expected change in intraocular pressure due to treatment of the eye to be treated with the IOL compared to the intraocular pressure of the eye to be treated before treatment of the eye to be treated with the IOL, an expected intraocular pressure after treatment of the eye to be treated with the IOL, and / or the intraocular pressure before treatment of the eye to be treated with the IOL are taken into account.

[0024] That is, (1) an expected change in intraocular pressure, (2) an expected post-treatment intraocular pressure, and / or (3) a pre-treatment intraocular pressure can be used for the procedure.

[0025] A computer-implemented method can be understood as a method in which one, several or all steps of the method are carried out at least partially by a computer or a data processing device.

[0026] For the definition of an IOL, please refer to the above statement. Furthermore, the above statements regarding IOL calculation also apply analogously to the disclosed solution. It is therefore conceivable that the method takes into account an (optionally predetermined) target refraction and / or an (optionally predetermined) target visual acuity when calculating the IOL. This target refraction can be 0 D. In this case, the target refraction does not necessarily have to be entered into the algorithm that calculates the IOL. However, it is also conceivable that a target refraction other than 0 D is used, which is entered as a further input into the algorithm, for example, into the AI ​​model described later. IOLs are generally only available in graduated powers, and a finite residual refractive error is therefore to be expected post-op. The target refraction can therefore also be understood as a target refraction range, for example -2.5 ± 0.25 D.Analogously, target visual acuity can be understood as a target visual acuity range, such as a target visual acuity > 0.9 for a selected reading distance of, for example, 0.4 m.

[0027] The eye to be treated with the IOL, or in short the eye to be treated, can be understood as the eye, for example of a human, into which an IOL is to be implanted or inserted.

[0028] Intraocular pressure can be understood as the physical excess pressure inside the eye to be treated compared to the surrounding atmospheric pressure.

[0029] The intraocular pressure before treatment of the eye to be treated with the IOL, or preoperative intraocular pressure, can be understood as the intraocular pressure measured during an (optionally predetermined) period before the treatment. Preoperative intraocular pressure can therefore be the result of tonometry. Since intraocular pressure generally follows a daily pattern (e.g., with maximum pressure typically reached in the morning, for example, at 8:00 a.m.), tonometry can also be a series of tonometric measurements taken over a predetermined period of time, e.g., a day and / or a night. The intraocular pressure can be described, for example, by a mean value (optionally with a range of variation) and / or minimum and / or maximum value(s).Intraocular pressure can exhibit a heartbeat-dependent modulation, the so-called ocular pulse amplitude (OPA), which typically ranges from 2 to 4 mm Hg, but in extreme cases can also range from 1 to 8 mm Hg. Therefore, tonometers are known in the state of the art that measure IOP values ​​using a pulse sensor in response to heart rate to increase accuracy (see US 7101335 B2). However, since intraocular pressure also changes with blood pressure, a combined intraocular and blood pressure measurement may be useful, for example, to better evaluate or, if necessary, discard outlier values ​​in intraocular pressure.

[0030] The expected intraocular pressure after treatment of the eye to be treated with the IOL or the expected postoperative intraocular pressure can be understood as an estimated or determined value for the intraocular pressure at an (optionally predetermined) preoperative time point.

[0031] The intraocular pressure after treatment of the eye to be treated with the IOL or the postoperative intraocular pressure can be understood as a value for the intraocular pressure measured at a post-operative time point.

[0032] The expected change in intraocular pressure due to treatment of the treated eye with an IOL can be understood as the delta, or difference, that results in the intraocular pressure from preoperative to postoperative. Thus, with the disclosed method, it is not mandatory to determine or estimate the postoperative intraocular pressure. Instead, the change in intraocular pressure caused by the treatment can also be used, for example, as a relative percentage and / or as an absolute difference value.

[0033] The method described above offers a number of advantages, as described below. Among other things, the method enables a more precise IOL calculation than the state of the art, i.e. one that is more suitable for the patient to be treated, and thus a better automated selection of a suitable IOL. The reasons for this are explained in detail below. Furthermore, when faced with several alternative solutions for potentially achieving a target refraction, e.g., using different selectable IOL types, it enables the physician to select the IOL that, if necessary in combination with other pressure-reducing measures, is most likely to achieve the most suitable reduction in intraocular pressure for the patient. For example, a maximum pressure reduction in the case of acute angle-closure glaucoma or no or the smallest possible pressure reduction in the case of a patient suffering from hypotony.

[0034] In detail: The conventional methods for calculating the IOL described above and the decision for selecting an IOL based on this sometimes lead to unsatisfactory surgical results, i.e. the refractive goal, for example emmetropia, is sometimes not sufficiently achieved with the conventional method.

[0035] According to the disclosure, it was surprisingly discovered that this unsatisfactory result may be due, among other things, to the fact that the biometric parameters or dimensions of the treated eye, in particular the axial length of the treated eye, may change after insertion, i.e., postoperatively. However, this change is not traditionally taken into account when calculating the IOL, so that the IOL selected based on the calculation is suitable for achieving a desired refractive goal, for example, emmetropia, for the preoperative eye but not for the postoperative eye.

[0036] It has also been recognized that intraocular pressure, preoperatively, postoperatively and / or a change in intraocular pressure due to IOL insertion, influences the change in the biometric parameters of the treated eye.

[0037] More specifically, cataract surgery often results in a reduction in intraocular pressure. It is generally assumed that the opening of the chamber angle that occurs when the natural lens is removed relaxes the tissues there (trabecular meshwork) and the tissues in the region of the uveoscleral outflow pathway, and / or that the application of phaco energy during phacoemulsification of the crystalline lens causes “SLT-like” pressure-reducing effects (i.e., tissue loosening due to cytokine release similar to that seen during selective laser trabeculoplasty in glaucoma therapy), and / or that irrigation (i.e., rinsing) of the interior of the eye during surgery loosens and removes materials (e.g., pigment particles) that may be blocking the trabecular meshwork. Such pressure reductions are highly desirable for patients with ocular hypertension (particularly those diagnosed with glaucoma), and in cases of angle-closure glaucoma (ACG), they are often an important therapeutic option.

[0038] However, the axial length (AL) of the eye also changes with intraocular pressure, for example, after trabeculectomy or pharmacological interventions. C. Leydolt, O. Findl, and W. Drexler ("Effects of change in intraocular pressure on axial eye length and lens position," Eye (2008) 22, 657-661; doi:10.1038 / sj.eye.6702709) cite, for example, a reduction of 46 µm with a 5.7 mm Hg pressure reduction. A systematic study (also by C. Leydolt et al.) with briefly induced increases in intraocular pressure also provided the following relationship between the axial length of the eye and changes in intraocular pressure: approximately ~2 µm / mm Hg.

[0039] Furthermore, it is assumed that other biometric parameters change as intraocular pressure decreases. For example, pig eye experiments have shown that keratometry flattens by approximately 2 dpt / 10 mm Hg (KJ Maverick, Effect of Intraocular Pressure on Keratometry in Porcine Eyes, ARVO 2004).

