Method and device for determining the size of a phakic intraocular implant for correcting ametropia

Abstract

The invention relates to a method and device for determining the size of a phakic corrective implant for correcting the vision of the eye (3) of a patient. On the basis of a horizontal section (6) of the anterior segment of the eye and an optical power Po, the size of a plurality of implants taken from a base Ii is determined (34), the image of the respective positions of these implants is geometrically simulated (35) on reconstructed curves s1, s2,...si of the section, then the value Vtsci of the theoretical vault without compression and the total theoretical compression CTi are calculated (37), the probability of obtaining a satisfactory vault between two threshold values S1 and S2 is calculated (42) by taking into consideration the cases of patients who have been operated for whom the respective differences between their Vtsci and vaults postoperatively for a given compression CTi have been calculated, and the implant size that has the highest probability between S1 and S2 is selected (43).

Description

METHOD AND DEVICE FOR ELABORATING THE SIZE OF AN INTRAOCULAR PHAQUE IMPLANT FOR THE CORRECTION OF REFLECTIVE VISION

[0001] The present invention relates to a method for developing the size of a phakic vision correction implant for a patient's eye and more particularly of an intraocular phakic implant (IIP) for the correction of ametropias (such as myopia, hyperopia and astigmatism).

[0002] It allows for a distance between the implant and the patient's lens that significantly limits the risks of subsequent complications.

[0003] It also relates to a device implementing such a development process.

[0004] It finds a particularly important, although not exclusive, application in the field of eye surgery, allowing the correction of vision defects observed in implanted patients, particularly between 18 and 60 years of age, between the natural lens and the patient's iris often posing reliability problems.

[0005] We know of methods which seek to predict the distance between the posterior face of the implant and the anterior face of the natural lens, called vault (in Anglo-Saxon terminology) and referred to hereafter as Vault, in order to determine the size of the optimum phakic intraocular lens which should allow effective correction of a patient's vision while limiting postoperative complications.

[0006] Such predictive processes or models are based on encrypted biometric measurements of the eyes of the patients concerned and present several drawbacks.

[0007] First, their accuracy depends on the sample (size, representativeness) on which the model is developed. For example, the model's performance depends on the patient's ethnicity, with the algorithmic formulas used performing better on patients of the same ethnicity as those used to develop the model.

[0008] They depend on the devices used to perform biometric measurements on patients.

[0009] Indeed, the biometric parameters used can vary from one machine to another, thus requiring the development of a new model and a new algorithm corresponding to each device.

[0010] It is observed that the algorithms of prior art models produce results that do not take into account the margins of error inherent in said models, and do not allow surgeons to assess the reliability of these results.

[0011] They do not allow for the effective separation of the lateral compression phenomenon of the implant from the compatibility between the patient's anatomy and the implant's geometry, thus leading to a tendency to overestimate small vaults and underestimate large vaults. Furthermore, they are less able to properly address the specific cases of certain atypical patients. A small vault is defined as a dimension (height) less than 200 microns, and a large vault as a dimension greater than 850 microns.

[0012] The present invention aims to provide a method and device for developing the optimal size of a phakic corrective implant that better meets the requirements of practice than those previously known, particularly in that it will allow the dissociation of the phenomenon of horizontal compression of the implant by the walls of the eye from that of the adequacy between the patient's anatomy and the geometry of the implant, in that it allows the use of the patient's structures directly in contact with the implant, which will make it possible to avoid deformations leading to distressing postoperative constraints, in that it offers a solution independent of considerations related to the ethnicity of the patients who were used to develop the model and / or that of the patients treated, the solution being independent of the imaging machines used, and allowing the surgeon to take into account the margin of error of the model, the uncertainties it could generate,and all this while producing a result that is visually interpretable by the latter.

[0013] Indeed, by using numerical parameters, older methods make approximations about the structures actually in contact with the implant. These approximations vary from one ethnic group to another. By actually detecting the structures involved (namely those corresponding to the iris, the lens, and the ciliary body, as well as its interparietal portion), the invention eliminates these approximations.

[0014] With the invention it is possible to almost completely eliminate defects related to errors in the geometry of the implant by detecting the structures that will serve as support and mechanical constraints for the implant, to accurately simulate the positioning of the implant in the eye, and this while taking into account the margin of error and uncertainty related to compression, which the known processes of the prior art did not do sufficiently.

[0015] More specifically, in the embodiment described here, almost all structures in direct contact with or in the immediate vicinity of the implant are detected directly. The only part modeled is then the non-visible portion between the lens and the ciliary body.

[0016] This small portion is modeled by a polynomial function, that is, of the form y = f(x) = a i x i + a i-1 x i-1+ .... a1x + a0 in the cutting plane, presents a regular shape which in fact turns out to be very well represented by this function.

[0017] The uncertainty in this area is low. The rest is actually detected. Furthermore, by knowing the exact geometries of all potential implants from the outset and integrating them into the design process, we avoid certain errors of older methods related to a mismatch between the implant geometry and the anatomy of the eye.

[0018] This results in a sharp increase in the reliability and success rate of surgical operations without deleterious postoperative effects leading to the need for reoperation.

