Improved Calculation of Ophthalmic Lenses
The surface model-based method addresses the computational challenges in ophthalmic lens calculation by optimizing model parameters, resulting in faster and more efficient surface determination with lower resource usage.
Patent Information
- Application Number
- JP2023508061
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Priority Date
- 2020-08-07
- Filing Date
- 2021-08-04
- Publication Date
- 2024-12-09
- Estimated Expiration
- 2041-08-04
AI Technical Summary
Existing methods for calculating ophthalmic lenses face challenges with increasing computational costs and storage requirements as the number of order parameters increases, leading to longer computation times and reduced quality in optimization.
A computer-implemented method using a surface model to directly calculate ophthalmic lens surfaces from order parameters, reducing the need for iterations and minimizing storage requirements by employing a training data set to optimize model parameters for faster computation.
The method significantly reduces computational cost and time, allowing for efficient calculation of ophthalmic lens surfaces with improved computational efficiency and reduced resource consumption.
Smart Images

Figure 0007596511000029 
Figure 0007596511000030 
Figure 0007596511000031
Abstract
Description
[Technical field]
[0001] The invention relates to a method for determining a surface model, a method for determining at least one surface of at least one ophthalmic lens with the aid of a surface model and a corresponding manufacturing method. The invention also relates to a computer program product and a device. [Background technology]
[0002] In calculating ophthalmic lenses, such as spectacle lenses, one objective is to calculate the shape of the surfaces of an ophthalmic lens or pair of lenses and their location (i.e. orientation and position) relative to each other, so that they have defined geometrical properties (e.g. a predefined thickness at a defined point of the lens) and optical properties adapted to the subsequent wearing situation (e.g. fit in optical and possibly also physiological terms to the eye or eyes looking through the ophthalmic lens).
[0003] Examples of ophthalmic lenses are contact lenses and spectacle lenses, such as single vision lenses and contact lenses, multifocal spectacle lenses and multifocal contact lenses, and spectacle lenses with variable refractive power (eg variable focus lenses).
[0004] Given a conventional spectacle lens, the fitting scheme is largely limited to the refractive error of the eye, the refractive index of the material used in the lens, and the size and / or shape of the edge of the spectacle lens. The refractive errors may include sphere, cylinder and axis, possibly addition or near refraction, and / or prism prescription.
[0005] Assuming what is known as an individualized spectacle lens, individual parameters such as the orientation and distance of the spectacle lens from the viewing eye (given by the centration parameters), the distance or location of the centers of rotation of the eye relative to the lens and / or each other, the individual object distances at defined observation points (e.g. reference points) on the spectacle lens, and the individual positions of these observation points on the glass are added as additional parameters.
[0006] Given a personalized spectacle lens, additional parameters may be added, for example viewing situations specifically provided for the spectacle lens, viewing behavior (e.g. the interaction of head and eye deflections), biometric parameters describing the eye (e.g. different viewing directions, eye length, curvature and position of the refractive surfaces of the eye, wavefront errors including higher orders assuming the refractive index of the medium (e.g. as a set of Zernike coefficients), pupil size and / or position), or other parameters specific to the provided wearer.
[0007] TIFF0007596511000001.tif30169
[0008] Examples of order parameters can be found in standards established for spectacle lenses (see, for example, EU guideline 93 / 42 / EWG on medicinal products).
[0009] For this reason, given the calculation of an ophthalmic lens from a set of order parameters as a whole, there is an objective to calculate a set of surface parameters from which an ophthalmic lens or lens pair may be manufactured.
[0010] Given conventional spectacle lenses, this can be done by direct calculation of the surface parameters (e.g. by superimposing a determined surface determined for one product with another surface in order to adjust the prescription at the reference or measurement points). Given individualized or personalized spectacle lenses, optimization is often used, which requires the shape and location of a starting surface as well as one or more objective functions to be optimized.
[0011] Such a starting surface may be held constant for multiple order parameters, or multiple starting surfaces belonging to different order parameter sets may be interpolated and / or extrapolated.
[0012] Examples of optimization methods using objective functions are described, for example, in EP 1 091 233, DE 10 2012 000 390, EP 2 384 479, EP 2 177 943.
[0013] Exemplary methods for direct calculation are described in EP 0 654 692 A1 or US 4 514 061 A. An example for calculation by interpolation is described in EP 2 449 420.
[0014] The disadvantage of traditional calculation methods for ophthalmic lenses is that with an increasing number of order parameters, either a direct calculation (e.g. by superposition) is no longer possible or is associated with a loss of quality, or the calculation times of the optimization passes become increasingly longer, since the objective functions to be evaluated in the optimization become more complex.
[0015] In fact, there are methods for directly calculating the surface of an ophthalmic lens from a set of several parameters with low computational cost, for example the methods described in EP 0 654 692 A1 or US 4 514 061 A (Winthrop), but these methods cannot be used to calculate currently common ophthalmic lenses, since with these methods it is only possible to calculate optimally determined surfaces using criteria that are no longer up-to-date (for example the optimization of multifocal lenses using surface properties but not use position properties).
[0016] Calculations of ophthalmic lenses, in particular the computationally intensive optimization of individualized or personalized spectacle lenses, are usually performed anew each time, even if the ophthalmic lenses have identical or very similar order parameters and therefore the surfaces of the ophthalmic lenses so calculated are identical or similar. Such calculations are usually performed in the manufacture of ophthalmic lenses (in the creation of the design of an ophthalmic lens, to determine the surfaces to be manufactured or to check whether it is possible to manufacture an ophthalmic lens with the geometric properties of the surfaces). They are also used during consultations (e.g. at an optician) in order to clarify the optical and geometric properties of the individualized or personalized ophthalmic lenses (e.g. the location of the clear vision zone and the thickness assuming a multifocal lens) of the future wearer of such lenses. In order to keep waiting times short, in this particular application the calculations have to be completed in a short time.
[0017] In EP 2 449 420 a method is disclosed with the aid of which a fast calculation of the surface of a spectacle lens can be performed by interpolation of a surface already optimized in the order parameter space. However, in this case the surface variations depending on the order parameters must be calculated in advance, which leads to high storage requirements or else must be repeated in every calculation, which leads to longer calculation times. Summary of the Invention [Problem to be solved by the invention]
[0018] The object of the present invention is to reduce the computational costs required for the calculation of ophthalmic lenses and at the same time reduce the storage requirements, which allows for faster calculation of ophthalmic lenses, requiring less computing power and therefore more advantageous. [Means for solving the problem]
[0019] This object is achieved by a computer-implemented method, a corresponding device and a corresponding computer program product for manufacturing a surface model, a computer-implemented method, a corresponding device and a corresponding computer program product for determining at least one surface of at least one ophthalmic lens, and a method and a corresponding device for manufacturing an ophthalmic lens having the features set out in the respective independent claims.
[0020] According to a first aspect, there is provided a computer-implemented method for determining a surface model for calculating at least one surface of at least one ophthalmic lens (e.g., a contact lens or a spectacle lens) from a set of order parameters for the at least one ophthalmic lens and / or from variables dependent on the order parameters (e.g., variables derived from the order parameters, etc.).
[0021] "Calculation of at least one surface of an ophthalmic lens" in the sense of the present application includes the calculation of at least a portion of the surface or a part of the surface. Expressed differently, what is understood by "calculation of at least one surface of an ophthalmic lens" is the calculation of at least a portion of the surface or the calculation of the entire surface.
[0022] The at least one ophthalmic lens may be an individual lens. Similarly, it is possible to calculate one or both of a pair of ophthalmic lenses. For example, at least one pair of ophthalmic lenses (lens pair) including lenses designated for a right eye and a left eye of a person may be calculated using the surface model. In this case, the order values of both lenses of the lens pair (e.g., left and right spectacle lenses of a spectacle lens pair, as well as binocular order data, etc.) may be included in a set of order parameters (order parameter set).
[0023] At least one surface calculated according to the surface model may be parametrically described by at least one parameter. In this case, the calculation of the surface with the aid of the surface model comprises the calculation of at least one parameter (surface parameter) of the surface from the order parameters or from variables (auxiliary variables) that depend on the order parameters (e.g. variables derived from the order parameters, etc.).
[0024] The surface may be described, for example, by the curvature or primary curvature at at least one point, for example a reference point of the ophthalmic lens, as well as by the surface normal and, if applicable, the orientation of the principal cross sections. It is also possible to describe the surface by a local representation, for example a spline or polynomial representation with corresponding coefficients, or a non-local representation, for example a Zernike decomposition with corresponding coefficients.
[0025] It is equally possible to provide the surface directly, for example by a set of elevations at multiple raster points. In this case, the calculation of the surface with the aid of the surface model comprises the calculation of the elevations of the surface at multiple raster points from the order parameters or from variables (auxiliary variables) that depend on the order parameters (for example are derived from the order parameters).
[0026] In one example, one of the surfaces of the ophthalmic lens and / or the positioning of this surface relative to another surface of the ophthalmic lens is calculated from a set of order parameters with the aid of a surface model. The other surface may be a predefined surface, e.g., a spherical surface with a predefined curvature, that may depend on the order parameters (e.g., given a known base curve system). In another example, both surfaces of the ophthalmic lens and / or the positioning (i.e., orientation and / or position) of these surfaces relative to each other are calculated from a set of order parameters with the aid of a surface model.
[0027] The calculation of at least one surface from the set of order parameters of the ophthalmic lens with the aid of the surface model is preferably performed directly (i.e. without iterations) or in a few iterative steps, for example less than 30, 25, 15, 10, 5 or 3 iterative steps.
[0028] An ophthalmic lens, at least one surface of which is calculated from order parameters determined with the aid of a surface model, is designated in the scope of the present application as a lens calculated from this set of order parameters according to a surface model. As mentioned above, the lens calculated according to a surface model may be one of the lenses of a lens pair. In this example, one or both lenses of the lens pair may be calculated according to a surface model. The ophthalmic lens may be, for example, a spectacle lens, for example an individualized and / or personalized spectacle lens. The individualization of the spectacle lens, for example the orientation of the spectacle lens in front of the user's eye, is taken into account in the calculation or optimization of the spectacle lens. The orientation of the spectacle lens may be characterized, for example, by a pantoscopic inclination, a surface formation angle, a pupil distance, a corneal apex distance and / or further parameters. Assuming personalization, for example an adaptation of the perceived design may be performed to meet the personal intended use of one pair of glasses. The spectacle lens may be, for example, a monofocal spectacle lens, a multifocal spectacle lens or a progressive multifocal spectacle lens.
[0029] The surface model may be a parameterized model. The surface model may include at least one variable parameter. The surface model may also have at least one constant parameter (such as the position of the evaluation point on the lens).
[0030] The method for determining the surface model comprises: providing a training data set including a plurality of order parameter sets, each of the order parameter sets including values for at least a portion of parameters required to order at least one ophthalmic lens; providing at least one target value for at least one predetermined characteristic of at least one ophthalmic lens for each of the order parameter sets in the training data set; - providing at least one surface model parameterized by model parameters, with which - for given values of the model parameters - at least one surface of at least one ophthalmic lens can be calculated at least from the order parameter set and / or from variables derived from the order parameter set (providing an initial surface model and possibly an initial parameterization of the surface model); Obtaining a surface model for the calculation of at least one surface of at least one ophthalmic lens, determining optimized values of model parameters of the at least one surface model using the provided target values; Includes.
[0031] Determining optimized values of model parameters of at least one surface model; The method includes optimizing values of at least one model parameter of the surface model (optimizing the parameterization of the surface model) for the purpose of minimizing or maximizing a target function of the model parameters of the at least one surface model, the target function being dependent at least on the model parameters and the provided target values.
[0032] The objective function of each model parameter of the order parameter set includes at least one term that assumes a minimum or maximum when a provided target value of at least one characteristic of at least one ophthalmic lens for the respective order parameter set matches a value of the same characteristic of at least one lens that is computable or calculated using the surface model for a given value of the model parameter of the surface model for the corresponding order parameter set.
[0033] If the at least one lens is one of a pair of lenses, the multiple order parameter sets may include respective values of a portion of the parameters needed to order the pair of lenses. At least one characteristic may be a binocular characteristic of the pair of lenses.
[0034] For example, at least one lens pair including lenses determined for a right eye and a left eye of a person may be calculated using the surface model. In this case, the order parameter set includes order parameters for the right lens and the left lens, respectively. The properties for which target values are provided may also include at least one binocular property depending on the surface data of the left lens and the right lens. The target value of the at least one binocular property may include values input to the calculation at least one property (e.g., a surface property) at a first point of a first lens of the lens pair and the same property at a second point of a second lens of the same lens pair.
[0035] Optionally, an initial complexity of the surface model may also be provided, for example an initial number of model parameters may be predetermined.
[0036] Optionally, the complexity of the surface model may also be optimized or set in addition to the parameterization of the surface model. Optimizing the complexity of the surface model may, for example, include varying and / or regularizing the number of model parameters.
[0037] For example, providing at least one surface model parameterized by the model parameters may include providing at least two surface models of different complexity, the complexity of the surface models including one or more of the following variables: - the type and / or number of order parameters used in the model, - the type and / or number of variables depending on the order parameters, - the number of model parameters, - The type and / or strength of objective function regularization used in the optimization of model parameters.
[0038] The method comprises: providing a validation data set including a plurality of order parameter sets, each order parameter set including values for at least a portion of parameters required to order at least one ophthalmic lens; providing at least one target value for at least one characteristic of at least one ophthalmic lens for each of the order parameter sets in the validation data set; It may further include.
[0039] Obtaining a surface model for calculating at least one surface of the at least one ophthalmic lens includes: calculating, for the provided surface models having different complexities for given pre-determined optimized values of model parameters of the respective surface models, values of a validation objective function and / or values of variables derived from the validation objective function, the validation objective function depending on the provided target values and including at least one term for each of the order parameter sets in the validation data set, the term assuming a minimum or maximum when the provided target value of at least one characteristic of the at least one ophthalmic lens for the respective order parameter set matches a value of the same characteristic of the at least one lens that is computable or calculated using the surface model for the given optimized values of the model parameters of the surface model for the corresponding order parameter set; selecting or determining a surface model for the calculation of at least one surface of the at least one ophthalmic lens from surface models having different degrees of complexity that are parameterized with optimized values of model parameters based on the calculated values of the validation objective function and / or with values of variables derived from the validation objective function; It may further include.
[0040] The term "providing" within the meaning of this application includes "establishing," "sending," "obtaining," "reading," "retrieving from a memory, database and / or table," "receiving," and the like.
[0041] The term "define" also encompasses "establishing," "calculating," "determining," and the like within the meaning of this application.
[0042] Order Parameter Set To determine the surface model, a set of order parameters necessary to calculate at least two additional different ophthalmic lenses or pairs of lenses is provided. Advantageously, in determining the surface model, 10, 100, 1000, 10000, 100000 or more than 1 million order parameters necessary to calculate the additional ophthalmic lenses are used.