[0040] However, since current IOL calculation formulas are based on biometric parameters that are independent of intraocular pressure, both unintentional and intentional reductions in intraocular pressure following IOL insertion (e.g., during cataract surgery) traditionally result in an error in the IOL prediction or IOL calculation.

[0041] This error would be particularly significant if a significant pressure reduction were achieved during cataract surgery, for example, in the emergency treatment of narrow-angle glaucoma. The literature (D. Melancia, "Cataract Surgery and Intraocular Pressure," Ophthalmic Res 2015;53:141-148, DOI:10.1159 / 000377635) cites pressure reductions of 38 mm Hg for such cases. Based on the data cited by Leydolt et al. for pressure-dependent axial length changes, one would expect a substantial IOL calculation error of more than 1 dpt of hyperopia.

[0042] Furthermore, the situation is complicated by the fact that changes in intraocular pressure can also change the corneal refractive power (increase in refractive power when pressure is reduced), which can cause previously unforeseeable effects, particularly in the case of patients with unnatural corneal curvature, for example after refractive corneal surgery, which is why there are discussions in the literature about the associated “refractive surprises”.

[0043] The disclosed solution therefore offers the technical effect that this refractive error can be (significantly) reduced by taking the intraocular pressure into account when calculating the IOL.

[0044] Optional further training courses on the procedure described above are described in detail below.

[0045] The method may include determining the expected change in intraocular pressure and / or the expected intraocular pressure after treatment based on the intraocular pressure before treatment, an (optionally predetermined) property of the IOL, an eye geometry of the eye to be treated, patient data of a patient having the eye to be treated, information regarding a planned surgical intervention of the eye to be treated to be combined with the IOL treatment, an LVC parameter and / or biomechanical corneal parameter of a cornea of ​​the eye to be treated.

[0046] An example of a biomechanical corneal parameter is corneal hysteresis (CH), which is also determined in some NCT tonometry studies. Another example is elasticity information or movement assessments of ocular tissues, such as the pulsating movements of the trabecular meshwork, obtained using elastography techniques (optical coherence elastography, OCE, or Brillouin microscopy). Biomechanical parameters, such as corneal parameters, can vary depending on the patient's age, ethnicity, gender, or previous treatments, for example, whether and to what extent cross-linking was performed to stabilize the cornea during keratoconus treatment.

[0047] The IOL type can influence the pressure reduction. For example, the thickness of the IOL can influence the chamber angle opening and / or the aqueous humor flow. For example, there are phakic IOLs that have a hole to allow flow to the endothelium. Additionally or alternatively, external mechanical pressure on the ciliary body, applied by the IOL, e.g., by an IOL haptic pressing on the ciliary body, can influence aqueous humor production. Additionally or alternatively, a drug depot attached to the IOL haptic can reduce intraocular pressure for a specific period of time. The shape of the haptic can determine the post-operative position and / or location of the IOL, for example, its distance from the iris plane, and thus influence the aqueous humor flow through the pupil. The properties of the IOL can therefore include, among others: - a dimension, optionally a thickness, of the IOL, - a shape, optionally a haptic, of the IOL, - a mechanical pressure exerted by the IOL, optionally the haptics of the IOL, on an ocular structure, optionally a ciliary body, of the eye to be treated in the implanted state of the IOL, and / or - the presence, optionally the size, of a drug depot on the IOL.

[0048] An LVC parameter (LVC stands for Laser Vision Correction) refers to a patient's corneal parameter after refractive corneal surgery, such as LASIK (Laser Assisted In-Situ Keratomileusis), SMILE (Small Incision Lenticule Extraction), or LIRIC (Laser Induced Refractive Index Change). This can include, for example, a "modified" corneal refractive power, which may be limited to a specific pupil diameter, a change in pachymetry (corneal thickness distribution), residual stromal thickness, lenticule or flap thickness, or diameter.

[0049] Eye geometry can describe the size and / or shape of an eye structure and / or the relative position of eye structures. An assumed eye geometry can be described by a combination of measured biometric parameters relating to the eye geometry and geometric assumptions (e.g., from statistical data, approximate formulas, and / or extrapolation and interpolation formulas, etc.).

[0050] The eye geometry may include information regarding a chamber angle geometry and / or anterior chamber geometry and / or posterior chamber geometry and / or retinal geometry of the eye to be treated.

[0051] The anatomical structure formed by the cornea and iris in the anterior chamber of the eye is called the iridocorneal angle (angulus iridocornealis). A large portion of the aqueous humor drains through the iridocorneal angle, or rather, the trabecular meshwork within it with Schlemm's canal behind it. Depending on the geometry of the iridocorneal angle, i.e., the width of the iridocorneal angle, more (wide angle) or less (narrow angle) aqueous humor can drain, which in turn influences intraocular pressure.

[0052] Regarding ocular geometry, reference is also made to Lin et al., "PREDICTORS OF INTRAOCULAR PRESSURE AFTER PHACOEMULSIFICATION IN PRIMARY OPEN-ANGLE GLAUCOMA EYES WITH WIDE VERSUS NARROWER ANGLES," Trans Am Ophthalmol Soc 2017;115:T6[1-13], where they showed that pressure reductions were greater in patients with narrow angles than in those with wide angles. Even more pronounced differences were shown by D. Melancia, "Cataract Surgery and Intraocular Pressure," Ophthalmic Res 2015;53:141-148, DOI: 10.1159 / 000377635.

[0053] The eye geometry may, as described above, include information concerning an anterior chamber geometry and / or posterior chamber geometry and / or retinal geometry of the eye to be treated.

[0054] Relevant anterior chamber structures include the chamber angle, cornea, iris, and lens (natural or phakic IOL), Schlemm's canal itself (optionally its cross-section), and the location, number, and size of the collecting vessels adjacent to Schlemm's canal, which drain the aqueous humor into the episcleral venous system. Schlemm's canal and the collecting vessels can be measured, for example, using suitable OCT imaging or fluorescence imaging with contrast agents.

[0055] In addition to the aqueous humor-producing ciliary body, a relevant posterior chamber structure of the eye can also be the nerve fiber head. The nerve fiber head can be altered in shape and size in glaucoma patients and can thus have varying effects on the nerves, arteries, and veins running nearby. This can indirectly influence the expected intraocular pressure response of the eye during cataract surgery. Likewise, the thickness of the choroid beneath the retina, which is often thinned in myopes, can influence changes in intraocular pressure following surgery. Also located in the posterior chamber of the eye is the vitreous body, which may be attached to the retina or, optionally in older patients, may already be detached from the retina, partially liquefied, or in some cases even completely or partially surgically removed via vitrectomy.The vitreous body can thus directly influence the aqueous humor circulation in the eye. During cataract surgery, with the application of ultrasound energy during phacoemulsification, it can be observed that a previously undissolved vitreous body is detached at the nerve fiber head, and a vitreous opacification that may be present at this interface, the so-called white ring, becomes mobile and can be perceived postoperatively by the patient in more central areas of the visual field. In this respect, vitreous opacities can also be relevant ocular structures of the eye geometry, providing information about expected changes in intraocular pressure. Choroidal thickness, nerve fiber head, vitreous body and vitreous opacities, and ciliary body can be detected and measured using suitable OCT or ultrasound imaging techniques, for example.