[0019] With the present invention, in addition to the gain in quality of the result linked to the choice of the appropriate implant, greater ease of placement by the surgeon is obtained due to the excellent dimensioning.

[0020] To this end, the invention proposes, in particular, a method for elaborating the size of a phakic corrective implant for the vision of a patient's eye by a user, in which a raw image of a horizontal section of the anterior segment of the patient's eye is obtained, from said raw image on the one hand and an optical power Po of the implant determined beforehand, for example by a subjective examination of the refraction in a manner known in itself by an ophthalmologist, on the other hand, the structures corresponding to the iris I, the lens L, and the intraparietal portion M of the ciliary bodies C are identified on the image, the angle existing between a line X separating the outermost points A of the iris I and a horizontal line H (corresponding to the horizontal line of the image) is determined, a rotation is performed to make these two lines X and H coincide, four series of points are detected, namely the two points A substantially symmetrical with respect to the lens,formed by the outermost points of the iris I on the image, two points B substantially symmetrical with respect to the lens, formed by the innermost points on the image of the ciliary body C, two points D, substantially symmetrical with respect to the lens corresponding to the connection between the ciliary body C and its intraparietal portion M, a series of points E located on the anterior face F of the lens L, from these points we calculate by means of a polynomial function f(x) of degree n, with n greater than or equal to 4 at least two reconstructed curves (s1, s2, ...si) which correspond to a section of the support surface of the implant in the eye according to the plane of the image, we determine the geometry or size of a first implant I1 from the determined optical power Po and from the geometric characteristics of the implant I1 present in a determined implant database, hereinafter also referred to as the Supplier database,A theoretical or simulated uncompressed Vault Vtsci is determined by performing the following operations: the image of the respective positions of implant I1 on the reconstructed curves s1, s2, ... is geometrically simulated; from the image of these respective positions, the value Vti of the theoretical uncompressed Vault is calculated for each reconstructed curve s1, s2, ...; the theoretical or simulated uncompressed Vtsc1 retained for implant I1 is the average of the Vti values ​​obtained for the reconstructed curves; the calculations of the theoretical uncompressed Vtsci for each implant size Ii with the same optical power Po existing in the implant database are repeated; for each implant Ii, the total theoretical compression CTi is calculated, defined as the difference between the size of implant Ii and the distance between the two points A, for each implant Ii, and from its theoretical or simulated uncompressed Vtsci.From its theoretical total compression (CTi) and a database of operated patients (hereinafter also referred to as the Patients database) for which the respective differences between their theoretical Vault without compression and the postoperative Vaults actually obtained with said respective patients have been calculated, as well as their theoretical total compressions, the probability of obtaining a satisfactory Vault between two threshold values ​​S1 and S2 is calculated from the Patients database by considering the cases of patients in the database having a determined theoretical total compression (CTi), and the implant size with the highest probability of being in this range between these two threshold values ​​S1 and S2 is chosen.

[0021] The theoretical vault without compression depends only on the geometry of the implant and its position in the eye. Therefore, it does not take compressive forces into account (by its very definition). The reconstructed curves can also be described as portions of reconstructed surfaces or, by extension, reconstructed surfaces, although in 2D since they are functions f(x) (which are therefore necessarily curves).

[0022] Note that the section used is a horizontal section, but it may be, as specified for example below in the text, a section in a plane different from the horizontal plane, if we define the latter as the plane passing through the two ears of a patient in a vertical position.

[0023] In one embodiment, the theoretical Vault without compression can also be obtained by using Artificial Intelligence (AI), and more specifically by implementing an AI model (Deep Learning, Convolutional Neural Network (CNN) type, transformer or other) trained by the process described above in steps (i), (ii) and (iii).

[0024] To do this, we take as input the raw images and the size of the implants to predict the theoretical uncompressed Vault, using a large number of images, for example more than 500, for each of the four or five existing sizes (these are the sizes of the manufacturer in question). Note that there are four sizes for hyperopic eyes and four for myopic eyes as well. However, one size is specific to hyperopic eyes and another to myopic eyes.The others are common) for each image, by dissociating the effect of power on the Vault, by calculating a base Vault corresponding to the distance between the posterior face of the implant and a flat surface on which it would be placed (this base Vault therefore depends only on the size and not on the optical power of the implant), which gives a base Vault specific to each size (Vt) independent of the power and we calculate the added Vault (Vp) specific to the power p, the theoretical simulated Vault for a given size and power being given by the formula Vtp = Vt + Vp.

[0025] In other words, we already knew the base Vault (from the implant database). We then selected the implant with the lowest base Vault, subtracted the Vault related to the lens convexity for that power (also from the Implant database), and obtained the base Vault for the size Vt. We obtained Vp by subtracting Vtp from Vt for each power value.

[0026] Thus, we can train using only the size to predict a Vault without compression (therefore this time in the eye and relative to the anterior surface of the lens) for this size, then add Vp to the prediction in order to obtain the Vault without compression predicted for the power p and thus form the simulated patient database X.