[0043] The order parameter set preferably covers a large range, preferably the entire range, within which the ophthalmic lens may subsequently be ordered (see e.g. the refractive limits given by the ophthalmic lens manufacturer, the individual frame parameters in the spectacle lens order, other parameters of the lens such as the freely selectable object distance and other order parameters). For example, the order parameter set may lie within the range of refractive values, e.g. -20dpt to +20dpt for a sphere and -8dpt to +8dpt for a cylinder.
[0044] Thereby, an order parameter set may contain one order parameter, several order parameters or all order parameters necessary to order a single surgical lens or a pair of surgical lenses. Examples of order parameters can also be found in standards established for spectacle lenses (see, for example, EU guideline 93 / 42 / EWG on medicinal products).
[0045] To this end, the order parameter set may include at least one of the following order parameters: - parameters of the ophthalmic lens, such as, for example, the material (optionally having the refractive index of the lens), the desired thickness of the lens, coatings, etc. - refraction values, such as, for example, a sphere and / or a cylinder with an axis and / or an addition and / or a near refraction and / or a prism with a base, - geometric parameters of the eyeglass frames, - the intended use of the ophthalmic lens, e.g. for reading, working at a computer, playing sports, etc., - physiological parameters or characteristics of the future wearer of the ophthalmic lens, - one or more biometric parameters or characteristics of the eye of the future wearer, such as, for example, the location of the center of rotation of the eye, the individual structure of the eye, the pupil diameter, the individual measurements of the wavefront, - the preferred viewing behavior of the future wearer, - other known parameters for the individualization and / or personalization of ophthalmic lenses. Parameters for the individualization of ophthalmic lenses may, for example, characterize the orientation of the ophthalmic lens in front of the lens wearer's eye. Parameters for the personalization of ophthalmic lenses may, for example, characterize the perceived fit of the design to satisfy the personal intended use of one pair of spectacles. These parameters for the individual and / or personalization of ophthalmic lenses may, for example, be expressed in design characteristics, location of the reference points, progression length, etc.
[0046] Variables that depend on the order parameters (e.g., variables derived from the order parameters) may be, for example, the refractive index of a material, the mechanical properties of a material, the mechanical properties of a coating, the thickness distribution of an ophthalmic lens, the desired distribution of residual astigmatism at the position of use, the desired distribution of refractive error at the position of use, etc.
[0047] The order parameter set necessary to determine the surface model may - but does not necessarily - relate to an already provided, already calculated or already manufactured ophthalmic lens. Depending on the embodiment of the surface model, it may be advantageous that the ophthalmic lens is already provided, calculated or manufactured. It is therefore also possible that the order parameter set lies only within the tolerance limits of the order parameters. Thus, the refraction may be within the supply range of the ophthalmic lens, but the lens itself may never have been ordered, calculated or manufactured.
[0048] To reduce the number of data sets (e.g., given effects that are ordered very frequently), redundant orders in the order data set may optionally be removed before definition. Alternatively or in addition, for the same reasons, the order set may be stratified, but nevertheless ensuring high coverage of order parameter ranges in the order data set.
[0049] Characteristics and Target Values At least one target value of at least one predetermined characteristic of the at least one ophthalmic lens is also provided to determine the surface model for each of the order parameter sets in the training data set. The target values for different order parameter sets may be different or the same for at least a portion of the order parameter sets.
[0050] The term "target value" in the sense of the present application comprises a desired or required value of at least one characteristic of an ophthalmic lens, e.g. a spectacle lens or a contact lens. The target value may comprise a plurality of values or a combination of a plurality of values. The target value may for example be learned from a database and / or calculated using a predefined optimization algorithm.
[0051] The at least one predetermined characteristic of the ophthalmic lens may be an optical or geometric characteristic of the ophthalmic lens or of a lens pair including the ophthalmic lens. The at least one predetermined characteristic may be a physical characteristic of the lens, such as, for example, the bulge, the curvature, or variables derived therefrom, such as, for example, surface astigmatism, surface refractive index, etc. The at least one predetermined characteristic may also be an "indirect" characteristic, i.e. a characteristic that appears in conjunction with at least one model (e.g., object distance model, eye model, use position model, etc.). Examples of indirect characteristics are residual astigmatism, refractive error, etc.
[0052] For example, the property may be one of the following properties: - at least one surface elevation and / or a derivative of the elevation (e.g. in an orthogonal direction); - surface characteristics or surface parameters of at least one surface, for example at least one viewing position, such as coefficients of a parametric function describing the surface, for example a spline or polynomial function, - at least one surface characteristic, e.g., smoothness, ability to be continuously differentiated, manufacturability, etc.; - optical variables or properties of at least one surface or of the ophthalmic lens, such as, for example, refractive index or refractive error (preferably in the use position), astigmatism or residual astigmatism (preferably in the use position), vertical and / or horizontal prism (preferably in the use position), higher order imaging errors (preferably in the use position), etc. The optical variables or properties may, for example, be specified in the form of a power vector, - the gradient of an optical variable or property of at least one surface or of at least one ophthalmic lens, such as for example the gradient of the (residual) astigmatism and / or the refractive index or refractive error, - a distribution of optical variables or properties and / or their gradients of at least one surface or of an ophthalmic lens, such as, for example, a distribution of prisms, prism-based, prism vector components of the refractive error and / or of the magnitude and / or axis of the astigmatism or residual astigmatism in the position of use, or a distribution derived therefrom. By "distribution" is understood both the optical properties or their gradients in the sense of a function of the spatial position on the ophthalmic lens (e.g. the (x,y) position) and the frequency distribution of these variables in the sense of a probability distribution, - the width of the zone of good vision (for example the zone in which the residual astigmatism and / or refractive error is less than 1 dps, preferably less than 0.75 dpt or 0.5 dpt), - geometric parameters or characteristics (not included in the order parameters) of an ophthalmic lens, such as, for example, the center thickness of the lens, the edge thickness of the lens, the thickness of a coating, the diameter of the lens, the mass of the lens, etc. - material parameters or material properties (not included in the order parameters) of the surgical lens, - the geometric parameters of the spectacle frame (not included in the order parameters), - the manufacturability of the lens (e.g. the absence of surface undercuts), - binocular characteristics, or characteristics of a lens pair including ophthalmic lenses. Binocular characteristics may be characteristics whose calculation involves at least one characteristic (e.g. surface characteristics, geometric characteristics, optical characteristics, visual perception characteristics, etc.) at a first point of a first lens of a lens pair (e.g. a pair of spectacle lenses, etc.) and the same characteristic at a second point of a second lens of the same lens pair. At least one binocular characteristic of a lens pair may be, for example, a deviation or difference of at least one optical variable or characteristic between a first lens and a second lens of a lens pair. Examples of binocular characteristics of a lens pair are horizontal and / or vertical prism deviation or difference (minus a prescribed prism difference) at corresponding viewing locations (e.g. at least one reference point, such as at the prism reference point, etc.) between the right and left lenses of the lens pair, the difference in magnification at the left and right lenses, the base curvature difference between the left and right lenses, the locations of the design points or their differences, the visual perception characteristics of the future wearer of the ophthalmic lens calculated using a surface model, etc. The parameters or characteristics of the lens pair may also include a deviation in at least one geometric parameter between a first lens and a second lens of the lens pair, such as, for example, a center thickness, an edge thickness, a coating thickness, a deviation in the curvature of the front surface, etc. - discomfort with respect to the quality of vision and / or posture of the wearer of the ophthalmic lens, calculated using a surface model; - The properties of the visual perception of the future wearer of the ophthalmic lens calculated using the surface model.
[0053] Other relevant characteristics may also be considered.
[0054] The target values may be values (e.g. nominal values, etc.) of at least one ophthalmic characteristic of the lens that have already been calculated or will be calculated according to a known method (e.g. by minimizing or maximizing a known target function in an iterative optimization method) or will be generated for different sets of order parameters. Similarly, it is also possible to obtain the target values using measurements of an already manufactured ophthalmic lens. The order parameters of the already manufactured or to be manufactured lens are preferably already at least partially known.
[0055] For example, a measured value of at least one characteristic of an already manufactured ophthalmic lens, or a value determined or determinable from one or more measured values of an already manufactured ophthalmic lens, may be set as the target value of at least one characteristic of the at least one ophthalmic lens. Similarly, it is also possible to set a nominal value of the ophthalmic lens to be manufactured as the target value.
[0056] For example, a target value for the mean surface refractive index of a surface of an ophthalmic lens may be determined depending on the position of the lens, given a known refractive index from a measurement of the elevation of the glass surface (via the mean curvature of the surface). A target value for the astigmatism of a spectacle lens in a position of use (also depending on the position) may be determined from the elevation measurement, the pantoscopic tilt, the surface formation angle, the glass length and the glass height (order values), and the position of the center of rotation of the eye (order values or model assumptions).
[0057] In determining the surface model, in addition to the order parameter set of a plurality of ophthalmic lenses (base lenses), one or more surfaces of these lenses belonging to the order parameter set may be used. The ophthalmic lenses used in determining the surface model are also called base lenses within the scope of this application. The base lenses may be lenses calculated and optimized according to known methods, and optionally manufactured.
[0058] Additionally or alternatively, target functions and / or their derivatives for the surfaces, which are necessary for calculating an ophthalmic lens according to the prior art or for calculating a base lens on the basis of these order parameter sets, may also be used. Said derivatives are understood, for example, as the change in the target functions given the change in elevation. To this end, the target values may correspond to nominal values that are input to these target functions. Thereby, the target functions and / or their derivatives for the surfaces may preferably exist such that they can be evaluated for any surface. Thereby, the target functions and their derivatives are usually evaluated on the basis of a suitable parameterization of the surface of the ophthalmic lens.
[0059] If the surfaces of a number of basic lenses are used to determine the surface model, it may be advantageous if these surfaces have already been calculated. Similarly, it may be advantageous if the basic lenses have also been manufactured, since their calculations already have a purpose and can be reused to determine the surface model without additional computational power being consumed.
[0060] The methods described herein may also be performed without additional measures on the measured surfaces of the ophthalmic lens (and the distance of the surfaces relative to each other) instead of the calculated surfaces. In addition, other properties may also be measured, such as the refractive index.
[0061] The above description relates to characteristics and target values for the training data set. Obviously, characteristics and target values may also be determined or calculated identically or similarly for other order parameter sets (e.g., validation and test data sets). Thereby, the same characteristics with corresponding target values need not necessarily be used for training, validating and testing data sets. However, it may often be simpler to use the same characteristic target values in connection with various data sets.
[0062] Surface Model The surface model can be any model, such as a model based on machine learning, for example, as described in, for example, Jeremy Watt, Reza Borhani, Aggelos Katsaggelos: Machine Learning Refined: Foundations, Algorithms, and Applications, Cambridge University Press, 2020.
[0063] The surface model may be described by suitably determined model parameters which, together with at least a portion, or in the best case all, of the order parameters and variables derived therefrom, are used to calculate one or more surfaces of the ophthalmic lens.
[0064] When one or more order parameters can be reasonably expressed as one or more real numbers, in order to ensure continuity of the design of the ophthalmic lens with respect to these order parameters, it is advantageous for the surface model to be constructed such that the surface generated by the surface model is a continuous or even continuously derivable function of the order parameters.
[0065] The surface model determined by the model parameters may be implemented as or may include a regression model that receives at least a portion, or preferably all, of the order parameters and / or variables derived therefrom as input variables and calculates therefrom one or more surfaces of the ophthalmic lens or pair of ophthalmic lenses.
[0066] The surface model or a part thereof may additionally or alternatively be designed as a regression model and as a classification model or may include a classification model. For example, if only determined glass diameters are available in blanks within the scope of the manufacture of spectacle lenses and a diameter of a glass blank is to be selected, this may be done by a classification model. Such a classification model may for example calculate the probability of the suitability of the available glass blank diameters for the ophthalmic lens to be manufactured, whereby finally the glass blank with the highest probability and thus most suitable for the ophthalmic lens to be manufactured may be selected.
[0067] Similarly, the classification model may also be used in the manufacture of ophthalmic lenses (e.g., spectacle lenses) to calculate the optimal base curvature and / or optimal diameter of the ophthalmic lens (e.g., spectacle lens), where the probability of suitability of the available base curvatures is calculated and the base curvature with the highest probability is selected for the manufacture or calculation of the ophthalmic lens (e.g., spectacle lens).
[0068] To be able to flexibly represent different designs of modern ophthalmic lenses, it is advantageous for the surface model to possess a sufficiently large number of model parameters, for example 10, 30, 50, 100, 500 or even more than 1000, 10000, 100000 or even more model parameters.
[0069] If the surface model is designed, for example, as a regression model or comprises a regression model, determining the surface model may consist of starting from an order parameter set (for example an order parameter set of a basic lens) and adjusting the model parameters of the surface model, so that any ophthalmic lens whose surface or surfaces can be calculated from those order parameter sets with the aid of the surface model differs only slightly from an ophthalmic lens which can be calculated or has already been calculated from the same order parameters by a predetermined method according to the prior art, using predetermined criteria.
[0070] The regression model used in or as the surface model can be a linear regression model, which typically simplifies the calculations required to determine the surface model since the model parameters can be determined from a system of linear equations.
[0071] However, it is also possible to use non-linear regression models instead of linear regression models without further measures. Such models are more flexible and may represent more complex correlations between the order parameters and the surface or surfaces of the ophthalmic lens. At the same time, however, it is more difficult to properly determine the model parameters, since for this purpose non-linear optimization algorithms are common that do not necessarily converge on the global optimum of the model parameters. For example, neural networks, among which deep neural networks, may be used as non-linear regression models, but also other non-linear regression models known from the field of machine learning. These regression models, for example neural networks, may be trained with the provided set of order parameters in the training data together with the associated target values.
[0072] The surface model can also be a combination of linear and / or non-linear regression models, classification models and / or neural networks. The combination is envisaged to reduce the complexity of the surface model and to save computation time and / or resource consumption. Examples of combinations are: - Example: classification models before regression models and / or neural networks: the model with the minimum complexity that still produces a sufficiently good surface with the order data is selected from a set of regression models with different complexity. The advantage of this procedure is the possible improvement of the calculation time and / or the utilization of few resources, - Example: Regression model before classification model before model regression model and / or neural network: With the order data the shape of the lens is approximately determined with the aid of a low complexity regression model. A subsequent classification model determines the glass blank from the order data and the approximate shape. A subsequent regression model determines the final lens from the order data and the glass blank. The advantage of this procedure is the reduced complexity of the regression model for determining the lens.
[0073] Similarly, it is advantageous to control or optimize the complexity of the surface model so that the order parameter set can be used to calculate the surfaces of ophthalmic lenses not included in the training data set. For example, this can be done with the aid of regularization and / or by selecting the order parameters and / or by calculating a sufficient number of different types of variables (auxiliary variables) derived from the order parameters and / or by an appropriate selection of the number of model parameters and / or the type of model.