[0056] The information regarding the planned surgical intervention in the treated eye to be combined with the IOL treatment may include information on the number and / or type of MIGS and / or TM aspiration.

[0057] The planned surgical intervention to be combined with IOL treatment may be (microinvasive) glaucoma surgery, which aims to improve aqueous humor outflow and thus influence, and in particular, reduce, intraocular pressure. Information about the number, position, and / or type of implants or stents used (so-called MIGS) may be considered, particularly trabecular meshwork shunts (e.g., iSTENT), Schlemm's canal stents (e.g., HYDRUS), suprachoroidal stents (e.g., MINIJECT), or subconjunctival stents (e.g., XEN). Microinvasive glaucoma surgery may also involve trabecular meshwork aspiration (TM-aspiration) or one or more excimer laser trabeculotomies (ELT) or one or more trabecular meshwork trephinations (iAccess) or canaloplasty, particularly viscocanaloplasty of Schlemm's canal.In TM aspiration and trabeculoplasties, the proportion of the treated trabecular meshwork (usually expressed in "clock hours") is a key parameter for the degree of pressure reduction. Examples of more invasive glaucoma surgical procedures that can also be combined with IOL treatment are trabeculectomy or the implantation of a tube shunt, both of which drain into a subconjunctival drainage bleb. Less common but in principle combinable is cyclodialysis, in which an incision is made to create a gap between the ciliary body and the scleral spur that drains aqueous humor. In principle, it can be combined with cyclocoagulation, for example, cyclophotocoagulation (CPC) or cyclocryocoagulation (CKC), or ultrasound cycloplasty (UCP), in which reduced aqueous humor production in the ciliary body is achieved through a generally destructive alteration of the ciliary body.

[0058] The biometric parameters can include axial length, corneal diameter, corneal refractive power (including astigmatism), corneal thickness, anterior chamber depth, and / or lens thickness of the treated eye. These biometric parameters can be determined using standard commercial biometers, such as the IOL-Master 700®. Refractive errors of the patient's eye can also be determined using phoropters, autorefractors, or wavefront measuring devices. However, preoperative measurements of these devices include the refractive component of the crystalline lens to be removed during cataract surgery. Therefore, there are known proposals in the state of the art to determine biometric parameters related to refraction using intraoperative wavefront measurements, if necessary, and to adjust IOL calculations and selections if necessary.

[0059] The patient data may include data or information about the patient’s age, gender, and / or ethnicity.

[0060] The patient data may include information regarding the number and / or type of prior treatment of the patient and / or regarding existing and / or planned medication and / or information about the patient's blood pressure (optionally for systolic and diastolic measurements and, if applicable, distribution data such as minimum, maximum and mean values ​​for measurements over time intervals of minutes to days) and / or (ocular) implants of the patient.

[0061] The above-mentioned parameters or information directly and / or indirectly influence the change in intraocular pressure and / or the intraocular pressure before and / or after treatment, so that by taking them into account, the prediction of the expected change in intraocular pressure and / or the expected intraocular pressure after treatment and thus the prediction of the postoperative biometric parameters (optionally the axial length) can be improved, which in turn enables a more accurate calculation of the IOL.

[0062] An initial model based on artificial intelligence can be used to determine the expected change in intraocular pressure and / or the expected post-treatment intraocular pressure.

[0063] Consequently, a first model based on artificial intelligence can be provided in the form of or as part of a computer program. That is, the first model based on artificial intelligence can be designed to be used by a computer as part of or in the form of a computer program.

[0064] In this context, an artificial intelligence-based model can be understood as a model created using machine learning. Machine learning can be understood as the "artificial" generation of knowledge from experience. An artificial system or model learns from one or more examples and can generalize them after the learning phase or training is completed. To do this, machine learning algorithms build a statistical model based on training data, which can optionally be tested against test data. This means that the examples are not simply memorized, but patterns and regularities are recognized in the training data. This allows the model to evaluate even unknown data (so-called learning transfer).

[0065] The model may comprise an artificial neural network. Such networks can be based on the networking or connection of many artificial neurons. It is conceivable that the artificial neural network can be trained or learn using one or more of the following methods: developing new connections between the neurons, deleting existing connections between the neurons, changing the weighting (i.e. a weight of one of the connections between the neurons), adjusting a threshold value of one of the neurons (if this has a threshold value), adding and / or deleting neurons, and / or modifying the activation, propagation and / or output function. It is also conceivable that the learning behavior of the artificial neural network changes when the activation function of the neurons or the learning rate of the network is changed.Such a network typically "learns" primarily by modifying the weights of the connections between neurons. Adjusting the threshold can be accomplished by an on-neuron. This enables such networks to learn complex nonlinear functions using a "learning" algorithm that attempts to determine all the function's parameters from existing input and desired output values ​​using iterative or recursive procedures.

[0066] The first model can be trained to automatically generate output data based on input data, which in turn corresponds to an automated determination or prediction of the expected change in intraocular pressure and / or the expected intraocular pressure after treatment. The first model can use one, several, or all of the above-mentioned information as input data, including: - the intraocular pressure before treatment, - a property of the IOL, - the intraocular pressure of the eye to be treated before treatment, - the eye geometry of the eye to be treated, - the patient data of the patient who has the eye to be treated, - information regarding the planned surgical intervention to be combined with the IOL treatment, - the LVC parameter, - the biomechanical corneal parameters of the cornea of ​​the eye to be treated, - the biometric parameters of the eye to be treated, and / or - predetermined parameters dependent on a type of IOL, optionally comprising the A-constant described above.

[0067] It should be noted that the use of an artificial intelligence-based model to predict or determine the expected change in intraocular pressure and / or the expected intraocular pressure after treatment is only one of a multitude of possibilities that can be considered as being in accordance with the disclosure.

[0068] The use of the first model based on artificial intelligence to predict or determine the expected change in intraocular pressure and / or the expected intraocular pressure after treatment has the advantage, among other things, that the model can be changed or adapted automatically, i.e. by means of machine learning, during the development of the software on which the model is based. This advantage then continues to have an impact when the first model can be used in the field. Even then, the first model can be changed; for example, it can be adapted to a user of the model or personalized automatically. In this case, the AI ​​model is not limited to a predefined, fixed set of rules when determining the output data, unlike a comparable deterministic algorithm, but can use a set of rules developed during the learning phase or the development phase to determine the output data.(Expert) knowledge acquired during the training phase, which allows a better representation of complex relationships compared to conventional, deterministic algorithms.

[0069] The method may comprise providing a post-treatment measured change in intraocular pressure and / or a post-treatment measured intraocular pressure of the treated eye, and adjusting the prediction model based on the measured change in intraocular pressure and / or the post-treatment measured intraocular pressure of the treated eye.