[0027] The advantage of having an AI model that directly predicts the theoretical Vault without compression allows for a gain in speed, proves to be more robust to image quality (can be trained with variants of images whose quality is artificially degraded), and will allow the correction of certain inaccuracies in the geometric simulation model (if some structures are poorly detected by the segmentation model, the result can indeed be inaccurate).

[0028] Furthermore, the AI ​​model, having been exposed to a large number of cases, immediately corrects this kind of outlier.

[0029] Finally, once trained this model can be readjusted with real data just by retraining it, the readjustment of the geometric model having to be done on a case-by-case basis.

[0030] In another embodiment, it is possible to further improve the speed of obtaining results by taking compression into account from the outset. This is achieved by starting with a model predicting the theoretical Vault without compression, which is then readjusted in a known manner using algorithms implemented by the model, based on data from a large number of eyes, greater than or equal to 500, for example, 800 eyes that have actually undergone surgery, for which the implanted size and the actual postoperative Vault obtained are known.

[0031] The rest of the process remains the same in both cases, whether or not compression is taken into account from the outset, that is to say that for each implant Ii and from its theoretical or simulated Vault without compression Vtsci, its total theoretical compression CTi and the Patients database, we calculate the probability of obtaining a satisfactory Vault between two threshold values ​​S1 and S2 from the Patients database by considering the cases of patients in the database having a determined total theoretical compression CTi, and we choose the implant size presenting the highest probability value of being in this range between these two threshold values ​​S1 and S2.

[0032] In other words, we establish a basis for the difference between the Vault predicted by the model and the actual Vault, which we then stratify into compression categories according to the two thresholds Sr1 and Sr2, identical to before (for example, 0.5 and 1 mm). This basis then allows us to calculate the probability as before.

[0033] Regarding certain characteristics mentioned in the described process, the following observations should be noted:

[0034] By two points substantially symmetrical with respect to the crystalline, we mean two points symmetrical in the plane of section used with respect to a vertical axis of the crystalline coincident or as close as possible geometrically to what its axis would be if the crystalline formed a body or polygon perfectly regular, to more or less several tens of microns, for example < 100 or 50 microns.

[0035] Size refers to the overall total diameter of the implant as seen from above, that is, of the circle in which the implant is inscribed.

[0036] Raw image means the greyscale image obtained for example by optical coherence tomography, or obtained, without limitation, by high-frequency or very high-frequency ultrasound UBM (Anglo-Saxon initials for "Ultrasound Biomicroscopy"), an imaging technique that allows visualization of the internal structures of the eye (iris, ciliary body, ...) with excellent resolution.

[0037] Artificial Intelligence is understood in a way known in itself, as a set of programs and / or algorithms including learning, reasoning, problem solving, perception or decision making in this case applied to the field of the present invention.

[0038] In this case, it refers, for example, to architectural models such as ResNet described by Microsoft or EfficientNet or ViT (Vision Transformer) described by Google.

[0039] The geometric simulation of the image of the implant position on the reconstructed curve or surface is described in more detail below in an embodiment given by way of non-limiting example.

[0040] The implant databases used are those of companies in the market such as the American company STAAR Surgical.

[0041] The identification on the image of the structures corresponding to the iris, etc.… is done in a way known in itself, for example by using a “deep learning” segmentation model, such as, for example and in no way limitingly, the convolutional neural network known as UNet from the University of Freiburg (Germany), which was trained on manually labeled images.

[0042] Rotations are, for example, performed using Python code that utilizes existing libraries.

[0043] The points E are variables, automatically detected in a known way, for example by being spaced at the same determined length, for example a few microns, for example between 10 microns and 50 microns… knowing that in practice we could take all the points of the anterior face F of the lens L.

[0044] A polynomial function is, for example, a fourth-degree polynomial, but it can also be of higher degree. Its coefficients vary from one image to another and are fitted to the data points in a way chosen by a person skilled in the art, for example, using the least squares method. A person skilled in the art knows that the least squares method is a statistical regression technique used to find the line or curve that best fits a set of data points. The method consists of minimizing the sum of the squared differences between the observed values ​​and the values ​​predicted by the model.

[0045] But there are others within the reach of a person skilled in the art, such as least absolute values, weighted least squares, Bayesian methods, or even orthogonal least squares regression...

[0046] An example of a polynomial fit of a set of points using the least squares method is, for example

[0047] Points used on a sinusoidal curve: x: 0, 800, 203, 227, 271, 332, 400, 468, 529, 573, 597 with corresponding y values: y: 300, 300, 385, 320, 267, 232, 220, 232, 267, 320, 385. With the polynomial y = a⁴x⁴ + a³x³ ​​+ a²x² + a¹x + a⁰, the coefficients obtained by the least squares method are: a⁰ = 300,495; a¹ = 3.24191; a² = 0.0226687; a³ = 4.65407e⁻⁵; a⁴ = -2.9088e⁻⁸

[0048] The invention thus makes it possible to take into account the margin of error and uncertainty related to compression from a database of operated patients and to perform a probability calculation which will make it possible to obtain a Vault considered satisfactory (for example in the range between 250 and 750 microns) taking into account the results (more or less satisfactory) obtained with implants actually placed on patients, thanks to a combination of the theoretical Vault without compression with an estimate (by calculation) of the total theoretical compression CTi (horizontal) of the implant.