[0074] For reasons of numerical stability, it is also reasonable to transform the order parameters and / or auxiliary variables derived therefrom or relying on them before passing them to the surface model so that they have a mean value of zero and, for example, a standard deviation of 1 over the distribution of the order parameters of ophthalmic lenses used in the determination of the model parameters.
[0075] Determining the model parameters of the surface model and adjusting its complexity Defining or determining model parameters First, at least one surface model parameterized by model parameters is provided or predetermined, using which - for given values of the model parameters - at least one surface of at least one ophthalmic lens can be calculated at least from the order parameter set and / or from variables that are dependent on the order parameter set.
[0076] Thereby, an initial parameterization and an initial complexity of the surface model may be provided or determined. Providing an initial parameterization of the surface model may include providing initial values for the model parameters of the surface model. Providing an initial complexity of the surface model may include determining or predetermining an initial number of model parameters of the surface model. Final model parameters, and possibly the complexity, are determined or are determined by a suitable optimization method. Thus, the final model parameters form an optimal set of model parameters.
[0077] The model parameters of the surface model are preferably determined to represent an optimal set of model parameters that minimizes or maximizes a predetermined objective function of the model parameters.
[0078] The optimal set of model parameters can be obtained using established arithmetic optimization algorithms (e.g., simple gradient descent, conjugate gradient descent, statistical gradient descent or similar algorithms). If a neural network is used as a regression model, the backpropagation algorithm - itself the only gradient-based algorithm adapted for this type of model - may be used to minimize the objective function.
[0079] The objective function of each model parameter of the order parameter set may include at least one term that assumes a minimum or maximum when a provided target value of at least one characteristic of at least one ophthalmic lens for the respective order parameter set matches a value of the same characteristic of at least one lens that is computable or calculated using the surface model of the corresponding order parameter set for a given value of the model parameter of the surface model.
[0080] The objective function may include a single term, a sum of multiple terms, or a weighted sum of multiple terms. For example, different objective functions may be used depending on whether (i) there are already existing surfaces (e.g., surfaces calculated according to conventional methods) for the order parameter set and whether these surfaces should be used in the objective function, or (ii) a surface corresponding to the order parameter set needs to be calculated first, or (iii) the objective function should not be determined with the aid of a surface.
[0081] Furthermore, it is advantageous if the gradient of the objective function can be quickly calculated with respect to parameters in the surface parameterization output by the surface model, for example as an analytical function.
[0082] The optimization of the parameterization and, where applicable, the complexity of the surface model may be performed such that the deviation of the value of at least one characteristic of an ophthalmic lens for at least one surface calculated from a set of order parameters with the aid of the surface model and a corresponding target value (possibly depending on the order parameters) of this characteristic is minimized.
[0083] The deviation of at least one value of at least one predetermined characteristic of the lens, calculated or calculable according to a surface model from a determined order parameter set, from at least one target value of this characteristic for the same order parameter set, can be quantified in various ways: the difference between at least one value of at least one predetermined characteristic of the lens, calculated or calculable according to a surface model from a determined order parameter set, and at least one target value of this characteristic for the same order parameter set, or a convex or concave function of the difference (for example the square, the magnitude or the negative of it) may be used as a measure of this deviation.
[0084] Thus, at least one term of the objective function of the model parameters may comprise a difference, or a convex or concave function of the difference, between at least one value of at least one predetermined characteristic of the lens, at least one surface of which is calculable for order parameters according to the surface model, and at least one target value of this characteristic for the same set of order parameters. Convex functions may be used in particular in the minimization of the objective function. Given the maximization of the objective function, concave functions may be used.
[0085] It is likewise possible to use other functions of at least one value of at least one predetermined characteristic of the lens, calculated or calculable from the determined order parameter set according to the surface model, and at least one target value of this characteristic for the same order parameter set, to describe or quantify the deviation between these values. Such a function is, for example, the ratio of at least one value of at least one predetermined characteristic of the lens, calculated or calculable from the determined order parameter set according to the surface model, and at least one target value of this characteristic for the same order parameter set. Other functions, for example logarithmic functions of the ratio, are likewise possible.
[0086] It is also possible to use an objective function for at least one optical characteristic used in a conventional optimization or calculation of at least an ophthalmic lens. Thus, one or more terms of the objective function of the model parameters may form or be understood as an objective function for the optimization or calculation of at least one ophthalmic lens for a given parameter set, the objective function being evaluated for various order parameter sets. The objective function may rely in a known manner on the actual values of at least one optical characteristic (evaluated for an ophthalmic lens for which at least one surface has been calculated or can be calculated from the determined order parameter set and / or variables derived therefrom according to a surface model) and the corresponding target values. Similarly, it is possible to use different objective functions for different order parameter sets. In this case, it is not mandatory to determine the surface model using an already calculated and / or manufactured surface. By way of example, such a target function may be the cumulative sum over multiple viewing locations of the squared refractive error deviations and the respective target astigmatism calculated directly from the order parameter set, or the average value - calculated over multiple viewing locations - which is then summed or averaged over the order parameter set.
[0087] The objective function of the model parameters may include multiple terms quantifying or describing the difference between the surface of the ophthalmic lens calculated using the surface model and the corresponding target value using different properties (e.g. optical and / or geometric properties, direct and / or indirect properties). Thereby, the additional properties (e.g. additional optical and / or geometric properties, direct and / or indirect properties) may depend on the order parameters. Examples of properties are vector components or magnitudes of the residual astigmatism at the position of use, refractive errors at the position of use, deviations of minimum or maximum glass thickness from the corresponding order values of the glass thickness, etc.
[0088] In addition, the objective function may also include terms that quantify or describe the differences between the binocular properties of two pairs of ophthalmic lenses (one pair calculated with the aid of a surface model and one pair calculated using prior art methods).
[0089] The objective function may also include at least one term that includes a design difference between ophthalmic lenses having various order parameters, such that such sub-term no longer relates only to a single ophthalmic lens, but to differences between two or more adjacent ophthalmic lenses in the order range, and represents an advantageous property of a product including multiple ophthalmic lenses.
[0090] In this way, for example, the perceived design similarity (but not necessarily consistency) of variable focus lenses across various refractions may be formulated as an objective function, whereby a desired design need not be specified for, for example, a single effect that occurs particularly frequently, and assuming other effects, designs result therefrom without having to be specified separately. The advantage of such a term involving the difference between two or more ophthalmic lenses is that it is often difficult to specify a surface design that is constant across an order parameter range, since these cannot be constant due to more fundamental principles (e.g., Minkwitz's theorem).
[0091] The objective function may include a weighted or unweighted sum of terms arranged for each of the order parameter sets across all order parameter sets in the training data set. Instead of a sum, it is also possible to calculate an average or median. Instead of a sum, it is also possible to use a more complex function, e.g., a non-linear function.
[0092] As discussed above, various optimization algorithms (e.g., simple gradient descent, conjugate gradient descent, statistical gradient descent, or similar algorithms) may be used to determine an optimal set of model parameters. For example, optimizing the values of the model parameters may include regularizing an objective function used in optimizing the model parameters.
[0093] If an optimization method requiring the gradient of the objective function (e.g. assuming the backpropagation algorithm in a neural network as the surface model, or - for example - gradient descent) is used to optimize the parameters of the surface model, these may be calculated analytically, numerically or with the aid of a combination of analytical and numerical methods. In particular, assuming the backpropagation algorithm, in the backpropagation step - instead of the residuals usually used - the gradient of the objective function is propagated backwards through the network (the residuals usually used are the gradients of the quadratic objective function usually used).
[0094] The above description relates to an objective function for the determination of model parameters. Obviously, other objective functions, such as a validation objective function or an objective function for testing a surface model (test objective function), may also be established or pre-determined in the same manner.
[0095] The objective function may be normalized, for example, a cumulative sum over the characteristics of one or more ophthalmic lenses (lens pairs) may be divided by the number of ophthalmic lenses (lens pairs) in the training, validation or testing data sets relative to the number of lenses (lens pairs) in the training, validation or testing data sets, depending on whether the objective function is used to determine model parameters, to validate a model, or for testing, respectively.
[0096] Adjusting the complexity of the surface model As already mentioned above, at least two surface models of different complexity may first be determined in order to determine the optimal complexity of the model. The complexity of the surface model may include one or more of the following variables: - the type and / or number of order parameters used in the model, - the type and / or number of variables derived from the order parameters, - the number of model parameters, - The type and / or strength of objective function regularization used in the optimization of model parameters.
[0097] Adjusting or optimizing the complexity of the surface model may also include providing: - a validation data set including a plurality of order parameter sets, each of the order parameter sets including values for at least a portion of parameters required to order at least one ophthalmic lens; - for each of the order parameter sets in the validation data set, a target value for at least one characteristic of at least one ophthalmic lens;
[0098] Further, obtaining a surface model for calculating at least one surface of the at least one ophthalmic lens may include: calculating values of the validation objective function and / or values of variables derived from the validation objective function for the provided surface models of various complexities for given previously determined optimized values of the model parameters of the respective surface models; selecting or determining a surface model for the calculation of at least one surface of the at least one ophthalmic lens from surface models having different degrees of complexity that are parameterized with optimized values of model parameters based on the calculated values of the validation objective function and / or with values of variables derived from the validation objective function; may include:
[0099] As already mentioned above, the validation objective function relies on the provided target values, the validation objective function includes, for each of the order parameters in the validation data set, at least one term that assumes a minimum or maximum when the provided target value of at least one property of at least one ophthalmic lens for the respective order parameter set matches a value of the same property of at least one lens that is computable or calculated using the surface model for a given optimized value of the model parameters of the surface model for the corresponding order parameter set.
[0100] The validation objective function may be constructed identically or similarly to the model parameter objective function. However, different objective functions can be used.
[0101] For example, if no regularization parameters are used in the optimization of the model parameters, the objective function for the optimization of the model parameters and the validation objective function may contain the same terms depending on the respective target values (i.e., for the training data set or the validation data set, respectively). If regularized, the terms of the objective function for the optimization of the model parameters that contain the regularization parameter(s) may be omitted for the calculation of the validation objective function. The same can also be achieved in that the regularization parameter(s) are set such that the corresponding terms do not contribute to the validation objective function (e.g., by setting the regularization parameter(s) to 0). Of course, for all of them, the corresponding target values are replaced with those based on the validation data set rather than the training data set. However, in addition, additional or exclusively different properties of the ophthalmic lens may also be utilized in the validation (compared to the optimization of the model parameters). For example, in optimization, the square of the difference in elevation may be minimized (i.e., the corresponding characteristic is the elevation of a given evaluation point on the ophthalmic lens), and in validation, the square of the difference between the effect of the ophthalmic lens and the corresponding target value (where the characteristic is, for example, a power vector, the effect as a sphere / cylinder / axis, or one or more components of the effect) is minimized.
[0102] Training with already calculated or known surfaces If an already calculated surface of the ophthalmic lens is already partially available (e.g. surfaces or lenses calculated according to conventional optimization methods), the model parameters may be selected such that the surface output by the surface model matches optimally well with the already calculated surface (target surface). Possible criteria for this may be determined in an objective function of the model parameters. In the simplest case, the objective function contains a term which is a sum of convex functions (e.g. squared) of the differences of the elevations of the surfaces, this elevation being calculated or computable from the surface model with the given model parameters and the target surface. The summation is thereby performed point-wise over all pairs of elevations of the target surface in the surface calculated or computable with the surface model and over the base lens or order parameter set. The already calculated surface of the base lens together with the associated order parameters and / or variables derived therefrom may be considered as a training data set.
[0103] Differently weighted sums across the ridges of the ophthalmic lens may also be used. For example, weighting may be greater at points of the ophthalmic lens that are evaluated as being essentially important than at other points. For example, weighting may be greater at areas of the ophthalmic lens that are more frequently found to ensure high optical quality (e.g., assuming a spectacle lens, areas of uncut spectacle lenses located in frames after polishing, or assuming a variable focus lens, areas where residual astigmatism is below a determined threshold, e.g., 0.5 dpt).
[0104] As an alternative to the difference in elevation, their ratio or the logarithm of the ratio may also be used.
[0105] In addition, other terms in the objective function may also be used for the model parameters. For example, it is possible that an ophthalmic lens calculated from a surface model has a diameter different from the nominal diameter. In this event, the objective function of the model parameters of the surface model may also include a term that penalizes the deviation of the diameter calculated by the surface model or implicitly obtained from the calculation results (e.g. due to an excessively high curvature) (e.g. the objective function rises very strongly if the calculated diameter is smaller than the nominal diameter).
[0106] If the already calculated surface of the basic lens and the surface calculated by the surface model are given in different parameterizations (e.g. the bumps are specified in various point rasters), it is advantageous to convert the already calculated surface of the basic lens into the parameterization output by the surface model, for example via interpolation. However, the adjustment of the surface parameterization may also be done in the opposite direction, or a completely different parameterization may be chosen (e.g. a representation by Zernike polynomials).
[0107] Additionally or alternatively, the objective function of the model parameters also includes other terms quantifying the difference between the surface of the ophthalmic lens calculated using the surface model and the surface of the basic lens using optical and / or geometric properties, whereby the optical or geometric properties may also depend on order parameters (e.g. vector components or magnitudes of the residual astigmatism at the position of use, refractive errors at the position of use, or deviations of minimum or maximum glass thickness with corresponding order values of glass thickness).
[0108] These terms usually consist of a sum or weighted sum of point-by-point differences of the optical parameters of the lens or base lenses calculated using a surface model, which are summed over all base lenses. The points for the summation may be predetermined by a raster of evaluation points of the ophthalmic lens or by a raster of viewing directions. Instead of a (weighted) sum, other functions may be used, for example non-linear functions.
[0109] As explained above, the objective function may also include a term that quantifies the difference between the binocular characteristics of two pairs of ophthalmic lenses (one calculated with the aid of a surface model and one calculated according to conventional optimization methods).
[0110] Global optimization using objective functions In a second example, an objective function of the model parameters is used that does not rely on the surfaces of the base lenses, for example, this may be the case when these surfaces have not yet been fully calculated to determine the model parameters, or in the event that correspondingly calculated surfaces do not yet exist for the order parameter set of the base lenses.
[0111] In such a target function, the difference between certain properties of the surface calculated using the surface model and the desired target values of these properties (possibly depending on the order parameters) may be calculated. Possible properties are the optical properties mentioned above, for example optical properties (e.g. refractive error; vector components; and / or magnitude and / or axis of astigmatism or residual astigmatism in the use position; prism, prism base, distribution of vector components of the prism; or distribution of variables derived therefrom), geometric properties, binocular properties, or properties of the visual perception of the future wearer of the ophthalmic lens calculated using the surface model.