[0070] In other words, training of the first model as described above may be provided so that its accuracy can be improved as the period of use progresses.

[0071] A second model based on artificial intelligence can be used to calculate the IOL.

[0072] What was described above with reference to the first model based on artificial intelligence also applies analogously to the second model.

[0073] Therefore, an IOL-calculating AI is proposed. This AI can be trained, for example, by using not only the preoperative parameters described above but also the postoperative outcome parameters, such as refraction and / or visual acuity, each determined subjectively and / or objectively (e.g., determined one to three months after treatment, but possibly even earlier (e.g., one week and / or one day after treatment), since the healing process can provide additional information to support the AI's training).

[0074] The method may include providing a post-treatment measured refraction and / or visual acuity of the treated eye, and adjusting the second model based on the post-treatment measured refraction and / or visual acuity of the treated eye.

[0075] In other words, it is proposed to train the second model based on the refraction and / or visual acuity of the treated eye measured after treatment. It is conceivable that the model could be adjusted in such a way that the refractions and / or visual acuities measured after treatments of the treated eyes increasingly approach the target refractions and / or target visual acuities as the training progresses or as intraocular pressure predictions and IOL calculations improve.

[0076] The method may comprise providing information regarding an actual surgical intervention of the treated eye combined with the IOL treatment, and adapting the second model based on the information regarding the actual surgical intervention of the treated eye combined with the IOL treatment.

[0077] In other words, the adjustment or training can also take into account the actual surgical intervention of the treated eye combined with the IOL treatment.

[0078] During treatment, it is possible that a deviation from the planned surgical intervention in the treated eye, which was combined with the IOL treatment, is carried out. For example, it may have been planned to implant a certain number or type of MIGS, but these are then not implanted and / or not implanted to the planned extent. Another example of this could be if a capsular bag tear occurred during IOL implantation, resulting in the use of a different IOL (with a different haptic structure, such as sulcus-supported IOLs or iris-claw lenses). Such a deviation can influence the overall change in intraocular pressure achieved by the treatment. The actual surgical intervention in the treated eye, combined with the IOL treatment, can therefore be a valuable input for learning or training the model.

[0079] Additionally or alternatively, it is conceivable that information concerning phacoemulsification parameters recorded during the treatment (e.g. duration, phaco energy, aspiration rate, e.g. minimum and / or maximum, or the average) is provided, and that the second model is adapted based on the information concerning the phaco parameters recorded during the treatment.

[0080] Calculating the IOL may include determining a target refractive power of the IOL. Additionally or alternatively, calculating the IOL may include determining a target position and / or a target location of the IOL in the eye to be treated.

[0081] It is also conceivable that the target location and / or position of the IOL in the eye to be treated is assumed to determine the target refractive power of the IOL.

[0082] The IOL can be calculated based on a parameter that characterizes the IOL. This can be the A-constant described above or an equivalent.

[0083] For example, for the IOL calculation of a spherical IOL, only the (target) refractive power of this IOL can be determined or calculated. In this case, the assumed or expected position and orientation can be implicitly taken into account in the ACD and A-constant, which are included in the calculation. The orientation and position of the IOL are then not calculated, but assumed. For example, if a toric IOL is being calculated, the IOL calculation can also specify the rotational position in which the lens should be inserted. If additional IOL positions (ELP) and the IOL orientation (tilt), for example due to custom-made haptics, are degrees of freedom, then it is also conceivable that the IOL calculation could also provide these values. The position of the IOL can, for example, be understood as the position of a predetermined point on the IOL, such as the center point of the central optical lens of the IOL, in space.The position can be specified relative to the position of one or more structures of the eye to be treated. The position of the IOL can, for example, specify the orientation of the IOL in space, for example, relative to the position of at least one structure of the eye to be treated. However, the term "position" can also be understood in a broader sense, so that the position indicates whether the IOL is intended to be an anterior or posterior chamber lens.

[0084] To the extent that the method comprises determining the expected change in intraocular pressure and / or the expected intraocular pressure after treatment based on the property of the IOL, it is conceivable that the method comprises comparing the determined expected change in intraocular pressure and / or the expected intraocular pressure after treatment with a comparison change in intraocular pressure and / or a comparison intraocular pressure after treatment, and optionally outputting a result of the comparison.

[0085] This offers the advantage of making it possible to select IOL variants with regard to a desired pressure reduction effect. For example, if different IOL types are considered for a target refraction, the IOLs can be suggested that, due to their configuration (thickness, haptics, etc.), come closest to the desired pressure reduction.

[0086] In detail, when performing the procedure for several different IOLs, the IOL can be selected with which, for example, the smallest deviation from the target change in intraocular pressure and / or the target intraocular pressure after treatment can be achieved.

[0087] The method can therefore be carried out for several different IOLs, i.e. for several IOLs whose properties (e.g. thickness, haptics, etc., see also above) that are taken into account in the method are different. If the method is carried out for several different IOLs, it is conceivable that by outputting the result of the comparison, at least one of the different IOLs is suggested which has the smallest deviation from the target change in intraocular pressure to the expected change in intraocular pressure or from the target intraocular pressure after treatment to the expected intraocular pressure after treatment. Other output forms are also conceivable; for example, the output can also be in list form, e.g. sorted in ascending or descending order according to the amount of deviation. The output can generally be auditory and / or visual.The method may include controlling a human-machine interface to output the result of the comparison. The same also applies to outputting the result of the IOL calculation.

[0088] Furthermore, a computer-implemented method for training an artificial intelligence-based model can be provided. The model is configured to determine or predict a result of a calculation of an IOL. This is done based on a biometric parameter of an eye to be treated with an IOL, and an expected change in intraocular pressure due to treatment of the eye to be treated with the IOL compared to an intraocular pressure of the eye to be treated before treatment of the eye to be treated with the IOL, an expected intraocular pressure after treatment of the eye to be treated with the IOL, and / or an intraocular pressure before treatment of the eye to be treated with the IOL.The method comprises providing a post-treatment measured refraction and / or visual acuity of the treated eye, and adapting the model based on the post-treatment measured refraction and / or visual acuity of the treated eye, optionally taking into account a predefined target refraction and / or target visual acuity.

[0089] Refraction is the refractive power of the necessary optical correction (e.g., glasses or contact lenses) with which an eye, without accommodation, produces a sharp image of an object at infinite distance. A normal-sighted eye (emmetropic eye) therefore has a refraction of 0 dpt. When determining refraction, a distinction must be made between objective and subjective refraction. In objective refraction, the refractive properties of the eye are determined using a measuring device (e.g., an autorefractor), while in subjective refraction, information from the person being examined about the perceived image sharpness when holding up various test lenses (e.g., a phoropter or trial glasses), which can then take into account the influence of other variables on perception, such as the color or contrast of the test image.Both objective and subjective refractions can be performed under different illumination conditions that affect pupil diameter, such as photopic, mesopic, and scotopic conditions.