[0049] Another range can be chosen, for example between 300 and 800 microns, or 200 and 700 microns or even between 100 and 500 microns.

[0050] With the process according to the embodiment of the invention described more particularly here, two steps are therefore particularly important.

[0051] The first step involves compiling information about the data of operated patients, which will be stored in a structured database called the Patient Database. This is a dynamic database, as it is fully configurable and modifiable according to, for example, a simple architecture.

[0052] For example, this is a table with two columns and N rows, where N is the number of patients (i.e., patients' eyes, as they often have a record for each of their two eyes) in the database. The first column is the difference between the simulated Vault without compression and the obtained Vault. The second column, called "comp_max" and corresponding to the CTi, is the difference between the implant size and the distance between the two points A.

[0053] The Patient database uses, on the one hand, the difference between the implant size and the distance between the two horizontal walls (points A) in contact with the implant (therefore independent of ethnicity), and on the other hand, the difference between the calculated (predicted) Vault without compression and the actual Vault obtained. This difference incorporates the prediction errors of the Vault without compression and the errors related to compression.

[0054] The errors in the uncompressed Vault are indeed related to structure detection and curve modeling using the polynomial function. These elements are therefore independent of ethnicity and depend on the imaging capabilities of the measuring machine.

[0055] Indeed, errors related to compression are linked to the implant's capacity to absorb compression.

[0056] This capacity can be broken down into two components: one related to the implant itself, which has a design that allows it to absorb some of the compression (independent of ethnicity), and the other related to the patient's anatomy, which can absorb more or less compression. Since different anatomies exist in all ethnicities, risk thresholds can be objectively determined from the patient database. These thresholds are, in fact, fixed.

[0057] All these errors are further limited thanks to the use of AI models as specified above.

[0058] In one embodiment, the risk thresholds are developed for example as follows: the Patient database is separated into three categories according to the variable "size_diff", that is, according to the extent of the differences observed between the theoretical Vault without compression and the Vault actually obtained in the operated patient.

[0059] The low-risk category has the smallest range of difference values ​​(obtained Vault close to the simulated Vault without compression). These patients have uncompressed or substantially uncompressed implants.

[0060] The high-risk category, on the other hand, exhibits the widest range of difference values ​​(the obtained Vault varies considerably compared to the simulated Vault). These are patients with significant horizontal compression, which is more or less absorbed.

[0061] The moderate risk category falls between the two categories specified above, and this is generally the one of interest. However, in some cases, it may be necessary for the implant to be either not compressed at all or, conversely, only slightly compressed to achieve the desired vault, depending on the patient's anatomy.

[0062] The categories therefore come from the size_diff variable (or total compression).

[0063] For example, in one embodiment, there will be a low risk if size_diff < 0.5 mm, a moderate risk if size_diff is between 0.5 and 1 mm, and a high risk if size_diff > 1 mm.

[0064] The second step consists of implementing an algorithm using on the one hand commercial CAD / CAM software, for example written in python using open source libraries of Mathematics, image analysis and deep learning (non-exhaustively and for example known under the names SciPy, Scikit-Image, PyTorch Pandas, Numpy or Matplotlib) allowing to perform the measurements of the image structures, to apply rotations, the determination of point series and the calculations of the reconstructed curves, and on the other hand a specific software allowing the simulation of the positions of the implants chosen in the Supplier databases, allowing the calculation of the theoretical Vaults without compression.

[0065] According to the embodiment of the invention more particularly described herein, the specific algorithm or software is developed to fulfill the following functionalities: Reconstruction of the curve corresponding to the support surface by the polynomial function; Determination of the y coordinates of the curve corresponding to the support surface for the positions x1 and x2 corresponding to each of the points of application of the effective compression of the implant centered horizontally with respect to the central axis of the lens.

[0066] These elements can be easily programmed by a person in the profession, an ophthalmologist with a minimum of computer knowledge and / or combined with a computer scientist, thus making it possible to obtain the theoretical Vault without compression.

[0067] For example, the reconstruction of the curve to simulate the position is the curve formed by each value of x of the maximum between the polynomial function f(x) and the highest point of the ciliary body.

[0068] Such a process is therefore independent of measurement methods.

[0069] In the event of a change in Supplier databases, the method remains consistent and will allow positioning to be simulated in the same way.

[0070] The Patient database will be adapted or modified (if necessary) to be consistent with the use of competing implants in the new (different) Supplier database.