[0112] The objective function of the model parameters may include terms representing weighted deviations of properties of the ophthalmic lens calculated or calculable using the surface model from the desired curvature. For example, the fulfillment of effects required in standards (e.g. high weighting at the reference points assumed for the ophthalmic lens) may be controlled by weighting. A term having a desired central thickness at the edge or a minimum value in the thickness distribution may also be included in the objective function of the model parameters. A desired mechanical property, e.g. actual or simulated breaking strength, may also be represented in the objective function terms.
[0113] The objective function can be one of the objective functions described above. In particular, all the examples of objective functions of the model parameters from the previous section that do not rely on the surfaces of the base lenses may be used with appropriately selected target values by substitution of the corresponding properties of the base lenses, where applicable, relying on the order parameters.
[0114] The objective function of the model parameters may also include terms that have already been used as objective functions in the optimization of ophthalmic lenses with the aid of a general optimization method, where the objective function of the model parameters includes terms that sum or average the objective functions of the optimization method over additional ophthalmic lenses.
[0115] Thus, determining the model parameters by optimizing an objective function of the surface-independent model parameters of the base lens corresponds to the simultaneous optimization of multiple ophthalmic lenses obtainable from the order parameter set by calculation and / or optimization. Instead of directly varying the surface parameterization (e.g., ridges or spline coefficients) of every single lens, as is common, the model parameters of the surface model controlling the surface curvature are thereby varied to minimize the sum of the individual objective functions for the optimization of every ophthalmic lens that can be calculated or optimized from the order parameter set for use in determining the model parameters.
[0116] Obviously, new and as yet unknown objective functions that represent advantageous properties of an ophthalmic lens may also be used for individual ophthalmic lenses.
[0117] In addition, i.e. in addition to the terms in the objective function for the individual ophthalmic lenses, the objective function for the model parameters of the surface model can also include terms that include the design differences between ophthalmic lenses having various order parameters. The advantages of such terms have already been discussed above.
[0118] For this reason, if the surface calculated from the surface model is a continuous or even continuously derivable function of order parameters and / or of variables derived therefrom, it is expected that, assuming an appropriately set complexity of the surface model, only small differences will exist between the surface calculated from the surface model and the surface calculated by optimization of the same objective function by conventional optimization methods.
[0119] Determining the complexity of a surface model using the quality of the computed surface. In order to appropriately set the complexity of the surface model so that a correct ophthalmic lens can be calculated using the surface model even from an order parameter set that was not present in the training data set, additional terms weighted with one or more different factors, known as regularization, may also be used in addition to or instead of a direct variation of the number of model parameters of the surface model (and also the order parameters and / or variables derived therefrom). These terms are usually quadratic terms in the model parameters. However, other exponents may also be used (e.g. the absolute value of the model parameters may be used) or other functions of the model parameters may also be used instead of the model parameters themselves (e.g. the difference of the spline coefficients of adjacent splines of the representation of the ophthalmic lens surface).
[0120] In order to check the quality of the surface model after adjusting the model parameters, it is proposed to train the model parameters of the surface model using a part of the data set for validation or final testing of the model, rather than using the entire available data set (i.e. a data set containing at least the order parameters, the variables derived therefrom, if applicable, and the associated calculated surfaces, if applicable). The validation of the adjusted model complexity and the final testing of the surface model may be performed using the same objective function of the model parameters used to adjust the model parameters of the surface model, but preferably without the terms resulting from the regularization. It is also possible to use an objective function for validation or testing of the surface model that is different from the objective function for the definition of the model parameters.
[0121] For example, a check may be made whether a sufficiently different set of order parameters not present in the training data set would lead to a slight deviation of the surface calculated from the surface model from the target surface. This may typically be achieved in that the order data set, together with the surfaces of the base lenses, is subdivided into a training data set and a validation and / or test data set. As explained above, the complexity of the surface model may be selected using the validation data set.
[0122] Thus, the method for determining a surface model comprises the following steps: providing a validation data set comprising a plurality of order data sets; providing at least one target value for at least one predetermined characteristic of the ophthalmic lens for each of the order parameter sets in the validation data set; validating the obtained surface model for the calculation of at least one surface of the at least one ophthalmic lens based on a validation data set; may include:
[0123] Alternatively or additionally, the method for determining a surface model may comprise the following steps: providing a test data set comprising a plurality of order data sets; providing at least one target value for at least one predetermined characteristic of the ophthalmic lens for each of the order parameter sets in the test data set; testing the obtained surface model for calculation of at least one surface of at least one ophthalmic lens based on a test data set; may include:
[0124] Different parts of the same data set may be used for validation and testing. The purpose of validation may be the determination of an appropriate model architecture (here also called model complexity) or an appropriate value of the regularization parameter. The purpose of testing may be a check of the trained and selected model to avoid overfitting.
[0125] The function evaluated in validation and testing may be the same, e.g., the validation function (also called the test objective function) described above. As explained above, the validation objective function typically does not include additional terms, including the regularization parameter.
[0126] In testing, the validation objective function may be evaluated for a test data set assuming fixed model parameters of the surface model and compared to the value of the validation objective function and / or the objective function for the model parameters (but without the regularization term) evaluated for the validation data set. Testing is successful if the value of the validation objective function evaluated for the validation data set and the value of the validation objective function evaluated for the test data set are similarly large. How significantly they actually differ from each other depends on the data in the respective validation or test data set (among other things, the number of data has a large impact if a normalized objective function for the number of data is not used) and the underlying model.
[0127] The values of the validation and test objective functions that are not normalized to the number of data (e.g. the sum of the quadratic deviations of the projections from the lens calculated using the conventional method and the lens calculated using the surface model at the determined evaluation point on the lens) may simply be compared to each other when the test and validation data sets contain the same amount of data. If these functions are divided by the number of data, what is obtained are normalized objective functions that can be compared even when the test and validation data sets contain different amounts of data points.
[0128] As long as normalized objective functions are used, it is advantageous that the values of the objective functions to be compared (i.e. the values of the validation objective function in the evaluation with the test data set and the validation data set) do not differ too greatly from each other (e.g. the absolute value of the difference between the two values of the objective function should be less than a predefined threshold). The value of such a threshold depends largely on the type of objective function used and should be a small percentage (e.g. 0.3-0.01 times) of the variation (e.g. maximum-minimum) of the validation objective function assuming evaluation with different models or different values of the regularization parameter. Assuming the use of non-normalized objective functions, these may be pre-divided by the number of ophthalmic lenses (lens pairs) in the respective data sets and thus normalized.
[0129] Calculation of variables based on order parameters As already mentioned above, order parameters derived from the above order parameters, for example size-dependent variables (auxiliary variables), may be used as input variables of the surface model.
[0130] For example, if the surface model includes or consists of a regression model, it may be advantageous for the input variables of the regression model from which the surface is calculated to include one or more variables calculated from the order parameters (auxiliary variables) in addition to or instead of the order parameters.
[0131] Examples of auxiliary variables are: the desired distribution of residual astigmatism and / or refractive error assuming a variable focus lens, preferably depending on the viewing direction of the eye in the position of use, the desired distribution of the residual astigmatism normalized by addition and / or of the refractive error normalized by addition, depending on the viewing angle of the eye, assuming a variable focus lens, preferably in the optimal position of use, a desired thickness distribution at one or more points of the ophthalmic lens (e.g. at the edge), which may also depend for example on the materials and / or layers or coatings of the ophthalmic lens or on the optical and / or mechanical properties of the materials and / or layers or coatings, - the optical and / or mechanical properties of the material of the ophthalmic lens (e.g. refractive index, modulus of elasticity, coefficient of thermal expansion), - The optical and / or mechanical properties of ophthalmic lens coatings (e.g. thickness distribution, elastic modulus, coefficient of thermal expansion).
[0132] Thereby, preferably parameters are selected which have a large influence on the surface of the ophthalmic lens or which are expected to have a large influence on the surface.
[0133] Neural Networks as Surface Models If the surface model includes or consists of a neural network, the input layer of the neural network is populated with order parameters and / or auxiliary variables calculated therefrom.
[0134] Thereby, the weights of a neural network (i.e., the strengths of the neuronal links) represent at least a portion of the model parameters.
[0135] The output layer may represent the entire calculated surface of the ophthalmic lens or a portion of the calculated surface (e.g., as bumps in a determined raster or grid having a desired, and if applicable, yet to be set, resolution).
[0136] In addition to the input and output layers, a neural network may also contain one or more hidden layers.
[0137] The way in which a neural network is constructed, e.g. the number of layers, the number of neurons in different layers, the type of links of layers between each other, etc., is implicitly determined by the model parameters (e.g. their number).
[0138] In particular, in addition to or instead of the order parameters, it may be advantageous to populate the input layer with one or more variables (auxiliary variables) calculated from the order parameters (see, e.g., section "Calculation of variables derived from order parameters" for auxiliary variables that may be used as input variables for the surface model).
[0139] Similarly, it may be advantageous to design the neural network such that one or more of these auxiliary variables are at least approximately represented in the network or appear during training of the network. Model parameters that are not used as weights in the neural network may also affect the calculation of the auxiliary variables.
[0140] Likewise, it may be advantageous to limit the size of the neural network in that the output layer represents a relatively coarse rasterization of the surface of the ophthalmic lens (e.g. a raster of only 10x10 or 20x20 ridges). In order to compute therefrom a representation of the surface of the ophthalmic lens suitable for manufacturing, it is proposed to interpolate the ridges output from the neural network to a higher resolution raster (e.g. a raster of e.g. 100x100 points with the aid of linear or bicubic interpolation) and, if applicable, to post-optimize by several steps of an optimization method according to the prior art. This post-optimization usually requires only a few iterations for convergence, as long as the neural network is trained using the results of an optimization method with the same objective function.
[0141] The optimized model parameters and, if applicable, the surface model determined or defined with the optimized complexity may be suitably stored and subsequently made available for the calculation of an ophthalmic lens from an order parameter set. For example, the surface model or a part of the surface model (such as the model parameters) may be stored in a suitable storage, for example a database. At least a part of the order parameter set required to determine the surface model or the target values associated therewith may likewise be stored in the storage. As mentioned above, the target values may for example at least in part be surface values of an ophthalmic lens (base lens) already calculated according to an established method or variables derived therefrom. The target values may likewise be nominal values to be input into an objective function (for example an objective function according to the prior art) for optimizing the ophthalmic lens.
[0142] The surface model determined as described above may be further modified. The method may result in a modification of the surface model. Examples of modifications are the addition of additional layers in a neural network, or embedding the surface model in additional functions, which for example interpolate or transform the surface elevations (for example, transforming the neural network into a support vector machine, a decision tree or any other regression model).
[0143] The surface model calculated according to the method and embodiment variants according to the above-mentioned aspects of the invention, as well as the method using this surface model, preferably comprises at least one of the following characteristics or advantages: The evaluation of the surface model for an order parameter set has a low computational cost compared to optimization by prior art methods. Typically, less than 90% of the computational power or computation time for calculating the surface, preferably less than 50%, 20%, 10%, 5%, 2% or 1%, is used for the iterative modification of the surface of the ophthalmic lens. This can be advantageous compared to common iterative algorithms, since the computational cost in each iteration remains the same, but the modification of the surface decreases with each iteration. the surface model does not necessarily require the provision of an initial surface (starting surface) as an input that is entered into the calculation as an input variable and changed during the optimization method, - No need to pre-calculate and store surface modifications based on order parameters; - A hybrid calculation of the surface with subsequent corrections (e.g. post-optimization) is possible. Such a hybrid calculation may have, as a final effect, a shorter optimization time due to a better initial surface; - It is possible to continuously improve the surface model.
[0144] Thus, by using the surface determined by the method according to the above-mentioned aspects and embodiment variants, it is possible to quickly and efficiently calculate the surface of an ophthalmic lens for any order parameters of a customer and to generate and / or visualize this surface.
[0145] The method according to the above-mentioned embodiment also has advantages in the development of a series of ophthalmic lenses (for both individual ophthalmic lenses and pairs of ophthalmic lenses) since the surface model can be used unchanged as long as the order parameters (e.g. refractive index) or variables based thereon (e.g. base curvature system) that differ between the two different series are present in the training data set and are subjected to variation.
[0146] Another application of the method according to the above-mentioned embodiment is the interpolation between different families of ophthalmic lenses (e.g. between different products, such as families of multifocal lenses offered for different intended uses, or between multifocal and monofocal lenses, for example). For this purpose, the ordered set only needs to be extended by one variable corresponding to the family of ophthalmic lenses. Given N families, a tuple (s1, s2, ..., s N ) is proposed, whereby the sequence of ophthalmic lenses i is expressed as a tuple (δ 1,i , δ 2,i ,…,δ N,i ), where δ j,i is the Kronecker delta symbol. Then, the interpolation between the different series of ophthalmic lenses is done using values s between 0 and 1. j may be selected, where s j The sum of these tuples is 1. Then, such tuples are used together with the current order data set to calculate at least one surface of the ophthalmic lens with the aid of a correspondingly trained surface model.
[0147] Further Aspects A second aspect of the present invention relates to a computer-implemented method and corresponding device for determining at least one of the surfaces of one or more ophthalmic lenses by a surface model previously determined from order parameters and / or from variables derived therefrom. The term "define" in the sense of the present invention includes the determination or calculation of at least one surface of one or more ophthalmic lenses.
[0148] The method is: providing an order parameter set for at least one ophthalmic lens; providing a function for calculating at least one surface of the at least one ophthalmic lens from a set of order parameters of the at least one ophthalmic lens and / or from variables dependent on the order parameters, the function being a surface model or a function which approximately performs a mapping of the set of order parameters to the at least one surface of the at least one ophthalmic lens, the mapping being performable using the surface model; determining surface data of at least one surface of at least one ophthalmic lens from the provided set of order parameters (not iteratively but directly) with the aid of a provided function; Includes.
[0149] The surface model may be a surface model as described above, i.e. a surface model determined or verified according to the method as described above. The surface data of at least one surface is preferably determined directly (i.e. not iteratively) from the provided order parameter set or using a few iterative steps, for example less than 30, 25, 15, 10, 5 or 3 iterative steps. As mentioned above, this leads to a significant reduction in the calculation time required to calculate a surface or lens for any order parameter set. It is also not necessary to pre-calculate and store surface modifications in dependence on the order parameters, which reduces storage space requirements. It is also possible to continuously update and improve the surface model in a simple manner.
[0150] The surface model may be used directly to calculate at least one lens surface.
[0151] Alternatively, functions of the surface model may be used, for example functions which approximately perform the calculations with the surface model determined according to the invention. Such functions may be generated within the scope of a simplification of the surface model, for example in that neurons of a neural network having similar activation patterns are combined, or within the scope of different transformations of the surface model used.
[0152] Furthermore, the above-mentioned preferred embodiment variations or advantages described above also apply pertinently with respect to this method for this device.
[0153] The method optionally further comprises the determination of other variables related to the manufacture of the surface (e.g. the current diameter and type of the blank from which the lens is to be manufactured), whereby the surface calculated in such a manner no longer requires further optimization or needs to be corrected with only relatively little computational effort.