[0090] When determining visual acuity, however, for a specified optical correction, for example, without correction or one that corresponds to the target refraction, it is tested whether the patient can normally recognize and name optotypes of different sizes, such as letters or the position of Landolt rings, using their eye(s). In this case, the test does not determine the patient's exact refraction, but rather whether normal vision is achieved with the selected optical correction, such as the target refraction, or what visual acuity is achieved compared to a person with normal vision.Such a relative assessment of visual acuity is also referred to as a visual acuity determination and is usually expressed decimal (e.g. 1.0 for normal vision) or as a fraction, where the denominator indicates the distance at which a person with normal vision should have seen a test object and the numerator indicates the distance at which the person examined actually saw the test object, e.g. 6m / 12m = 6 / 12 = visual acuity 0.5. It is conceivable to determine the absolute visual acuity of an eye for a specific optical correction using the image structure widths that it can resolve and then to express this, for example, in visual field angles. A person with normal vision can resolve structures of 1', while a person with a visual acuity of 0.5 can only resolve 2'. Since, in contrast to refraction, with a single visual acuity determination it is generally not clear whether a sub-optimal visual acuity compared to the target correction represents an over- or under-correction, this information could possiblybut can be obtained by carrying out at least one further visual acuity test during a second optical correction in order to be able to determine the course of the visual acuity depending on the optical correction.

[0091] The procedure described above is therefore a training procedure for the second artificial intelligence-based model described above.

[0092] The training or teaching of the model can involve the machine learning described above. Supervised learning can also be used. Supervised learning is a subfield of machine learning. Learning can refer to the ability of an artificial intelligence or the model to replicate regularities. The results to be predicted by the model are known from the training data and are used to train the model. A learning algorithm attempts to find a hypothesis that makes the most accurate predictions possible. A hypothesis is a mapping that assigns each input value to the presumed output value. The method is therefore based on a predetermined output to be learned, the results of which are known.The predicted results of the model being improved by the learning process can be compared with the actual results known, for example, from measurements, i.e., "monitored." The predicted result can be the target refraction described above, which can, for example, be set to 0 D by default. The model is therefore trained or adapted in such a way that the IOLs calculated by the model for the target refractions and implanted by the surgeon result in actual and thus post-operatively measurable or measured refractions of the eyes as actual results, which increasingly approach the target refractions as predicted results.

[0093] It is conceivable that the training method includes an initial training of the AI ​​model. For the initial training, a (training) dataset can be used that includes the at least one biometric parameter and a corresponding IOL calculation result. The latter can be obtained, for example, using a conventional IOL calculation formula (see above). After the initial training, an AI model (for example, referred to as a pre-trained AI model) can be obtained. This pre-trained AI model can be configured to mimic a result of an IOL calculation of the conventional formula used to generate the training dataset.

[0094] This pre-trained AI model can be further adapted or fine-tuned in a second or subsequent training step, i.e., the adaptation of the AI ​​model described above. In this further or subsequent training step, the actually achieved and postoperatively measured refraction and / or visual acuity can be used as a feedback variable, and the intraocular pressure (in the form described above) can be provided to the model for calculating the IOL as further input. As described above, a target refraction can be set for training in the subsequent training step. The target refraction can be 0. The goal of the subsequent training step can be to adapt the AI ​​model so that the result of the IOL calculation changes such that the actual postoperatively measured refraction approaches the target refraction as the training period progresses. The same applies analogously to a target visual acuity.

[0095] What has been described above with reference to the method for calculating the IOL also applies analogously to the previously described method for training the (second) model based on artificial intelligence and vice versa.

[0096] Furthermore, a (further) computer-implemented method for training an artificial intelligence-based model can be provided. The model is configured to determine or predict an expected change in intraocular pressure due to treatment of a treated eye with an IOL compared to the intraocular pressure of the treated eye before treatment of the treated eye with the IOL, and / or an expected intraocular pressure after treatment of the treated eye with the IOL.This is done based on an intraocular pressure before treatment of the eye to be treated with the IOL, a property of the IOL, an eye geometry of the eye to be treated, patient data of a patient having the eye to be treated, information regarding a planned surgical intervention of the eye to be treated to be combined with the IOL treatment, an LVC parameter of the eye to be treated, and / or biomechanical corneal parameters of a cornea of ​​the eye to be treated. The method comprises providing a measured change in the intraocular pressure and / or an intraocular pressure measured after the treatment of the treated eye, and adapting the model based on the measured change in the intraocular pressure and / or the intraocular pressure measured after the treatment of the treated eye.

[0097] The procedure is therefore a training procedure for the first artificial intelligence-based model described above.

[0098] What was described above with regard to training the second model also applies analogously to training the first model. As described above, a learning algorithm attempts to find a hypothesis that makes the most accurate predictions possible. The method is therefore based on a predetermined output to be learned, the results of which are known. In this case, the results can be understood as the postoperative intraocular pressure to be predicted by the first model and / or the expected change in intraocular pressure to be predicted by the first model.

[0099] The above descriptions regarding the method for calculating the IOL and the above descriptions regarding the method for training the second model also apply analogously to the previously described method for training the (first) artificial intelligence-based model, and vice versa. The two training methods can be partially or completely combined.

[0100] A data processing device comprising means for carrying out one, several or all of the methods described above may be provided.

[0101] The data processing device or the computer can be a local processing unit. Additionally or alternatively, it is conceivable that parts of the data processing device are arranged remotely and in turn are connected to the local processing unit wirelessly and / or wired. This means, for example, that a computer program according to the disclosure is provided at least partially in or via a cloud, e.g., as Software as a Service (SaaS). It is conceivable that in such a case, the input data required for the computer program are output from the local computer to a server, e.g., via the Internet. The computer program can then be executed in the cloud or by the respective cloud component, which, as part of the server, is also part of the data processing devices. In this case, the output data of the AI ​​computer program determined on the server, i.e.For example, the result of an IOL calculation can be output to the local computer, again via the Internet. The server and thus the computer program can be operated, maintained, and further developed by a software manufacturer, possibly based on input data from multiple local computers and thus users.

[0102] What has been described above with reference to the methods also applies analogously to the data processing device and vice versa.

[0103] A computer program comprising instructions which, when executed by a computer, cause the computer to perform one, several or all of the methods described above may be provided.

[0104] A computer program can be understood as a sequence of instructions that complies with the rules of a specific programming language in order to process or solve certain functions, tasks or problems, including, in this case, the calculation of an IOL, with the help of a computer.

[0105] A computer program is part of a computer's software. It can be present (e.g., on a data storage medium) as a computer-executable program file, often in so-called machine code, which can be loaded into the computer's memory for execution. The computer program is processed as a sequence of machine or processor instructions by the computer's processor(s) and thus executed.

[0106] A computer-readable medium comprising instructions that, when executed by a computer, cause the computer to perform one, some, or all of the methods described above may be provided.