[0071] In advantageous embodiments, one or more of the following provisions are also used: as indicated above, the Patient database is divided into three risk categories, Gi, based on the extent of the differences observed between the theoretical Vault without compression and the Vault obtained after the patient's operation, namely a low-risk category G1 defined by a total theoretical compression Cti below a first threshold Sr1, with the smallest range of difference values; a high-risk category G2 defined by a total theoretical compression Cti above a second threshold Sr2, with the widest range of difference values; and a moderate-risk category G3 situated between the two previous categories. Sr1 is, for example, chosen between 0.2 mm and 0.8 mm, for example 0.5 mm, and Sr2 is chosen between 0.8 mm and 1.5 mm, for example 1 mm; to estimate the highest probability value,For each simulated implant size Ii, the risk category Gi is determined using the theoretical total compression threshold values ​​Sr1 and Sr2. The risk category Gi is determined by calculation based on the theoretical total compression of implant Ii located between these threshold values. The thresholds S1 and S2 are the lower and upper limits of Vault, respectively, of the desired target implant, which has a theoretical Vault without compression Vtsci. Among the number N of patients (patient eyes) in the database that have the same risk category, the number of patients P that have a difference between the theoretical Vault without compression and the actual Vault between the values ​​S1 – Vtsci and S2 – Vtsci is calculated. Then, the value P is divided by the number N of patients in the relevant Gi category in the Patients database.This yields the highest estimated probability P / N. S1 is a fixed value equal to or greater than 250 microns and S2 a fixed value equal to or less than 750 microns (S1 > 250 microns and S2 < 750 microns), advantageously S1 = 300 microns and S2 = 600 microns and / or S1 > 300 microns and S2 < 600 microns; the detection of point series is performed manually on the image by the user. There is then no automatic detection of structures by Deep Learning, but rather the identification of points of interest by hand on the image, the rest of the process remaining unchanged. Such an arrangement can be useful for correcting an inaccuracy in the automated model.or to adapt to any imaging machine if no automated model has yet been specifically trained for these images; the patient's lens is not the patient's natural lens but an artificial one. In other words, the procedure is used in the case of an "add-on" implant, that is, not on top of the patient's natural lens but on an artificial lens after cataract surgery, for example, in order to readjust the refractive result.

[0072] These are the same implant references as phakic implants, bearing in mind that current methods are not at all suitable for this case because they take into account measurements of the eye before cataract surgery which are not at all the same after surgery (in particular the height of the lens and the depth of the anterior chamber).

[0073] By directly using the post-operative image of the cataract (on which there is no longer a lens but a thinner artificial lens (implant)) we can find the true anatomical landmarks of the structures surrounding the future implant and predict the correct implant size.

[0074] Furthermore, the target acceptable Vault is no longer the same because much lower Vaults can be tolerated and Vaults will be generally higher (because the artificial crystalline lens is thinner than the natural crystalline lens).

[0075] Adding new components does not disrupt the rest of the model. Note that this is impossible in a regression formula, which is fixed once it has been developed. The process is repeated on other radial sections of the eye to obtain a 3D map of the implantation space, which will improve accuracy, particularly when the implant is not placed along the standard axis, and / or allow the surgeon to choose a different orientation to achieve the desired result. The patient database is accessed via an internet portal.

[0076] Developing a patient database allows, in a secure manner using known means, global access via an internet portal, a service which does not currently exist; The patient database used to calculate probability is regularly updated collaboratively with data from surgeons worldwide or locally with data from a given surgeon.

[0077] With the invention it is also possible to adapt the method to other types of implants than those more specifically targeted (IPI) such as the IpCL (Intra Ocular Phakic contact Lens) from the Indian company Care Group.

[0078] These implants, with their very similar shape, require adaptation as follows: a new database of the geometric characteristics of each implant is established. Actual data from operated patients is obtained to create a new Patient database, enabling the calculation of the difference between the uncompression Vault and the obtained Vault. The definitions of the new thresholds Sr1 and Sr2 for calculating the probability are defined in the same way.

[0079] In a complementary and advantageous embodiment, the patient's iridocorneal angle is taken into account.

[0080] The iridocorneal angle is the one in which the trabeculum is located, which constitutes the drainage pathway for aqueous humor in the eye.

[0081] If the angle is closed (iris pressed against the cornea) the fluid can no longer drain from the eye, the pressure increases and this constitutes a risk of glaucoma (and therefore blindness).

[0082] However, when the Vault is too large, the iris is pushed towards the cornea and can close the angle.

[0083] While some patients have a particularly wide pre-operative angle, for whom vaults larger than the maximum S2 value of 750 microns can be tolerated, conversely, some patients have a narrower than average pre-operative angle. For these patients, a vault of 750 microns is often already too large.

[0084] The idea here is therefore to adapt the maximum value S2 in order to take into account the patient's pre-operative angle.

[0085] To do this: we consider that the post-operative angle should not be less than a determined value v0, for example a value between 10° and 35°, for example 20° from the pre-operative angle we determine a maximum angle delta (ex: pre-op angle = 50°, maximum delta = -30°) we calculate what value S2' of Vault, this maximum delta would correspond to we replace S2 with this value S2' and we calculate the probability as previously indicated.

[0086] Advantageously, the results are presented differently, in the form of intervals. In other words, instead of reasoning on simulated or predicted uncompression Vault values ​​to determine the size of the implant, we reason in the same way but this time on an interval, namely the interval between the uncompression Vault (low value) and a high value consisting of the value corresponding to the uncompression Vault + a fraction of the total theoretical compression (around 20-30%).