[0154] The method for determining at least one surface of at least one ophthalmic lens may also comprise performing a correction of at least one surface calculated using a surface model, the correction being an optimization of the surface calculated using the surface model and / or a superposition with a superposition surface and / or a correction of manufacturing-dependent deviations of the surface or of the optical properties of the ophthalmic lens and / or an ophthalmic lens required for manufacturing. Department This includes extending the surface to the diameter of the intended lens. Examples of methods for extending the surface are described in EP 2087396.
[0155] The method for determining at least one surface of at least one ophthalmic lens may also include storing surface data of the at least one surface calculated with the surface model and, if applicable, corrected and / or extended. The surface data may optionally be stored together with at least a portion of the order parameters used to determine the surface data. For example, the surface data may be stored on a suitable data medium or storage device. The storage device may also be a computer cloud or a data cloud.
[0156] The method for determining at least one surface of at least one ophthalmic lens may also comprise transmitting surface data of the at least one surface calculated using a surface model, corrected, if applicable, and / or extended to an external unit, e.g. a manufacturer of the ophthalmic lens, a manufacturing unit, a manufacturing device, etc. The surface data may optionally be transmitted together with at least a portion of the order parameter set used to determine the surface data.
[0157] The method for determining at least one surface of at least one ophthalmic lens may also comprise checking at least one surface calculated using the surface model for fulfillment of desired or requested characteristics and storing information regarding fulfillment or non-fulfillment of the requested characteristics together with at least one portion of the order parameters used to determine the surface data and / or at least one surface calculated using the surface model and, if applicable, corrected and / or extended, and / or at least one particular value of the desired or requested characteristics which may be provided as a target value of the characteristics in the definition of the surface model according to the first aspect of the present invention.
[0158] Further, the method for determining at least one surface of at least one ophthalmic lens may comprise fitting model parameters of the surface model after all surfaces have been determined and / or stored, or after a predetermined number of surfaces have been calculated and, if applicable, corrected using the surface model.
[0159] Correction of calculated surfaces and check of design conformity using surface models At least one surface of the ophthalmic lens calculated using the surface model may be further corrected, for example if the ophthalmic lens calculated using the surface model does not fulfill at least one desired or required optical and / or geometric property. For this purpose, the ophthalmic lens calculated using the surface model may be checked for conformity with the desired or required property, for example by being above or below an appropriately selected threshold value.
[0160] For example, assuming a multifocal lens, one or more of the following characteristics may be checked: - residual astigmatism along the primary line of sight (optimally in the position of use), - fulfillment of standard optical effects at the reference points or permissible deviations of optical effects (e.g. spheres and / or cylinders and / or prisms), - the desired or acceptable vertical and / or horizontal prism in at least one prism reference point; - the desired or tolerable misalignment of the vertical and / or horizontal prism at least in one prism reference point between the left and right lenses, - the maximum tolerable gradient of the residual astigmatism and / or refractive error (respectively, optimally in the position of use), - the distribution of the residual astigmatism and / or refractive error (respectively, optimally in the position of use), - the desired width of the zone of good visibility, - Sufficient smoothness of the surface, - manufacturability of the surface (e.g. absence of undercuts), - Absence of surface undercuts to ensure manufacturability; - Sufficient thickness to achieve breaking strength (e.g. quantified using static or dynamic load tests).
[0161] Other or further characteristics may also be checked.
[0162] However, prior to the calculation with the surface model, it may be advantageous to already determine whether a correction of the surface calculated with the surface model is necessary, for example this may be determined using the proportion of the surface that does not meet the required properties (for example if this proportion exceeds a determined threshold value).
[0163] Similarly, it may be advantageous to continuously check each ophthalmic lens generated with the surface model for fulfillment of desired or required characteristics to determine individually for each lens whether correction is necessary or not. Such a check is preferably applied only if the computational power applied to the check, averaged over the ophthalmic lenses to be calculated, is less than the saved computational power of the correction calculation (e.g., post-optimization or post-calculation).
[0164] The correction of the ophthalmic lens calculated using the surface model may be carried out starting from these surfaces, as a result of which the fulfilment of the required properties of the ophthalmic lens is checked independently of whether it is true or not.
[0165] For example, the correction may be performed by methods according to the prior art. For example, it is possible to calculate the surface of the ophthalmic lens in a post-optimization, which includes the optimization of several established optimization methods for ophthalmic lenses. The surface output by the surface model may be used as a starting point (known as the starting surface) for the post-optimization.
[0166] Likewise, it is possible to generate the surface of the ophthalmic lens in a post-calculation by superposition of the surface calculated by the model with one or more superposition surfaces. For example, a simple ellipsoid-annular zone superposition surface or a more complex superposition surface, as described for example in US 2018 / 0088353 A1 or EP 1 240 541 B1, may be used as superposition surface.
[0167] Alternatively, the calculations may be performed using a second surface model, which may include, for example, a regression model.
[0168] Likewise, it may be advantageous to carry out further corrections of surfaces calculated using a surface model and possibly already corrected: if it is known, for example, which systematic deviations of the surface or optical properties of the ophthalmic lens occur in production, these may likewise be corrected with the aid of established methods, for example with the aid of the method described in WO 2014 / 076155 A1.
[0169] Furthermore, the surface or surfaces of the ophthalmic lens may be checked again for fulfillment of the desired or required properties after the correction. If the ophthalmic lens does not fulfill the desired or required properties, the production of the lens may be stopped to manually check the order. This ensures that the production or even delivery of an inappropriate lens is prevented. In particular, this may allow failed post-optimization to be discovered.
[0170] Likewise, it may be advantageous to store information regarding the fulfillment or non-fulfillment of requested characteristics together with the order parameters, for example in a database, for later evaluation.
[0171] When an ophthalmic lens is finally post-optimized after calculations via a surface model, its surface or surfaces together with the order parameters and, if applicable, other variables derived therefrom may be stored as a new data set in the database. This data set may be utilized to improve the surface model in that the surface model is redetermined or redefined under consideration of newly arising data. Thus, the database for the method according to the invention steadily grows. Also, the redefinition of the surface model may be performed using only a portion of the data set stored in the database. Thus, the quality of the surfaces calculated using the surface model may be improved with all the post-optimized lenses.
[0172] The correction of the surface or surfaces calculated using the surface model may be omitted, for example, if the deviation of the order parameter set of the ophthalmic lens to be calculated and / or the variables derived therefrom from the order parameter set necessary to determine the surface model most similar to the order parameter set of the lens to be calculated is less than a predefined threshold. The deviation may be measured using a suitably determined distance measure, which may be selected such that the sensitivity of the surface or surfaces is taken into account by the order parameters. For example, the deviation may be described or quantified using the square of the difference in elevation or desired properties of two ophthalmic lenses having different order parameters.
[0173] Continuous refinement of the surface model It is possible to check and / or modify the model parameters of the surface model continuously or at regular intervals. If an ophthalmic lens has to be post-optimized or newly optimized, the data of the ophthalmic continuous surface required for this may be added and used together with the associated order parameters for the adaptation of the model parameters. For the adaptation, an optimization algorithm that uses only a part of the data, e.g. statistical gradient effect or limited memory BFGS, may preferably be used. However, optimization algorithms that require the complete data set may be used as well.
[0174] The checking and / or adaptation of the model parameters of the surface model may be performed, for example, after every recalculated or reoptimized ophthalmic lens, or after a predefined number of recalculated or reoptimized ophthalmic lenses. In the simplest scenario, this number may be constant.
[0175] Another possibility is to perform the fitting only when a fixed percentage (e.g. 10%) of the data already used for the determination of the model data is to be re-optimized or post-optimized. It may also be proposed to perform a fitting of the model parameters of the surface model when computation time is available (e.g. when several ophthalmic lenses need to be calculated).
[0176] If the model parameters of the surface model are to be adapted in that exclusively new data sets from the posterior optimization are used, it is also possible to choose the learning rate in the adaptation of the model parameters (i.e. the strength of the adaptation of the model parameters in the adaptation step) to be proportional to the ratio of the number of new data sets to the total number of data sets used for training. The learning rate is thus reduced at each adaptation, ensuring the convergence of the model parameters.
[0177] Hybrid computation and optimal choice of complexity for surface models According to one example, a hybrid method for determining at least one of the surfaces of one or more ophthalmic lenses may be provided, the method including determining at least one of the surfaces according to a predefined surface model from order parameters and / or variables derived therefrom, and a subsequent correction or post-calculation of the at least one surface determined using the surface model. The correction may be one of the corrections described above.
[0178] It may be advantageous to adjust the complexity of the surface model such that, on average, the computational cost of the surface of the ophthalmic lens using the surface model is minimized together with the computational cost of the post-computation or correction.
[0179] Z(F_m(a_m)) is the quality of the surface F_m(a_m) calculated with the surface model, this quality being measured with the aid of the objective function Z. Thus, Z(F_n(a_n;F_m(a_m))) is the quality of the surface F_n(a_n;F_m(a_m)) calculated starting from the surface F_m(a_m) by a posteriori calculation having a computational cost a_n. In order to minimize the computational cost, the complexity of the surface model may be chosen such that the improvement of the objective function by the calculation with the surface model is, for each additional computational cost a_m, strictly as much as or greater than the initial improvement of the objective function by a posteriori calculation starting from the surface calculated with the surface model assuming the missing computational cost a_n=0. For this reason, the evaluation of the surface model must have the smallest computational cost a_m, for which dZ(F_m(a_m)) / da_m(a_m)>=dZ(F_n(a_n;F_m(a_m=0))) / da_n applies.
[0180] In this context, note that the derivatives on both sides have negative algebraic signs because the objective function becomes smaller with increasing computational cost.
[0181] In order to keep the model complexity low and thus minimize the computational cost averaged over all ordered ophthalmic lenses, it is possible to restrict the range of order parameters for which the surface model is used to a frequent order range. For example, this may be reasonable if a certain set of order parameters occurs rarely and others occur very frequently. For example, the surface model may be used only for lenses with standard individual parameters in the main order range and standard design (e.g. sphere between -4dpt and +4dpt, cylinder absolute value less than 2dpt, addition between 1.5dpt and 2.5dpt). The prior art method may then be used for the remaining range of order parameters.
[0182] Also, the complexity of the surface model may be limited, for example, so that only the order parameters with the greatest influence on the surface or surfaces of the ophthalmic lens are processed. The influence of further order parameters on the surface may be corrected after the fact using a posteriori calculation or a posteriori optimization. In one possible case, assuming a calculation of a spectacle lens, the order parameters represented in or used to determine the surface model are reduced to prescribed values and centering parameters, together with the pantoscopic tilt, the surface formation angle, and the corneal apex distance of the ophthalmic lens. Thereby, the calculation of the surface or surfaces of the ophthalmic lens remains more efficient than the previously performed optimization based on a standard starting surface.
[0183] In the most extreme case, refraction alone - in the absolute most extreme case, spherical equivalents and, where applicable, additions alone - may be used in the calculations using the surface model.
[0184] It is likewise proposed to determine the extent to which a surface model is used for the calculation with the fulfillment or non-fulfillment of the properties required for the ophthalmic lens. For this purpose, information on the fulfillment or non-fulfillment of these properties may be stored on the basis of order parameters. The extent to which a surface model is used may be determined with the aid of a classification algorithm established, for example, by means of logistic regression or a support vector machine.
[0185] The range of order parameters in which the surface model is used for the calculation of the surface may likewise be continuously extended, for example if a sufficient surface, finally calculated using the methods according to the prior art, is available outside the range of order parameters in which the surface model is used for the calculation, the model parameters of the surface model may be retrained and then the extended range may also be recalculated as proposed above.
[0186] In the most general case, multiple surface models, differing in terms of complexity and / or implementation, may be used depending on the order parameters.
[0187] In particular, assuming individualized or personalized ophthalmic lenses, it may occur that a portion of the order parameters more frequent than average assumes a standard value in the distribution of the order parameters. Assuming a partially individualized ophthalmic lens, some order parameters may be set to a standard value and may not be changed depending on the order. Now, in the case where a portion of the order parameters has a standard value, it is proposed to use a surface model that is reduced in terms of complexity, which allows a simpler and therefore faster calculation, since the deviation of a portion of the order parameters from the standard value does not have to be represented by the surface model. The remaining orders may be calculated using a surface model of higher complexity or identical to the method according to the prior art.
[0188] The deviation of the surface model used in any order parameter range may be determined experimentally. The goal of optimal deviation may be to achieve fast and resource-saving calculation of ophthalmic lenses.
[0189] According to a third aspect of the invention, a computer program product is proposed which, when loaded into the memory of a computer and executed on said computer, has the effect that the computer performs a method according to one of the above aspects, which computer may also be a computer system.
[0190] The method according to any of the above aspects may be implemented by a correspondingly designed device.
[0191] A fourth aspect of the invention relates to a device for determining a surface model for calculating at least one surface of at least one ophthalmic lens (e.g. a contact lens or a spectacle lens) from at least one set of order parameters for the ophthalmic lens and / or from variables dependent on the order parameters, the device comprising a computing device designed to implement the method according to the first aspect of the invention.
[0192] The device for determining a surface model comprises in particular: a device for providing a training data set including a plurality of order parameter sets, each of the order parameter sets including values for at least a portion of parameters required to order at least one ophthalmic lens; a device for providing, for each of the order parameter sets in the training data set, at least one target value for at least one predetermined characteristic of at least one ophthalmic lens; - a device for providing at least one surface model parameterized by model parameters, with which - for given values of the model parameters - at least one surface of at least one ophthalmic lens can be calculated at least from an order parameter set and / or from variables dependent on the order parameter set; and a computing device designed to obtain or determine a surface model for the calculation of at least one surface of at least one ophthalmic lens, wherein the obtaining or determining of the surface model comprises: Determining optimized values of model parameters of the at least one surface model using the provided target values.
[0193] Determining optimized values of model parameters of at least one surface model may, for example, include: optimizing values of the model parameters of the at least one surface model for the purpose of minimizing or maximizing an objective function for the model parameters of the at least one surface model based on the model parameters and the provided target values, wherein the objective function of the model parameters for each of the order parameter sets includes at least one term that assumes a minimum or maximum when a provided target value of the at least one characteristic of the at least one ophthalmic lens for the respective order parameter set matches a value of the same characteristic of the at least one lens that is computable or calculated using the surface model for a given value of the model parameters of the surface model for the corresponding order parameter set.
[0194] A fifth aspect of the present invention relates to a device for determining at least one of the surfaces of one or more ophthalmic lenses by means of a surface model previously determined from order parameters and / or from variables that depend on the order parameters. The device is designed to implement a method for determining at least one of the surfaces of at least one ophthalmic lens according to any of the above aspects. The device for determining at least one surface of an ophthalmic lens is in particular a device for providing an order parameter set for at least one ophthalmic lens; a device for providing a surface model for calculating at least one surface of at least one ophthalmic lens from a set of order parameters for an ophthalmic lens and / or from variables dependent on the order parameters; a computing device configured to perform a determination of surface data of at least one surface of at least one ophthalmic lens with the aid of a surface model from the provided set of order parameters; Equipped with.