[0107] Below, an optional embodiment of the disclosure is described with reference to Fig. 1 to 5 described. Fig. 1 shows a data processing device according to the disclosure, which is designed to carry out one of the computer-implemented methods according to the disclosure, Fig. 2 shows a flowchart of a computer-implemented method for calculating an IOL according to the disclosure, Fig. Figure 3 shows a step of calculating the IOL of the disclosed method in detail, Fig. 4 shows a flow diagram of a computer-implemented method according to the disclosure for training a second model based on artificial intelligence for calculating an IOL, and Fig. 5 shows a flow diagram of a computer-implemented method according to the disclosure for training a first model based on artificial intelligence for determining a change in an intraocular pressure and / or an expected intraocular pressure after treatment of an eye with an IOL.

[0108] The Fig. The data processing device 1, shown only schematically in FIG. 1, comprises a data carrier or a computer-readable medium 2, a main memory 3, and a processor 4. The computer-implemented methods described in detail below are each stored as a computer program in the form of processor instructions on the computer-readable medium 2. To execute the computer program or instructions, the respective computer program is loaded into the main memory 3 and is processed and thus executed by the processor 4 as a sequence of processor instructions.

[0109] In the following, a computer-implemented method 100 for calculating an intraocular lens (IOL) according to the disclosure is described in detail with reference to Fig. 2, which schematically illustrates a flow chart of the method 100.

[0110] In a first step 101 of the method, information regarding an intraocular pressure and biometric parameters of an eye to be treated, optionally including a chamber angle of the eye to be treated, and optionally further information are provided to an artificial intelligence (AI) model 10. In addition to or as an alternative to the AI ​​model 10, a conventional deterministic algorithm can also be used, which, for example, uses a calculation formula described above, but which additionally takes into account information regarding the intraocular pressure. The following explanations regarding the AI ​​model 10 therefore apply analogously to such an algorithm.

[0111] The information regarding intraocular pressure may include an expected change in intraocular pressure due to treatment of the treated eye with the IOL compared to an intraocular pressure of the treated eye before treatment of the treated eye with the IOL, an expected intraocular pressure after treatment of the treated eye with the IOL, and / or an intraocular pressure before treatment of the treated eye with the IOL.

[0112] The optionally provided further information may include information concerning a property of the IOL, an eye geometry of the eye to be treated, patient data of a patient having the eye to be treated, information concerning a planned surgical intervention of the eye to be treated to be combined with the IOL treatment, an LVC parameter of the eye to be treated, and / or biomechanical corneal parameters of a cornea of ​​the eye to be treated.

[0113] The eye geometry may include information regarding a chamber angle geometry and / or anterior chamber geometry and / or posterior chamber geometry and / or retinal geometry of the eye to be treated.

[0114] The information regarding the planned surgical intervention to be combined with the IOL treatment of the treated eye may include information on the number and / or type of MIGS and / or TM aspiration (see below for further details on the planned combined surgical intervention).

[0115] The biometric parameter may include an axial length, a corneal diameter, a corneal refractive power, a corneal thickness, an anterior chamber depth, and / or a lens thickness of the eye to be treated. The biometric parameters may also include information about the patient, such as age, gender, ethnicity, and the number and type of prior treatments.

[0116] The further information may comprise a parameter characterizing the IOL, for example information concerning a dimension and / or a shape and / or a refractive power of the IOL, so that the calculation of the IOL described below can optionally also be carried out based on this parameter characterizing the IOL.

[0117] In a second step 102 of the method 100, the information provided in the first step 101 of the method (i.e., the information concerning the intraocular pressure, the biometric parameters, and the information concerning the surgical intervention) is input into the AI ​​model 10.

[0118] In a third step 103 of the method 100, the AI ​​model 10 calculates the IOL based on the information entered into the AI ​​model 10 in the second step 102. A result of the calculation includes a target refractive power of the IOL. To calculate the IOL, a target position and / or a target location of the IOL in the eye to be treated can be assumed. Alternatively, it is conceivable that calculating the IOL includes determining the target position and / or location of the IOL in the eye to be treated in addition to determining the target refractive power of the IOL.

[0119] Steps 101–103 can be performed for IOLs with different properties. The results described above can thus be obtained for each of the calculated IOLs.

[0120] In an optional fourth step 104 of the method 100, the result of the calculation (see step 103), optionally together with the expected change in intraocular pressure used for this purpose and / or the expected intraocular pressure after treatment of the eye to be treated with the IOL, is output to a surgical planning system 5. This can be done for each of the calculated IOLs. The surgical planning system 5 can be configured to output guidance information (such as a rotational orientation of a toric IOL to be realized), notes and / or warnings (such as regarding inconsistencies, for example, between the two eyes, or unusual parameters that may need to be verified) to a user of the surgical planning system 5, in particular a physician, during treatment of the eye with the calculated IOL.

[0121] In an optional fifth step 105 of the method, a combination of parameters that are required and / or available for surgical planning and a determination of options that can be selected by the physician (for example, selection from options of IOL types that can be used) and the output of surgical parameters to be implemented by the physician (such as the position of an incision to be performed by the physician (limbal relaxing incision) or the alignment of the IOL to be achieved in the eye for the treatment of astigmatism) can be carried out by means of the surgical planning system 5. It is conceivable that in the optional fifth step 105, in a first sub-step 1051 thereof, a comparison of the expected change in the intraocular pressure obtained in the fourth step 104 and / or the expected intraocular pressure after the treatment with a comparison change (iea desired change) in the intraocular pressure and / or a comparison intraocular pressure (i.e., a desired intraocular pressure) after the treatment. The comparison change can be an (optionally predetermined) target change in the intraocular pressure. The comparison intraocular pressure can be an (optionally predetermined) target intraocular pressure after the treatment. The comparison change or the comparison intraocular pressure can be a change in the intraocular pressure expected for a hypothetical application of a different IOL (i.e., a different IOL variant or a different IOL type) and / or the intraocular pressure expected for this after the treatment. This can be done for each of the calculated IOLs. In a second sub-step 1052 of the optional fifth step 105, a result of the comparison can be output, or in the case of multiple calculated IOLs, the comparisons can be output.It is conceivable that the physician be suggested one of the IOLs with which the smallest deviation from the target change in intraocular pressure and / or the target intraocular pressure (i.e., a desired intraocular pressure) can be achieved after treatment. It is conceivable that this could be done, for example, by means of a visual display on the surgical planning system 5.

[0122] The above-described third step 103 of the method 100 will be described in detail below with reference to Fig. 3, which shows a flowchart of this step 103.

[0123] As can be seen from Fig. 3, the third step 103 of the method can be divided into three sub-steps 1031 - 1033 in one possible embodiment.

[0124] In a first of the three sub-steps 1031, the expected change in intraocular pressure and / or the expected intraocular pressure after treatment is determined using a first AI model referred to as prediction model 11. As already described above, a deterministic algorithm can also be used for this purpose, so the following description of the third sub-step 1033 also applies analogously to a deterministic algorithm.

[0125] The prediction model 11 receives as input the biometric parameters described above with reference to the second step 102 of the method 100 as well as the intraocular pressure before the treatment of the eye with the IOL and optionally the further information described for the first step 101 of the method (see above).

[0126] Based on the input, the prediction model 11 determines the expected change in intraocular pressure and / or the expected intraocular pressure after treatment, which is also the output of the prediction model 11.