[0087] This fraction is adapted according to the risk category, the value of the total theoretical compression or the power of the implant (positive powers are in fact with a greater risk of transmitted compression, which leads to a higher value of the total theoretical fraction mentioned above).

[0088] It should be noted that a compression actually transmitted (not absorbed) of a given value is responsible for an increase in Vault of approximately the same value.

[0089] Advantageously, the results are presented differently,

[0090] More specifically, regarding the determination of this additional fraction, the following modus operandi can be proposed as an example:

[0091] For total theoretical compression values ​​less than 0.5, this fraction will be negligible.

[0092] For total theoretical compression values ​​greater than 0.8, this fraction will be 25%.

[0093] Thus, if we have a Vault without compression of 400 microns the interval will be [400 / 400 + 0.25 x800] or [400 / 600].

[0094] This interval thus gives a more precise idea of ​​the possible results.

[0095] Because compression is often largely absorbed, it is important for surgeons to have access to the uncompressed Vault value, which represents an estimate of the minimum possible value (often close to reality given the significant absorption of compression). The largest value in the range represents a realistic estimate of the maximum Vault value. It is, in fact, quite rare to observe compression transmission exceeding 30%.

[0096] Advantageously, the size of the implant ultimately chosen takes into account the risk of implant rotation, which is displayed / communicated to the surgeon during their final selection.

[0097] This risk is estimated based on the total theoretical compression value. If this value is less than 0.1 mm or negative, the implant is possibly smaller horizontally than the space into which it will be inserted.

[0098] In the total absence of horizontal compression, the implant may then rotate, which has visual consequences, for example in the case of an implant correcting astigmatism (the majority of implants placed).

[0099] The present invention also relates to a device implementing the processes described above.

[0100] It also relates to a device for developing the size of a phakic corrective implant for the vision of a patient's eye from a raw image of a horizontal section of the anterior segment of the eye on the one hand and an optical power of the implant Po previously determined on the other hand, characterized in that it comprises means for identifying on the image the structures corresponding to the iris I, the lens L, and the interparietal portion M of the ciliary bodies C, means for calculating the angle existing between a line X separating the outermost points A of the iris I and a horizontal line H, means for rotating to make these two lines X and H coincide, means for detecting four series of points, namely the two points A substantially symmetrical with respect to the lens, formed by the outermost points of the iris I on the image, two points B substantially symmetrical with respect to the lens,formed by the innermost points on the image of the ciliary body C, two points D, substantially symmetrical with respect to the lens corresponding to the connection between the ciliary body C and its intraparietal portion M, a series of points E located on the anterior face F of the lens L, means of calculation by means of a polynomial function f(x) of degree n with n greater than or equal to 4 from these points of at least two reconstructed curves (s1, s2,...si) for the eye on which the implant is planned to be placed, means of calculation of the geometry of a first implant I1 from the determined optical power Po and from the geometric characteristics of the implant I1 present in a determined implant database called the Supplier database,means for determining a theoretical or simulated uncompressed Vault Vtsci comprising (a) means for geometric simulation of the respective positions of the implant I1 on said reconstructed curves s1, s2, ... si, (b) means for calculating, from the image of these positions, the value Vti of the theoretical uncompressed Vaults, for each reconstructed curve s1, s2, ... si, the theoretical or simulated uncompressed Vtsc1 retained for the implant I1 being the average of said Vti values ​​obtained for said reconstructed curves, and this for each prospective implant size Ii having the same optical power Po existing in said implant database, (c) means for calculating the total theoretical compression CTi defined as the difference between the size of the implant Ii and the distance between the two points A, and for each implant Ii and from its theoretical or simulated uncompressed Vtsci,of its total theoretical compression CTi and a database of operated patients (hereinafter also referred to as the Patients database) for which the difference between the theoretical Vault without compression for them and the actual postoperative Vault obtained and the total theoretical compression has been calculated, calculation methods arranged to calculate the probability of obtaining a satisfactory Vault between two threshold values ​​S1 and S2 from the Patients database by considering the cases of patients in the database having a determined total theoretical compression CTi, allowing the size of the implant to be chosen as the one with the highest probability of being located in this range between S1 and S2.

[0101] Advantageously the means of determining the theoretical or simulated Vault include an Artificial Intelligence which is trained by means (a) and (b).

[0102] Also advantageously, Artificial Intelligence is further trained to take compression into account from the outset, starting from the model predicting the theoretical Vault without compression, which is readjusted in a known way thanks to the algorithms implemented by the model, with data from a large number of eyes, for example 800 eyes actually operated on for which the implanted size and the actual post-operative Vault obtained are known.

[0103] In an advantageous embodiment, the device includes computing means arranged to divide the Patient database into three categories G1, G2, G3 of total theoretical compression according to the extent of the differences observed between the theoretical Vault without compression and the Vault obtained after operation, namely a low risk category has the smallest range of difference values, a high risk category has the largest range of difference values ​​and a moderate risk category situated between the two previous categories.

[0104] Advantageously, the system includes means of accessing the patient database via an internet portal.