[0195] The surface model may be a surface model determined or verified according to a method according to any of the aspects described above.
[0196] The above-mentioned preferred embodiment variations or advantages described above also apply pertinently with respect to the above-mentioned devices.
[0197] A sixth aspect of the present invention relates to a data set comprising surface data of at least one surface of at least one ophthalmic lens, the at least one surface having been determined according to the method for determining at least one of the surfaces of at least one ophthalmic lens according to any of the above aspects. The data set may be or may be stored permanently or non-permanently in a suitable data medium or storage device, such as a database, a computer cloud or a data cloud, etc.
[0198] The above-described method and corresponding device for determining at least one of the surfaces of one or more ophthalmic lenses by means of a surface model may be used in the manufacture of ophthalmic lenses, for example in the process of designing ophthalmic lenses, to determine the surfaces to be manufactured, or to check the manufacturability of ophthalmic lenses using the geometrical characteristics of the surfaces. In other words, the method for determining at least one of the surfaces of one or more ophthalmic lenses may be a component of a manufacturing or fabrication method of ophthalmic lenses.
[0199] For example, it is also possible to use the method for determining at least one of the surfaces of one or more ophthalmic lenses by means of a surface model in a consultation (e.g. with an optician) in order to clarify the optical and geometric properties of an individualized or personalized ophthalmic lens (e.g. the location of the clear vision zone and the thickness assuming a multifocal lens) for a future wearer of such a lens. A further aspect of the present invention therefore relates to a method and a corresponding device for manufacturing an ophthalmic lens.
[0200] The method for manufacturing an ophthalmic lens may in particular include: Determining at least one surface of at least one ophthalmic lens according to a method according to any of the aspects described above; Producing an ophthalmic lens having at least one surface.
[0201] The device for producing ophthalmic lenses includes in particular: A device for determining at least one surface of at least one ophthalmic lens according to any of the aspects described above; A manufacturing device for producing an ophthalmic lens having at least one surface.
[0202] The variants of the preferred embodiment described above or the advantages described above also apply pertinently with respect to this method or this device.
[0203] The above-mentioned devices for providing, determining, deciding or calculating data (e.g. variables derived from order parameters, model parameters, target values, surface data, weights, etc.) and / or evaluating functions, e.g. target functions, may be realized by suitably configured or programmed data processing devices (in particular dedicated hardware modules, computers or computer systems, e.g. computer clouds or data clouds) together with corresponding calculation units, electronic interfaces, storage and data transmission units. The devices may also include at least one interactive graphical user interface (GUI) allowing a user to view and / or input and / or modify data.
[0204] The above mentioned devices may also have suitable interfaces allowing the transmission or input or reading of data (e.g. order parameter sets, model parameters, target values, surface data, etc.) Likewise, the devices may include at least one storage unit, e.g. in the form of a database, for storing the data used, e.g. order parameter sets, target values, surface data, weights, etc.
[0205] For example, the manufacturing device may include at least one CNC machine for direct processing of blanks according to the determined optimization specifications. Alternatively, the ophthalmic lens may be manufactured by a casting method. The finished processed ophthalmic lens preferably has a simple spherical or rotationally symmetric aspheric surface and a surface determined according to any aspect of the method according to the invention with the aid of a surface model. The simple spherical or rotationally symmetric aspheric surface is preferably the front surface (i.e. the surface on the object side) of the ophthalmic lens. However, it is trivially possible to arrange a surface calculated with the aid of a surface model as the front surface of the ophthalmic lens. Also, both the surfaces of the ophthalmic lens and / or their arrangement relative to each other may be determined with the aid of a surface model.
[0206] Furthermore, the present invention provides the use of an ophthalmic lens manufactured according to the manufacturing method according to the invention, in a determined predetermined average or ideal use position of the spectacle lens in front of the wearer's eye, for the correction of the wearer's refractive error.
[0207] In the following, preferred embodiments of the present invention are described by way of example with the aid of the accompanying drawings. The individual elements of the described embodiments are not limited to the respective embodiments. Rather, the elements of the embodiments may be combined with each other in any combination, thereby creating new embodiments. [Brief description of the drawings]
[0208] [Figure 1] FIG. 1 illustrates an example of a method for calculating an ophthalmic lens with the aid of a surface model. [Diagram 2] FIG. 1 illustrates an example of a method for calculating an ophthalmic lens with the aid of a parameterized surface model. [Diagram 3] FIG. 13 illustrates a further example of a method for calculating an ophthalmic lens with the aid and correction of a surface model. [Figure 4] FIG. 13 illustrates a further example of a method for calculating an ophthalmic lens with the aid of a surface model and optional corrections. [Diagram 5] FIG. 2 shows an example of a method for determining a surface model with the aid of an already calculated ophthalmic lens. [Figure 6] FIG. 1 illustrates an example of a method for determining a surface model without a previously calculated ophthalmic lens. [Figure 7] FIG. 1 illustrates an example of splitting a data set, including multiple order parameter data sets, into a training data set, a validation data set, and a test data set. [Figure 8A] FIG. 2 shows the correlation between the central thickness of a spectacle lens calculated with the aid of a first exemplary surface model and the central thickness of a test spectacle lens. [Figure 8B] FIG. 13 shows a histogram of the frequency of the center thickness residual. [Figure 8C] FIG. 2 shows the correlation between the rear curvature of the first principal cross section of a spectacle lens calculated with the aid of a first surface model and the rear curvature of the first principal cross section of a test spectacle lens. [Figure 8D] FIG. 13 shows a histogram of the frequency of the residual of the posterior curvature of the first principal cross section. [Figure 8E] FIG. 2 shows the correlation between the rear curvature of the second principal cross section of a spectacle lens calculated with the aid of a first surface model and the rear curvature of the second principal cross section of a test spectacle lens. [Figure 8F] FIG. 13 shows a histogram of the frequency of the residual of the posterior curvature of the second principal cross section. [Figure 8G] FIG. 2 shows the correlation between the vertex power in a first principal cross section of a spectacle lens calculated with the aid of a first surface model and the vertex power in a first principal cross section of a test spectacle lens. [Figure 8H] FIG. 3 shows a histogram of the frequency of deviation of the vertex power in the first principal cross section, calculated according to a first surface model, from the vertex power in the first principal cross section of a test spectacle lens. [Figure 8I] FIG. 2 shows the correlation between the vertex power in the second principal cross section of a spectacle lens calculated with the aid of a first surface model and the vertex power in the second principal cross section of a test spectacle lens. [Figure 8J] FIG. 4 shows a histogram of the frequency of deviation of the vertex power in the second principal cross section, calculated according to a first surface model, from the vertex power in the second principal cross section of a test spectacle lens. [Figure 9A] FIG. 13 shows the correlation between the central thickness of a spectacle lens calculated with the aid of a second exemplary surface model and the central thickness of a test spectacle lens. [Figure 9B] FIG. 13 shows a histogram of the frequency of the center thickness residual. [Figure 9C] FIG. 2 shows the correlation between the rear curvature of the first principal cross section of a spectacle lens calculated with the aid of a second surface model and the rear curvature of the first principal cross section of a test spectacle lens. [Figure 9D]FIG. 13 shows a histogram of the frequency of the residual of the posterior curvature of the first principal cross section. [Figure 9E] FIG. 2 shows the correlation between the rear curvature of the second principal cross section of a spectacle lens calculated with the aid of a second surface model and the rear curvature of the second principal cross section of a test spectacle lens. [Figure 9F] FIG. 13 shows a histogram of the frequency of the residual of the posterior curvature of the second principal cross section. [Figure 9G] FIG. 2 shows the correlation between the vertex power in the first principal cross section of a spectacle lens calculated with the aid of a second surface model and the vertex power in the first principal cross section of a test spectacle lens. [Figure 9H] FIG. 4 shows a histogram of the frequency of deviation of the vertex power in the first principal cross section, calculated according to a second surface model, from the vertex power in the first principal cross section of a test spectacle lens. [Figure 9I] FIG. 4 shows the correlation between the vertex power in the second principal cross section of a spectacle lens calculated with the aid of a second surface model and the vertex power in the second principal cross section of a test spectacle lens. [Figure 9J] FIG. 4 shows a histogram of the frequency of deviation of the vertex power in the second principal cross section, calculated according to a second surface model, from the vertex power in the second principal cross section of the test spectacle lens. [Figure 10] FIG. 13 shows the compatibility of the minimum edge thickness and center thickness of a spectacle lens calculated with the aid of a second surface model. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0209] Order Parameter Set i About Ophthalmic Lenses i Conventional methods for calculating Order Data i Provide an order parameter set, Calculating or optimizing at least one of the surfaces of an ophthalmic lens; and Order Data iOphthalmic lenses manufactured with respect to i Obtain the surface of.
[0210] The optimization is typically performed iteratively by minimizing or maximizing an objective function, to which a nominal value of at least one characteristic (e.g. optical characteristic) of the lens is input. The objective function is typically evaluated for a determined parameterization of the surface to be calculated. The parameters of the surface are varied until a given criterion is met.
[0211] Figure 1 shows an example of a method for calculating an ophthalmic lens or a pair of ophthalmic lenses with respect to order data with the aid of a surface model and a direct non-iterative calculation or a calculation having several iterative steps. The method comprises the following steps: S1-1: Order parameter set for one or more ophthalmic lenses k providing order data including: S1-2: Calculating / optimizing at least one surface of the lens with the aid of a surface model; S1-3: Order parameter set k Ophthalmic lenses manufactured with respect to k A step of obtaining a surface of the
[0212] The calculation of at least one surface of the lens with the aid of the surface model may be performed directly, non-iteratively or with several iterative steps in an iterative manner, thereby significantly reducing the time required for the calculation of the surface.
[0213] The surface model may be a model determined according to one of the previously described aspects and variants of the embodiment. For example, the surface model may be determined parametrically, and model parameters (parameters of a parametric representation of the surface model) together with at least a portion of the order parameters and / or variables derived therefrom are used to calculate the surface or surfaces of the ophthalmic lens. The surface model may be a linear or non-linear regression model. The non-linear regression model may be, for example, a neural network. In this regard, reference is made to the variants of the preferred embodiment described above, or to the above-mentioned advantages of the different surface models.
[0214] Figure 2 shows an example of a method for calculating an ophthalmic lens or a pair of ophthalmic lenses with respect to order data with the aid of a parametric surface model and by direct calculation. The method comprises the following steps: S2-1: Order parameter set for one or more ophthalmic lenses k providing order data including: S2-2: Providing parameters of a surface model; S2-3: Calculating / optimizing the ophthalmic lens with the aid of a surface model (either direct, non-iterative or involving several iterative steps by iterative methods); S2-4: Order parameter set k Ophthalmic lenses manufactured with respect to k A step of obtaining a surface of the
[0215] The surface model may be determined using an existing set of order parameters with associated target values. For this purpose, an initial complexity and an initial parameterization may be determined or defined. The model parameters may then be determined by an optimization method, in which the model parameters are iteratively varied. The goal of the optimization method is that the surface output from the surface model for different sets of order parameters and / or the properties of this surface optimally correspond well to the target values for the same set of order parameters.
[0216] The optimization of the parameterization and, where applicable, the complexity of the surface model may be performed by minimizing or maximizing an objective function of the model parameters as explained above, the objective function being preferably evaluated over all order parameter sets in the training data set, the objective function including at least one term depending on the deviation, determined for each order parameter set in the training data set, of one or more values of at least one predetermined characteristic of the ophthalmic lens calculated according to the surface model from at least one target value of this characteristic for the same order parameter set.
[0217] In one example, the objective function for the model parameters may include the following terms:
[0218]
number
[0219] A j-th value of at least one property Z of the lens may be determined using a current parameterization of the surface model or current model parameters. The j-th value of the at least one property Z of the lens may, for example, be a value of this property at a j-th evaluation point of the lens.
[0220] Function f i =f(Z i , Z Nominal) may also include additional terms that rely on other or more characteristics of the ophthalmic lens.
[0221] Function f i may be, for example, an objective function used in the optimization of ophthalmic lenses by conventional methods and evaluated for the current parameterization or current parameters of the surface model.
[0222] One or more target values Z i (j) Nominal may be equal to 0. Thus, for example, the astigmatism in the use position at one or more evaluation points of a lens may have a target value of 0 dpt.
[0223] The above function f on the surface i and / or its derivatives may then be evaluated across all order parameter sets in the training data set, with the evaluation occurring depending on the parameterization of the surface model. For example, the function f i and / or from its derivatives, a weighted or unweighted sum f may be calculated and evaluated.
[0224]
number
[0225] If as a result of the evaluation of the objective function f the predefined criterion is still not met, the model parameters are modified and the objective function f is re-evaluated, this is repeated iteratively until the predefined criterion is met.
[0226] The surface model with the model parameters so determined may be suitably stored and used as described above for the calculation of new ophthalmic lenses.
[0227] Figure 3 shows a further example of a method for calculating an ophthalmic lens or a pair of ophthalmic lenses with respect to order data with the aid of a surface model and direct calculation. This method is similar to the method shown in Figure 1 and further comprises a correction of the surface calculated with the aid of the surface model. The correction of the surface can be one of the corrections described above. The method comprises the following steps: S3-1: Order parameter set for one or more ophthalmic lenses k providing order data including: S3-2: Calculating / optimizing the ophthalmic lens with the aid of a surface model (direct, non-iterative); S3-3: Correcting the calculated surface using a surface model (post-calculation / post-optimization); S3-4: Order parameter set k Ophthalmic lenses manufactured with respect to k A step of obtaining a surface of the
[0228] The correction of the surface or surfaces calculated using the surface model may be one of the corrections described above. Due to the optimal starting surface for the post-calculation or post-optimization, such a correction usually requires only one or a few iterations. Thereby, the total calculation time may be significantly reduced.