[0127] In a second sub-step 1032 of the third step 103 of the method, the biometric parameters of the eye to be treated provided in the first step 101 of the method 100 are adapted based on the output of the prediction model 11.

[0128] In a third of the three sub-steps 1033, the actual calculation of the IOL is performed using a second AI model referred to as calculation model 12. As already described above, a deterministic algorithm can also be used for this, so the following description of the third sub-step 1033 also applies analogously to a deterministic algorithm.

[0129] The calculation model 12 receives as input the adapted biometric parameters described above with reference to the second sub-step 1032 and optionally the further information described above with reference to the first step 101 of the method 100.

[0130] The calculation model 12 calculates the IOL based on the input, wherein the calculation, as described above with reference to the third step 103 of the method, comprises determining the target refractive power, the target position and / or the target location of the IOL in the eye (see above for details), which in turn represents the output of the calculation model 12.

[0131] As described above with reference to the fourth step 104 of the method 100, the output of the calculation model 12 can be fed into the operation planning system 5 (see Fig. 2) must be entered.

[0132] Below, a training procedure 200 for the AI ​​model 10 with reference to Fig. 4, which, after training, is designed to perform the IOL calculation. This can be the calculation model 12.

[0133] In Fig. 4 schematically shows a flow chart of the training method 200 for the calculation model 12.

[0134] In a first step 201 of this training method 200, a refraction and / or visual acuity of the treated eye measured after the treatment is provided. It is conceivable that the first step 201 of the training method 200 also includes providing information regarding an actual surgical intervention of the treated eye combined with the IOL treatment.

[0135] In a second step 202 of the training method 200, the calculation model 12 is adapted based on the refraction and / or visual acuity of the treated eye measured after the treatment (actually achieved). It is conceivable that the calculation model 12 is also adapted based on the information regarding the actual surgical intervention of the treated eye combined with the IOL treatment.

[0136] The training method 200 can be completed as soon as the AI ​​model 10, optionally the calculation model 12, can perform the calculation of the IOL with a desired quality or accuracy for a data set unknown to the AI ​​model 10, ie a refraction and / or visual acuity of the treated eye achieved with an IOL calculated with the calculation model 12 lies within a predetermined tolerance range, which can correspond, for example, to daily fluctuations in refractions (for example ±0.25 dpt) or visual acuities.

[0137] The AI ​​model 10, optionally the calculation model 12, is consequently designed after training, i.e. after completion of the second step 202, to calculate a further IOL unknown to the AI ​​model 10 based on a further (optionally adjusted) biometric parameter of an eye to be treated with an IOL (and optionally a further expected change in an intraocular pressure due to a treatment of the further eye to be treated with the further IOL compared to an intraocular pressure of the eye to be treated before the treatment of the further eye to be treated with the further IOL, a further expected intraocular pressure after the treatment of the further eye to be treated with the further IOL and / or a further intraocular pressure before the treatment of the eye to be treated with the further IOL) with a predetermined quality.

[0138] In the following, a further, second training procedure 300 for the AI ​​model 10 with reference to Fig. 5. This may, in particular, be the prediction model 11. The basic principle of the second training method 300 is the same as that of the first training method 200, so the above statements apply analogously.

[0139] The second training method 300 comprises, in a first step 301, providing a measured change in the intraocular pressure and / or an intraocular pressure of the treated eye measured after the treatment.

[0140] The second training method 300 comprises, in a second step 302, adapting the prediction model 11 based on the measured change in the intraocular pressure and / or the intraocular pressure of the treated eye measured after the treatment.

[0141] The training method 300 can be completed as soon as the AI ​​model 10, optionally the prediction model 11, can calculate the expected intraocular pressure after the treatment and / or the expected change in the intraocular pressure with a predetermined quality or accuracy for a data set unknown to the AI ​​model 10, ie as soon as a deviation of the predicted intraocular pressure after the treatment or the predicted change in the intraocular pressure, which was determined with the prediction model 11, from an actually measured intraocular pressure after the treatment or an actually measured change in the intraocular pressure is smaller than a predetermined limit value, for example the measurement error of the intraocular pressure measurement method used or smaller than the mean daily fluctuations in the intraocular pressure.

[0142] It is further noted that the above with reference to Fig. 2 described method 100 can be combined with the first and second training methods 200, 300. Thus, with reference to Fig.2, the method 100 described above comprises steps 301, 302 of the second training method 300. A sixth step 106 of the first method 100, which corresponds to the first step 301 of the second training method 300, then involves providing a measured change in the intraocular pressure and / or an intraocular pressure of the treated eye measured after the treatment. A seventh step 107 of the first method 100, which corresponds to the second step 302 of the second training method 300, then adapts the AI ​​model 10, optionally the prediction model 11, based on a difference ora difference between the measured change in intraocular pressure and / or the intraocular pressure of the treated eye measured after treatment and the expected change in intraocular pressure and / or the determined expected intraocular pressure of the eye to be treated determined by the AI ​​model 10, optionally the prediction model 11, in the third step 103 (optionally the first sub-step 1031 thereof). Furthermore, in an eighth step 108 of the method 100, which corresponds to the first step 201 of the first training method 200, a refraction and / or visual acuity of the treated eye measured after treatment can be provided.In a ninth step 109 of the method 100, which corresponds to the second step 202 of the first training method 200, the AI ​​model 10, optionally the calculation model 12, can be adapted based on a comparison of the (actually achieved) refraction and / or visual acuity of the treated eye measured after the treatment and a predetermined or specified target refraction.

[0143] Finally, it should be noted that although the AI ​​model 10 used in the third step 103 can have the two submodels (i.e., the calculation model 12 and the prediction model 11) as described above, other implementation options for the AI ​​model 10 are also conceivable. For example, a single (end-to-end) AI model is conceivable, which generates the output of the calculation model 12 (see fourth step 104 of the method 100) based on the input of the prediction model 11 (see second step 102 of the method 100). The above statements regarding training the calculation model 12 and the prediction model 11 apply accordingly or analogously to such an end-to-end model. List of reference symbols 1 Data processing device 2 computer-readable medium 3 RAM 4 processors 5 Operation planning system 10 AI model 11 Prediction model 12 Calculation model 100 methods for calculating an IOL 101 - 109 Steps of the IOL calculation procedure 200 First training procedure for an AI model 201-202 Steps of the first training procedure 300 Second training procedure for an AI model 301-302 Steps of the second training procedure QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] US 7101335 B2

[0029] Cited non-patent literature

[0000] C. Leydolt, O Findl and W Drexler („Effects of change in intraocular pressure on axial eye length and lens position“, Eye (2008) 22, 657-661; doi:10.1038 / sj.eye.6702709

[0038] KJ Maverick, Effect of Intraocular Pressure on Keratometry in Porcine Eyes, ARVO 2004