[0105] Also advantageously, the device also includes calculation means allowing the compression to be taken into account from the outset by predicting the theoretical Vault without compression readjusted with data from more than 500 eyes, actually operated on and for which the implanted size and the actual post-operative Vault obtained are known.

[0106] In another advantageous embodiment the device further includes means for taking into account the iridocorneal angle of the patient arranged to adapt the maximum value S2 including calculation means in which, from a determined maximum value v0 of postoperative angle, a minimum angle delta dm is determined, it is calculated to what value S2' of Vault, this minimum delta would correspond and S2 is replaced by this value S2' before calculating the probability.

[0107] The invention will be better understood upon reading the following description of embodiments given below by way of non-limiting examples.

[0108] The description refers to the accompanying drawings in which:

[0109] is a schematic diagram illustrating the device for elaborating the size of an intraocular phakic implant (IIP) for the correction of ametropias according to an embodiment of the invention.

[0110] is a block diagram detailing the steps of the manufacturing process according to an embodiment of the invention.

[0111] schematically shows a horizontal section of a patient's eye obtained by optical coherence tomography, the following figures 4 to 8 successively showing the steps followed on this image according to an embodiment of the method of the invention.

[0112] More precisely :

[0113] shows the structure detection step.

[0114] shows the step of horizontalizing the image.

[0115] shows that of determining the points of interest used with the embodiment of the invention more particularly described here.

[0116] gives the step of reconstructing the curve which corresponds to a section of the surface which will serve as the support for the implant.

[0117] shows the simulation step of the implant on the surface and the calculation of the theoretical Vault without compression.

[0118] is an example of an interface screen rendering showing the simulated implant, complemented by an example of a table presenting to the user the results of the development process, allowing them to choose the implant most favorable to the patient whose eye section was used.

[0119] is a schematic cross-sectional representation of an implant model for parameterization usable with the invention.

[0120] is a top view of an embodiment of an implant usable with the invention.

[0121] is a cross-sectional view of two types of implants usable with the invention, presented in a superimposed manner.

[0122] illustrates the main steps for simulating the position of the implant on the surface and calculating the theoretical Vault without compression according to the embodiment of the invention more particularly described here.

[0123] shows three schematic cross-sectional views of the eye illustrating the consideration of the Iridocorneal angle to determine the maximum Vault value for a given patient.

[0124] Lamontre presents a device 1 implementing the method for determining the size of a corrective implant 2 for the vision of a patient's eye 3 (specification sheet 4) according to an embodiment of the invention. The determination is carried out by a user 5 using a raw image of a horizontal section 6 of the anterior segment of the patient's eye 3 and an optical power Po of the implant previously determined by the user in a manner known per se (shown in the patient's specification sheet 4).

[0125] This image is obtained, for example, using a high-frequency ultrasound machine.

[0126] The device includes means 8 (processor) for identification on the image 9 appearing on the screen 10 of the monitor 11 (computer) for rendering, of the structures of the eye (iris, lens, interparietal portions of the ciliary bodies...).

[0127] It includes means 12 for calculating angle, rotation, detecting points on the image 9, and implementing a polynomial function f(x) to obtain the geometry of a first implant I1 from the determined optical power Po of the patient and from the geometric characteristics of the implants in the Suppliers database 13 where the implants are listed by size and power, and are presented for example as follows:

[0128] [TABLE 1] - SUPPLIER DATABASE (example) Size Power Optical Zone Optical Vault 13.2-9.5 6.1 46 5 12.6-10.5 5.8 46 5 13.7-14.5 4.9 45 0

[0129] Column 1 gives the supplier reference regarding the size of the implant.

[0130] Column 2 gives the optical power of the implant in diopters

[0131] The optical zone is the central, round area of ​​the implant when viewed from above. It is responsible for the optical correction of the refractive error and is a lens whose surfaces can be convex, concave, planar, or toric. Outside this zone, light rays are only slightly deviated.

[0132] The optical vault is given in microns.

[0133] The functionalities of these means will be further developed below with reference to the process.

[0134] Device 1 further includes means for simulating S(y) 14 of the implant and means 15 for calculating the value Vti of the theoretical Vaults without compression as will be described more precisely with reference to the following figures.

[0135] It finally includes means 16 for calculating the total theoretical compression CTi (defined as the difference between the size of the implant Ii and the distance between the two points A), and means 17 for calculating arranged, for each implant Ii and from its theoretical or simulated Vault without compression Vtsci, its total theoretical compression Cti and a database 18 of operated patients (hereafter also referred to as the Patients database) for which the difference between the theoretical Vault without compression concerning them and the actual postoperative Vault and the total theoretical compression have been calculated, to determine the value 19 with the highest probability 20 of being located in this range between S1 and S2 allowing the choice of the size of the implant 21 to be retained.