[0229] 4 shows a further example of a method for calculating an ophthalmic lens or a pair of ophthalmic lenses with respect to order data with the aid of a surface model and by direct calculation as well as optional corrections. The method comprises the following steps: S4-1: Multiple order parameter sets for one or more ophthalmic lenses k providing order data including: S4-2: Calculating / optimizing the ophthalmic lens with the aid of a surface model (either direct, preferably non-iterative, or with several iterative steps); S4-3: Checking whether correction is necessary; S4-4: If a correction is necessary, carry out a correction of the calculated surface using the surface model (post-calculation / post-optimization with only one or a small number of iterations); S4-5: Order parameter set k Ophthalmic lenses manufactured with respect to k A step of obtaining a surface of the
[0230] 5 shows an example of a method for determining a surface model with the aid of an already calculated ophthalmic lens. The method comprises the following steps: S5-1: Multiple order parameter sets {d i}, and a plurality of ophthalmic lenses L i An order data set {d i ,L i} and splitting the data set into a training data set (and, if applicable, a validation data set) and a test data set. i is a lens calculated for a set of order parameters in an order data set according to known computational or optimization methods. The lens provided is a function of the objective function f i and S5-2: Providing an initial complexity and parameterization of the surface model; S5-3: Parameterization and, if applicable, optimization of the complexity of the surface model (iteratively) with the goal that the surface model optimally reflects the training data set (and, if applicable, the validation data set). The optimization is performed using an objective function G for the model parameters, whereby the surface of the ophthalmic lens BL i Or their surfaces are order data d iThe surface or lens calculated using the surface model is calculated using the objective function G for the provided lens L i For example, in the optimization of the model parameters, the objective function G(BL i ,L i ) is minimized over i. S5-4: Testing the optimization parameters and, if applicable, the optimized complexity of the surface model using a recreation of the test data set; S5-5: Obtaining optimized parameters of the surface model and, if applicable, an optimized complexity of the surface model in order to provide a direct calculation of the ophthalmic lens from the order data.
[0231] The above method may also be performed using measured surfaces and / or surface spacings of already manufactured lenses instead of calculated surfaces.
[0232] FIG. 6 shows an example of a method for determining a surface model without a previously calculated ophthalmic lens. The method comprises the following steps: S6-1: Order data set for multiple hypothetical ophthalmic lenses i and splitting the data set into a training data set (and, if applicable, a validation data set) and a test data set; S6-2: Providing an initial complexity and parameterization of the surface model; S6-3: Lens L calculated from surface model i is the objective function f for the optimization of individual lenses. i Optimizing the parameterization, and, where applicable, the complexity of the surface model, with the goal of minimizing the cumulative sum over the training data set (and validation data set, where applicable) of the objective function f i may be an objective function known from the prior art, S6-4: For the optimization of individual lenses according to the prior art, the objective function f i testing the optimization parameters and, if applicable, the optimized complexity of the surface model using a cumulative sum of S6-5: Obtaining optimized parameters of the surface model and, if applicable, an optimized complexity of the surface model in order to provide a direct calculation of the ophthalmic lens from the order data.
[0233] In the above example, the order parameter set d k may each include one or more order parameters required for a single ophthalmic lens or a pair of ophthalmic lenses. Examples of order parameters can be found in standards established for spectacle lenses (see, for example, EU guideline 93 / 42 / EWG on medicinal products). For further examples of order parameters and variables derived therefrom, as well as for further details, please refer to the above descriptions in the corresponding sections. All features, embodiment variations and / or advantages described therein apply duly to the above examples.
[0234] In the following, the calculation of ophthalmic lenses (spectacles lenses) is explained in detail using two further examples.
[0235] Example of lens calculation using regression model The first example concerns a glass calculation by means of a regression model. In this example, the back surface curvature and the central thickness of a monofocal lens are calculated directly from the spherical and cylindrical order values of refraction with the aid of a surface model designed as a regression model. The parameters and the complexity of the regression model are determined starting from the data of an already calculated ophthalmic lens. The glass diameter is given in this example at 65 mm.
[0236] Complex iterative algorithms are conventionally used to calculate the center thickness. In this example, the conventional iterative algorithms are replaced with a regression model that is evaluated without iterations.
[0237] The starting point for the calculations is a data set with a total of 825 lenses already calculated according to the method from the prior art, the sphere and the cylinder of which are varied in steps of 0.25 dpt. As shown in Fig. 7, the data set is subdivided into a training data set with 425 lenses, a validation data set with 192 lenses and a test data set with 208 lenses. Since the data form an equidistant grid on the sphere and the cylinder, the subdivision is not random, as is common, following a determined pattern (see Fig. 7). However, it is possible to randomly subdivide the initial data set into a training data set, a validation data set and a test data set.
[0238] The calculation of the curvature of the front surface KVFL is first performed with the aid of a table in which the basic curves are tabulated on the basis of spheres, cylinders. Naturally, the curvature of the front surface may also be calculated in more complex surface models using the classification models of spheres, cylinders and, if applicable, the refractive index of the material and, if applicable, the information on the glass blank available for production. However, for the sake of clarity, this has been intentionally ignored in this example.
[0239] The determination of the curvatures K1 and K2 of the principal sections of refraction is carried out first with the largest or smallest curvature (assuming the positive cylindrical convention). K1=sphere + cylinder and K2=sphere
[0240] The regression model is then determined, which in this example is designed with the aid of a spline function. A cubic spline is used in K1, K2 or KVFL, respectively, with an additional linear interaction term. The node points of the spline are equally distributed within the value ranges of K1, K2 and KVFL, respectively, meaning that the nodes form a grid with identical spacing in the respective parameters K1, K2 and KVFL. The spline coefficients represent the model parameters.
[0241] The sum of squared residuals (i.e., the difference between the calculated variables minus the corresponding values in the data set) was used as a measure of how well the model described the existing data. The quadratic deviation of the central thickness in mm, as well as both curvatures of the posterior surface in dpt, were minimized in separate fits.
[0242] To fit the training data set, the complexity of the models was varied in that the number of spline node points was varied. The model with the minimum sum of squared errors, calculated starting from the validation data set, was finally selected. The models used for the central thickness can be summarized as follows (see Table 1):
[0243] [Table 1]
[0244] The process may proceed similarly for the posterior surface curvatures. Finally, splines in K1, K2 and KVFL, each with three nodal points (within the same value range), were selected such that each curvature of the posterior surface was fitted using a model with 15 parameters each. (1(constant) + 2(K1) + 2(K2) + 2(KVFL) + 2^3(mixture term) = 15 parameters in total)
[0245] In this example, the posterior surface curvature is given in diopters for a refractive index of 1.525.
[0246] The monofocal lenses calculated here are fully described by the material refractive index (here 1.668), diameter (here 65mm), center thickness, and front and back surface curvatures (the latter since these lenses have a spherical front surface and a sphere-torus rear surface). In this case, there is also no tilt of the back surface relative to the front surface since in this example prism=0dpt is given and there is centration according to the eye's center of rotation requirement.
[0247] The model parameters included in Table 2 were determined by minimizing the deviation of the actual central thickness from the central thickness calculated by the spline (model parameters in mm) using the training data set. The elementary spline functions are numbered below with a multi-index, where 0 means that the spline function for the order parameter or derived variable is always equal to 1, and the higher indexes correspond to cubic elementary spline functions that assume, in ascending order, a space from close to the lower edge to the upper edge in the value range of the respective order parameter or derived variable.
[0248] [Table 2]
[0249] The following model parameters in dpt result analogously from the fit of the first principal cross section of the posterior curvature (see Table 3):
[0250] [Table 3]
[0251] The following model parameters in dpt result analogously from the fit of the second principal section of the posterior curvature (see Table 4):
[0252] [Table 4]
[0253] In the following, using spectacle lenses from a test data set (test spectacle lenses), it is shown that the method according to the invention leads to spectacle lenses possessing almost identical properties compared to the method according to the prior art. For this purpose, the central thickness, the back surface curvature and the vertex power of the spectacle lenses are plotted against each other or histograms of the deviations of these variables are calculated. Figures 8A to 8J show the corresponding results.
[0254] Figure 8A shows the correlation between the central thickness of the spectacle lens calculated with the aid of the surface model and the measured central thickness of the test spectacle lens. The central thickness (in mm) measured with the test data or with the test spectacle lens is plotted on the abscissa of Figure 8A, and the central thickness (in mm) calculated according to the model is plotted on the ordinate. Figure 8B shows the histogram of the frequency of the residual of the central thickness (in mm), which means the deviation of the calculated central thickness of the test lens from the measured central thickness of the test lens.
[0255] FIG. 8C shows the correlation between the rear curvature of the first principal cross section of the spectacle lens calculated with the aid of a surface model (curvature of the first principal cross section of the rear surface or rear curvature 1) and the measured rear curvature of the first principal cross section of the test spectacle lens. The rear curvature 1 (in dpt) measured with the test data or with the test spectacle lens is plotted on the abscissa in FIG. 8C. The rear curvature 1 (in dpt) calculated according to the surface model is plotted on the ordinate. FIG. 8D shows a histogram of the frequency of the residuals of rear curvature 1 (in dpt), i.e. the deviation of the calculated rear curvature 1 from the measured rear curvature 1.
[0256] FIG. 8E shows the correlation between the posterior curvature of the second principal cross section of the spectacle lens calculated with the aid of a surface model (curvature of the second principal cross section of the posterior surface or posterior curvature 1) and the measured posterior curvature of the first principal cross section of the test spectacle lens. The posterior curvature 2 (in dpt) measured with the test data or with the test spectacle lens is plotted on the abscissa in FIG. 8E. The posterior curvature 2 (in dpt) calculated according to the surface model is plotted on the ordinate. FIG. 8F shows a histogram of the frequency of the residuals of the posterior curvature 2 (in dpt), i.e. the deviation of the calculated posterior curvature 2 from the measured posterior curvature 2.
[0257] Figure 8G shows the correlation between the apex power in the first principal section of the spectacle lens calculated with the aid of a surface model (apex power in principal section 1) and the measured apex power in the first principal section of the test spectacle lens. The apex power in principal section 1 measured with the test data or with the test spectacle lens (in dpt) is plotted on the abscissa of Figure 8G, and the apex power in principal section 1 calculated according to the model (in dpt) is plotted on the ordinate. Figure 8H shows a histogram of the frequency of deviation of the apex power in principal section 1 calculated according to the model from the measured apex power in principal section 1 (in dpt).
[0258] Figure 8I shows the correlation between the apex power in the second principal section of the spectacle lens calculated with the aid of a surface model (apex power in the second principal section 2) and the measured apex power in the first principal section of the test spectacle lens. The apex power in the principal section 2 measured with the test data or with the test spectacle lens (in dpt) is plotted on the abscissa of Figure 8I, and the apex power in the principal section 2 calculated according to the model (in dpt) is plotted on the ordinate. Figure 8J shows a histogram of the frequency of deviation (difference) of the apex power in the principal section 2 calculated according to the model from the measured apex power in the principal section 2 (in dpt).
[0259] To calculate the central thickness and the two posterior curvatures, given arbitrary values of the sphere sph and the cylinder cyl, basic spline functions were evaluated at corresponding points in KVFL(sph,cyl), K1(sph,cyl), K2(sph,cyl), respectively, with the parameters generated from the fit (see table) and the sum calculated therefrom, without the need for iterative calculations.
[0260] Obviously, the surface model proposed here can also be extended by the order parameters of the prism and the prism base, in that, for example, the surface model is extended by the prism itself and, as a derived parameter, by the angle between the prism base and the axis of astigmatism.
[0261] Example for determining a regression model without using pre-calculated data In a departure from the conventional examples, in the following, the surface model designed as a regression model is determined without having to rely on data of already calculated ophthalmic lenses.Instead of these, the parameters of the surface model are calculated directly by minimizing the sum of the objective functions used according to the methods known from the prior art for the (iterative) calculation of multiple lenses.The surface model determined in such a way is therefore capable of carrying out the calculation of a single-vision lens with any effect exclusively specified by one or more objective functions and the desired effect and diameter.
[0262] Spherical front curvature KVFL i and diameter D i The vertex power is two principal sections K1 i and K2 i An example of a prior art objective function for a single lens i characterized by:
[0263]
number
[0264] The center thickness is d M,i and the order values of the two principal sections are
[0265]
number
[0266]
number
[0267]
number
[0268]
number
[0269]
number
[0270]
number
[0271]
number
[0272]
number
[0273]
number
[0274]
number
[0275]
number
[0276]
number
[0277] where r VFL,i , r RFL1,i and r RFL1,i are the radii of curvature at the front or rear faces at their two main cross sections, and p is a factor that penalizes undercutting the glass thickness (where p=100 was used).
[0278] In this example, the minimum allowable center and edge thicknesses of all lenses are constant (
[0279]
number
[0280]
number
[0281] Obviously, other objective functions, whether alone or with additional terms, may also be used, as long as they embody the desired properties of the spectacle lens, such as, for example, the desired distribution of refractive error or of undesired astigmatism in the position of use.
[0282] The objective function for the optimization of the parameters of the surface model is constructed from the sum of the objective functions of the individual lenses, calculated over all lenses i from the respective data sets (i.e., training, validation or testing data sets), where the central thickness and curvature of the posterior surface depend parametrically on the parameters of the surface model:
[0283]
number
[0284] As a result, θ=(θ dM ,θ K1 ,θ K2 ) specifies the parameters of the surface model, where the parameters may be divided into three separate parameter sets: the central thickness of the posterior surface, and the spline coefficients of the two principal curvatures. The values of the anterior surface curvature, lens diameter, nominal values of the principal curvatures of the vertex power, and the minimum central and edge thicknesses, which are predetermined for each lens, are collectively
[0285]
number
[0286] Here, values calculated with the aid of the surface model are used for the central thickness, and for the two posterior curvatures, the values for which depend on the current parameters of the surface model:
[0287]
number
[0288] The data set used for training, validation and testing of the trained regression model consists of the same values as in the conventional example for spheres and cylinders (see Fig. 7). The same spline-based regression model is also used, but this one initially has as a starting point for the optimization a set of parameters corresponding to lenses with a central thickness of 2 mm and a posterior curvature of -5 dpt each (independent of their anterior curvature) (i.e., only parameters corresponding to constants, thus parameters with multi-index (Spline(K1)Nr, Spline(K2)Nr, Spline(KVFL)Nr) = (0,0,0), are populated with 2 mm or -5 dpt in each different fit of KVFL, K1 and K2, all other parameters are 0).
[0289] To determine the optimal parameters of the surface model, here the Nelda-Mead algorithm (with 20000 function evaluations) was used first, since it is relatively robust and does not require deviations regarding the parameters. Then, an optimization with 200 iterations was performed with the BFGS algorithm (Broiden-Fletcher-Goldfarb-Shanno algorithm), since this algorithm converges more quickly at local optima. The gradients of the latter algorithm were calculated numerically, but they could also be specified analytically, again accelerating the optimization in this case. The number of iterations may be accelerated by a more appropriate choice of the starting point of the optimization. For example, if a new surface model is to be determined, whose objective functions for the individual lenses differ slightly from those of the first surface model (e.g. in the minimum thickness, the weighting of the reciprocal of the terms, or additional terms), it may be useful for the parameters of the already determined surface model to be used.
[0290] The following model parameters θ dM was obtained for the central thickness (in mm) (see Table 5):
[0291] [Table 5]
[0292] Please refer to the previous example for individual specifications.
[0293] For the first principal plane of the posterior curvature, the following model parameters θ K1 Obtained (in dpt):
[0294] [Table 6]
[0295] For the second principal plane of the posterior curvature, the following model parameters θ K2 Obtained (in dpt):
[0296] [Table 7]
[0297] In the following, using ophthalmic lenses from a test data set, it is shown that the method according to the invention leads to ophthalmic lenses possessing almost identical properties compared to the method according to the prior art. For this purpose, the central thickness, the back surface curvature and the apical power of the lens are plotted against each other or a histogram of the deviation of these variables is calculated. The minimum central thickness and the maximum edge thickness are fitted as well. Figures 9A to 9J show the corresponding results.