[0039] D. Melancia, „Cataract Surgery and Intraocular Pressure“, Ophthalmic Res 2015;53:141-148, DOI:10.1159 / 000377635 [0041, 0052] Lin et al. „PREDICTORS OF INTRAOCULAR PRESSURE AFTER PHACOEMULSIFICATION IN PRIMARY OPEN-ANGLE GLAUCOMA EYES WITH WIDE VERSUS NARROWER ANGLES“, Trans Am Ophthalmol Soc 2017;115:T6[1-13

[0052]

Claims

[1] A computer-implemented method (100) for calculating an intraocular lens (IOL), comprising: - Calculation (103) of the IOL based on a biometric parameter of an eye to be treated with the IOL, characterized by that the IOL is / are taken into account in the calculation (103): - an expected change in intraocular pressure due to treatment of the treated eye with the IOL compared to the intraocular pressure of the treated eye before treatment of the treated eye with the IOL, - an expected intraocular pressure after treatment of the eye to be treated with the IOL, and / or - an intraocular pressure before treatment of the eye to be treated with the IOL. [2] Computer-implemented method (100) according to claim 1, characterized bythat the method comprises determining (1031) the expected change in intraocular pressure and / or the expected intraocular pressure after treatment based on: - the intraocular pressure before treatment, - a property of the IOL, - an eye geometry of the eye to be treated, - Patient data of a patient who has the eye to be treated, - information regarding a planned surgical intervention in the eye to be treated, which is to be combined with the IOL treatment, - an LVC parameter of the eye to be treated, and / or - biomechanical corneal parameters of a cornea of ​​the eye to be treated. [3] Computer-implemented method (100) according to claim 2, characterized bythat a first model (11) based on artificial intelligence is used to determine (1031) the expected change in intraocular pressure and / or the expected intraocular pressure after treatment. [4] Computer-implemented (100) method according to claim 3, characterized by that the method (100) comprises: - providing (106) a change in intraocular pressure measured after the treatment and / or an intraocular pressure of the treated eye measured after the treatment, and - adjusting (107) the first model (11) based on the measured change in intraocular pressure and / or the intraocular pressure of the treated eye measured after treatment. [5] Computer-implemented method (100) according to one of claims 2 to 4, characterized bythat the eye geometry includes information concerning a chamber angle geometry and / or anterior chamber geometry and / or posterior chamber geometry and / or retinal geometry of the eye to be treated. [6] Computer-implemented method (100) according to one of claims 2 to 5, characterized by that the information concerning the planned surgical intervention of the eye to be treated, to be combined with the IOL treatment, includes information on a number, a position and / or a type of implants and / or trabeculotomies and / or the extent of trabeculoplasty or TM aspiration and / or trabeculectomy and / or cyclodialysis and / or cyclocoagulation. [7] Computer-implemented method (100) according to one of claims 1 to 6, characterized by , that: - the biometric parameter includes an axial length, a corneal diameter, a corneal refractive power, a corneal thickness, an anterior chamber depth and / or a lens thickness of the eye to be treated, - the patient data include information concerning the number and / or type of previous treatment of the patient and / or concerning existing and / or planned medication and / or (eye) implants of the patient and / or blood pressure of the patient, and / or - the patient data includes information regarding the patient's age, gender and ethnicity. [8] Computer-implemented method (100) according to one of claims 1 to 7, characterized by that a second model based on artificial intelligence (12) is used to calculate the IOL. [9] Computer-implemented method (100) according to claim 8, characterized by that the method (100) comprises: - providing (108) a refraction and / or visual acuity of the treated eye measured after the treatment, and - Adjusting (109) the second model (12) based on the refraction and / or visual acuity of the treated eye measured after the treatment. [10] Computer-implemented method (100) according to claim 9, as far as dependent on claim 2, characterized by that the method (100) comprises: -providing (201) information concerning an actual surgical intervention of the treated eye combined with the IOL treatment, and - adapting (202) the second model (12) based on the information concerning the actual surgical intervention of the treated eye combined with the IOL treatment. [11] Computer-implemented method (100) according to one of claims 1 to 10, characterized bythat calculating (103) the IOL comprises determining a target refractive power of the IOL, optionally based on an assumed target position and / or a target location of the IOL in the eye to be treated, and / or determining the target position and / or the target location of the IOL in the eye to be treated. [12] Computer-implemented method (100) according to one of claims 1 to 11, characterized by that the calculation (103) of the IOL is based on a parameter characterizing the IOL. [13] Computer-implemented method (100) according to one of claims 2 to 12, insofar as the method comprises determining (1031) the expected change in intraocular pressure and / or the expected intraocular pressure after the treatment based on the property of the IOL, characterized by that the procedure includes: - comparing (1051) the determined expected change in intraocular pressure and / or the expected intraocular pressure after treatment with a comparison change in intraocular pressure and / or a comparison intraocular pressure after treatment, and - optionally output (1052) a result of the comparison. [14] Computer-implemented method (200) for training an artificial intelligence-based model (10, 12) designed to obtain a result of a calculation of an IOL based on: - a biometric parameter of an eye to be treated with an IOL, and - to determine an expected change in intraocular pressure due to treatment of the treated eye with the IOL compared to an intraocular pressure of the treated eye before treatment of the treated eye with the IOL, an expected intraocular pressure after treatment of the treated eye with the IOL and / or an intraocular pressure before treatment of the treated eye with the IOL, characterized by that the procedure includes: - providing (201) a refraction and / or visual acuity of the treated eye measured after the treatment, and - Adjusting (202) the model (10, 12) based on the refraction and / or visual acuity of the treated eye measured after treatment. [15] Computer-implemented method (300) for training an artificial intelligence-based model (10, 11) designed to: - an expected change in intraocular pressure due to treatment of a treated eye with an IOL compared to the intraocular pressure of the treated eye before treatment of the treated eye with the IOL, and / or - an expected intraocular pressure after treatment of the eye to be treated with the IOL based on: - intraocular pressure before treatment of the eye to be treated with the IOL, - a property of the IOL, - an eye geometry of the eye to be treated, - Patient data of a patient who has the eye to be treated, - information regarding a planned surgical intervention in the eye to be treated, which is to be combined with the IOL treatment, - an LVC parameter of the eye to be treated, and / or - to predict biomechanical corneal parameters of a cornea of ​​the eye to be treated, characterized by that the procedure includes: - providing (301) a measured change in the intraocular pressure and / or an intraocular pressure of the treated eye measured after the treatment, and - adapting (302) the model (1, 11) based on the measured change in intraocular pressure and / or the intraocular pressure of the treated eye measured after treatment. [16] Data processing device (1), comprising means (2 - 4) for carrying out the method (100) according to one of claims 1 to 13, the method (200) according to claim 14 and / or the method (300) according to claim 15. [17] A computer program comprising instructions which, when executed by a computer, cause the computer to carry out the method (100) according to any one of claims 1 to 13, the method (200) according to claim 14 and / or the method (300) according to claim 15. [18] A computer-readable medium comprising instructions which, when executed by a computer, cause the computer to perform the method (100) according to any one of claims 1 to 13, the method (200) according to claim 14 and / or the method (300) according to claim 15.

Images