[0298] In particular, Fig. 9A shows the correlation between the central thickness calculated according to the surface model and the central thickness of a spectacle lens according to the prior art (test spectacle lens). The central thickness (in mm) of the spectacle lens according to the prior art is plotted on the abscissa of Fig. 9A, and the central thickness (in mm) calculated according to the model is plotted on the ordinate. Fig. 9B shows a histogram of the frequency of the residual of the central thickness (in mm), which means the deviation of the central thickness calculated according to the surface model from the central thickness of the spectacle lens according to the prior art.
[0299] Fig. 9C shows the correlation between the rear curvature (rear curvature 1) of the first principal section calculated according to the surface model and the rear curvature 1 of a spectacle lens according to the prior art (test spectacle lens). The rear curvature 1 (in dpt) of the spectacle lens according to the prior art is plotted on the abscissa in Fig. 9C, the rear curvature 1 (in dpt) calculated according to the model is plotted on the ordinate. Fig. 9D shows a histogram of the frequency of the residual of the rear curvature 1 (in dpt), i.e. the deviation of the calculated rear curvature 1 of the spectacle lens calculated according to the model from the rear curvature 1 of the spectacle lens according to the prior art.
[0300] Fig. 9E shows the correlation between the rear curvature of the second principal section (rear curvature 2) calculated according to the surface model and the rear curvature 2 of a spectacle lens according to the prior art (test spectacle lens). The rear curvature 2 (in dpt) of the spectacle lens according to the prior art is plotted on the abscissa in Fig. 9E, the rear curvature 2 (in dpt) calculated according to the surface model is plotted on the ordinate. Fig. 9F shows a histogram of the frequency of the residuals of the rear curvature 2 (in dpt), i.e. the deviation of the calculated rear curvature 2 of the spectacle lens calculated according to the model from the rear curvature 2 of the spectacle lens according to the prior art.
[0301] Fig. 9G shows the correlation between the vertex power calculated according to the surface model in the first principal cross section (vertex power in principal cross section 1) and the vertex power in the first principal cross section of a spectacle lens according to the prior art (test spectacle lens). The vertex power in principal cross section 1 of the spectacle lens according to the prior art (in dpt) is plotted on the abscissa of Fig. 9G, and the vertex power in principal cross section 1 calculated according to the surface model (in dpt) is plotted on the ordinate. Fig. 9H shows a histogram of the frequency of deviation (difference) of the vertex power in principal cross section 1 calculated according to the model from the vertex power in principal cross section 1 of the spectacle lens according to the prior art (in dpt).
[0302] Fig. 9I shows the correlation between the vertex power in the second principal cross section (vertex power in principal cross section 2) calculated according to the surface model and the vertex power in the second principal cross section of a spectacle lens (test spectacle lens). The vertex power in the principal cross section 2 of the spectacle lens according to the prior art (in dpt) is plotted on the abscissa of Fig. 9I and the vertex power in the principal cross section 2 calculated according to the surface model (in dpt) is plotted on the ordinate. Fig. 9J shows a histogram of the frequency of deviation of the vertex power in the principal cross section 2 calculated according to the model from the vertex power in the principal cross section 2 of the spectacle lens according to the prior art (in dpt).
[0303] Figure 10 shows the fit with minimum edge and center thickness. Center thickness (in mm) is plotted on the abscissa of Figure 10 and edge thickness (in mm) is plotted on the ordinate.
[0304] Given arbitrary values of the sphere sph and the cylinder cyl, the calculation of the central thickness and the two posterior curvatures is carried out as in the first example, with the aid of a surface model.
[0305] In the above, preferred embodiment variants of the present invention have been described by way of examples. The individual elements of the described embodiment variants are not limited to the respective embodiment variants. Rather, the elements of the embodiment variants may be combined with each other in any way, thereby creating new embodiment variants. Also, individual features may be modified. For this purpose, other suitable functions, for example polynomial functions, may also be used instead of spline functions to determine the surface model. Similarly, the number of model coefficients or model parameters (e.g. spline coefficients) may be modified. Furthermore, other representations of the surface to be calculated, other order parameter sets, target values, target functions and / or optimization methods may be used.
Claims
1. 1. A computer-implemented method for determining a surface model for calculating at least one surface of at least one ophthalmic lens from at least a set of order parameters for said at least one ophthalmic lens and / or from variables dependent on said order parameters, said computer-implemented method comprising: providing a training data set including a plurality of order parameter sets, each order parameter set including values for at least a portion of parameters required to order at least one ophthalmic lens; providing at least one target value for at least one characteristic of the at least one ophthalmic lens for each of said order parameter sets in said training data set; - providing at least one surface model parameterized by model parameters, with which - for given values of said model parameters - at least one surface of at least one ophthalmic lens can be calculated at least from a set of order parameters and / or from variables dependent on the set of order parameters; Obtaining the surface model for the calculation of at least one surface of at least one ophthalmic lens, determining optimized values for said model parameters of said at least one surface model using said provided target values; Including, The determining of optimized values for the model parameters of the at least one surface model comprises: optimizing values of the model parameters of the at least one surface model with the aim of minimizing or maximizing an objective function of the model parameters of the at least one surface model, the objective function depending at least on the model parameters and the provided target values, the objective function of the model parameters for each of the order parameter sets including at least one term that assumes a minimum or maximum when the provided target value of the at least one characteristic of the at least one ophthalmic lens for the respective order parameter set matches a value of the same characteristic of at least one lens that is computable or calculated using the surface model for a given value of the model parameters of the surface model for the corresponding order parameter set; Said calculation of at least one surface of at least one ophthalmic lens comprises: - calculating at least one parameter of a parametric function describing said surface from said set of order parameters for said at least one ophthalmic lens and / or from variables dependent on said order parameters; or calculating vertex depths of the surface at a plurality of raster points from the set of order parameters for the at least one ophthalmic lens and / or from variables dependent on the order parameters; 4. A computer-implemented method comprising:
2. The providing of at least one surface model parameterized by the model parameters comprises providing at least two surface models of different complexity, the complexity of the surface models being determined by the following variables: the type and / or number of the order parameters used in the model, the type and / or number of said variables depending on the order parameters, - the number of model parameters, the type and / or strength of regularization of the objective function used in the optimization of the model parameters, [0033] providing a validation data set including a plurality of order parameter sets, each of the order parameter sets including values for at least a portion of the parameters required to order at least one ophthalmic lens; providing at least one target value for at least one characteristic of the at least one ophthalmic lens for each of the order parameter sets in the validation data set; Further comprising: said obtaining a surface model for calculating at least one surface of at least one ophthalmic lens comprises: calculating values of a validation objective function and / or values of the variables derived from the validation objective function for the provided surface models having different complexities for a given predetermined optimized value of the model parameters of the respective surface models, the validation objective function depending on the provided target values and including at least one term for each of the order parameter sets in the validation data set, the term assuming a minimum or maximum when the provided target value of the at least one characteristic of the at least one ophthalmic lens for the respective order parameter set matches a value of the same characteristic of at least one lens that is computable or calculated using the surface model for a given optimized value of the model parameters of the surface model for a corresponding order parameter set; - selecting or determining, based on the calculated values of the validation objective function and / or with values of the variables derived from the validation objective function, a surface model for the calculation of at least one surface of at least one ophthalmic lens from the surface models with different complexities parameterized with the optimized values of the model parameters; The computer-implemented method of claim 1 , further comprising:
3. The computer-implemented method of claim 2, wherein the at least one term of the validation target function comprises a difference between at least one value of at least one characteristic of the lens, at least one surface of which is computable or calculated in accordance with the surface model for an order parameter set, and the at least one target value of this characteristic for the same order parameter set, or a convex or concave function of the difference.
4. A computer-implemented method as described in any one of claims 1 to 3, wherein the at least one term of the target function of the model parameters comprises a difference between the at least one value of the at least one characteristic of the lens, at least one surface of which is computable or calculated in accordance with the surface model for an order parameter set, and the at least one target value of this characteristic for the same order parameter set, or a convex or concave function of the difference.
5. A computer-implemented method as described in claim 2 or 3, wherein one or more terms of the validation objective function form an objective function for optimization or calculation of at least one ophthalmic lens for a given order parameter set.
6. A computer-implemented method as described in any one of claims 1 to 5, wherein one or more terms of the objective function of the model parameters form an objective function for the optimization or calculation of at least one ophthalmic lens for a given order parameter set.
7. a measurement of said at least one characteristic of an already manufactured ophthalmic lens, the order parameters of which are known; or a value determined or determinable from one or more measurements of an already manufactured ophthalmic lens; or Nominal values of the ophthalmic lens to be manufactured, the order parameters of which are at least partially known; The computer-implemented method of any one of claims 1 to 6, wherein is set as a target value for said at least one characteristic of said at least ophthalmic lens.
8. A computer-implemented method as described in claim 2, 3 or 5, wherein the validation objective function includes a weighted or unweighted sum of terms evaluated for each of the order parameter sets with all order parameters in the training data set and / or validation data set.
9. A computer-implemented method as described in any one of claims 1 to 8, wherein the objective function of the model parameters comprises a weighted or unweighted sum of terms evaluated for each of the order parameter sets with all order parameters in the training data set and / or validation data set.
10. The set of order parameters may each include the following parameters: At least one refraction value, at least one geometric or material parameter of said ophthalmic lens, At least one geometric parameter of the spectacle frame; At least one intended use of said ophthalmic lens; - at least one parameter for individualization and / or personalization of the ophthalmic lens, at least one physiological characteristic of a future wearer of said ophthalmic lens; at least one biometric characteristic of the eye or eyes of the future wearer of the ophthalmic lens, The computer-implemented method of any one of claims 1 to 9, comprising one or more of:
11. The at least one characteristic is one of the following characteristics: a vertex depth and / or a derivative thereof of said at least one surface; a surface parameter, or a combination of surface parameters, of said at least surface; Optical properties of said ophthalmic lens and / or its gradient and / or profile, the width of at least one zone of good vision of said ophthalmic lens; geometric characteristics of said ophthalmic lens; discomfort in relation to visual quality and / or posture of a wearer of said ophthalmic lens calculated using said surface model; - characteristics of the visual perception of a wearer of said ophthalmic lens calculated using said surface model; a geometric or material parameter of said ophthalmic lens not included in said order parameters; Geometric parameters of the eyeglass frame not included in the order parameters; The computer-implemented method of any one of claims 1 to 10, wherein
12. the at least one ophthalmic lens being one of a pair of lenses; the plurality of order parameter sets including respective values for at least a portion of the parameters required to order the pair of lenses; The computer-implemented method of any one of claims 1 to 11, wherein the at least one characteristic comprises a binocular characteristic of the pair of lenses.
13. A computer-implemented method according to any preceding claim, wherein the surface generated by the surface model is a continuous or continuously derivable function of the order parameters in the order parameter set.
14. the surface model is or includes a linear or non-linear regression model, the coefficients of the regression model representing at least a portion of the model parameters of the surface model; and / or the surface model is or includes a classification model; and / or A computer-implemented method according to any preceding claim, wherein the surface model is or includes a neural network.
15. The computer-implemented method of any one of claims 1 to 14, wherein the optimization of the values of the model parameters comprises regularization of the objective function used in the optimization of the model parameters.
16. The computer-implemented method comprises: providing a test data set including a plurality of order data sets; providing at least one target value for said at least one characteristic of said ophthalmic lens for each of said order parameter sets in said test data set; testing said obtained surface model for the calculation of at least one surface of at least one ophthalmic lens based on said test data set; The computer-implemented method of any one of claims 1 to 15, further comprising:
17. 1. A computer-implemented method for determining at least one surface of at least one ophthalmic lens, comprising: providing an order parameter set for the at least one ophthalmic lens; providing a function for calculating at least one surface of at least one ophthalmic lens from a set of order parameters of said at least one ophthalmic lens and / or from variables dependent on said order parameters, said function being a surface model determined according to a computer-implemented method according to any one of claims 1 to 16 or a function which approximately performs a mapping of a set of order parameters onto at least one surface of said at least one ophthalmic lens, said mapping being possible using a surface model determined according to any one of claims 1 to 16; - determining surface data of said at least one surface of said at least one ophthalmic lens from said provided set of order parameters with the aid of said provided function; 4. A computer-implemented method comprising:
18. The computer-implemented method comprises:
18. The computer-implemented method of claim 17, further comprising performing a correction of the at least one surface calculated using the surface model, the correction comprising an optimization of the surface calculated using the surface model, and / or a superposition with a superposition surface, and / or a correction of manufacturing-dependent deviations of the surface or of the optical properties of the ophthalmic lens, and / or an extension of the surface to the diameter of the ophthalmic lens according to manufacturing requirements.
19. The computer-implemented method comprises: reviewing the at least one surface calculated using the surface model to meet desired or required characteristics; - storing information regarding the fulfilment or non-fulfillment of said requested characteristic together with at least a portion of said order parameter set used to determine said surface data and / or together with said at least one surface calculated and possibly corrected using said surface model and / or together with at least one value of a desired or requested characteristic which can be provided as a target characteristic value during the determination of said surface model according to any one of claims 1 to 13; 19. The computer-implemented method of claim 17 or 18, further comprising:
20. The computer-implemented method comprises:
20. The computer-implemented method of claim 17, further comprising adapting the model parameters of the surface model after determining and / or storing each of the surfaces calculated and possibly corrected using the surface model, or after a predetermined number of surfaces have been calculated and possibly corrected using the surface model.
21. A computer program product which, when loaded into the memory of a computer and executed on said computer, has the effect that said computer performs a computer-implemented method according to any one of claims 1 to 20.
22. A device for determining a surface model for calculating at least one surface of at least one ophthalmic lens from a set of order parameters for said at least one ophthalmic lens and / or from variables dependent on said order parameters, said device comprising a computing device designed to perform the computer-implemented method according to any one of claims 1 to 16.
23. A device for determining at least one surface of at least one ophthalmic lens, said device comprising a computing device designed to perform the computer-implemented method according to any one of claims 17 to 20.
24. 1. A method for manufacturing an ophthalmic lens, comprising: Determining at least one surface of at least one ophthalmic lens according to a computer-implemented method according to any one of claims 17 to 20; manufacturing said ophthalmic lens having said at least one surface; A method comprising:
25. 1. A device for manufacturing an ophthalmic lens, comprising: A device for determining at least one surface of an ophthalmic lens according to claim 23; a manufacturing device for manufacturing said ophthalmic lens having said at least one surface; A device comprising:
Citation Information
Patent Citations
Method for optimizing optical design
JP1998068913A
A method for calculating and optimizing eyeglass lens pairs while considering binocular characteristics.
JP2011508271A