[0136] For example, the patient database (18) is structured as follows:

[0137] [TABLE 2] – PATIENT DATABASE (EXAMPLE)#comp_max = CTivault_error670.66125.0681.30894.0691.30177.0700.36-7.0710.86114.0720.96434.0730.9436.0740.98196.0750.35-223.0761.04196.0

[0138] Column 1 gives the (confidential) reference value for the patient in question (actually, that of one eye of that patient), column 2 gives the total theoretical compression (CTi) (maximum compression) of that eye, and column 3 gives the Vault error, that is, the difference between the theoretical uncompression Vault for the patient and the Vault actually obtained or observed in that patient. We can see here that for patients 70 and 76 the results are quite good, whereas this is not the case for patients 68 and 72.

[0139] Advantageously the device implements several Supplier databases 13, 13', ...and / or several Patient databases 18, 18'... which it has access to via the internet network 22.

[0140] Artificial Intelligence (AI) can also be used advantageously to optimize the process.

[0141] To do this, we will train an AI model to directly predict, from the image and implant size, the result of the geometric simulation obtained by the previous processes.

[0142] In one embodiment, a modified convolutional neural network or Vision transformer model (ResNet, Efficientnet, ViT for example) will be used to take a number (the size of the implant) as input in addition to the image, and obtain a number as output (predicted Vault) by following these steps:

[0143] Preliminary step: Simplification of the basic Vault (spontaneous Vault of the implant when placed on a flat surface = Basic Vault mentioned in [TABLE 3] below). This Vault can be separated into a basic Vault specific to each size Vt and a supplementary Vault specific to each power Vp.

[0144] We then generate a dataset from a large number of images (for example > 1000) for which we will calculate the theoretical uncompressed Vault for each of the five sizes using the previously described geometric simulation (Note that the difference with [TABLE 3] lies in using Vt here as the basic Vault in order to be independent of the power).

[0145] We thus obtain a dataset which matches each image with the five (5) sizes with the five (5) theoretical Vaults without compression.

[0146] [TABLE 3] below gives an example of such results.

[0147] Theoretical uncompressed vault (independent of power) for the following file sizes: Image file 11.612.112.613.213.7xxx.png-12850150245342Yyy.png0123170300450

[0148] After separating the dataset into a training set and a validation set, the model is then trained with the training set. At each iteration, the model is trained with an image / size pair as input and the corresponding Vault value of the array as output.

[0149] All combinations are thus presented several times to the model during training, including random modifications of the images (rotation, mirroring, modification of light intensity and contrast, translation) in order to make it more robust.

[0150] Once the model is trained, the resulting concrete use is as follows: For a given patient whose image is available, and the implant power P determined by the pre-operative refraction to be corrected, and for each existing implant size for the patient's ametropia, a prediction of the model is made with the image and the size (therefore generally four predictions). The value of Vault Vp specific to the power is added to each of the predictions. For each size, the theoretical Vault without compression is obtained "predicted" (and no longer simulated) specific to the eye, the size and the power.

[0151] The rest of the process is the same:

[0152] In particular, a new database is created from operated patients comparing the predicted Vault to the actual Vault obtained, and the same sr1 and sr2 thresholds are used to define the compression risk category and calculate the probability of obtaining a Vault between S1 and S2.

[0153] Once the model is trained, it can also be retrained, this time with real post-operative values. The goal is then to predict not the theoretical vault without compression, but the post-operative vault (including compression).

[0154] In this case the process remains the same: a new database is created from operated patients comparing the predicted Vault to the Vault actually obtained, and the same thresholds sr1 and sr2 are used to define the compression risk category and calculate the probability of obtaining a Vault between S1 and S2.

[0155] Since the prediction of the Vault post-op compression is not perfect, the maximum total compression remains the main source of error, and probabilities can be calculated in the same way using the same theoretical total compression thresholds. Generally, the means used in this device include, in particular, means for inputting various information via a graphical screen, computing means such as a computer programmed accordingly through its graphical capabilities, a modem and / or means of connection to the internet, means for creating the images, and means for printing these images. These means are known in themselves, but arranged in a specific way to perform the functionalities of the method according to the invention, an embodiment of which will now be described in more detail with reference to the following figures.

[0156] This is a block diagram describing the implementation of the process for developing the invention more particularly envisaged.

[0157] Subsequently, reference will also be made to figures 3 to 9, which illustrate certain steps of the process more precisely.

[0158] The first step 23 of the procedure consists of obtaining or measuring a raw image of the horizontal section 6 of the anterior segment of the eye 3 of the patient in whom the prosthesis will be implanted.

[0159] We have schematically represented in shades of grey such a horizontal section of the eye which will serve as a background for the operations which will follow.

[0160] Simultaneously, the patient's optical defect is measured in step 24 by means known in themselves and the optical power Po of correction which is necessary for him is determined (step 25).

[0161] This step can advantageously be carried out at the same station and with the same equipment as in step 23 of measuring the raw image of the horizontal slice of the patient, and / or additional equipment known to the person skilled in the art (not shown on the).

[0162] From the raw image of the section, we identify on the image (step 26) the structures corresponding to the iris I, the lens L (in mixed lines on the figures), and the intraparietal portion M (in dashed lines on the figures) of the ciliary bodies C, as well as the outermost points A of the iris I.

[0163] Lalle illustrates this on the cross-section of the physical identification of these different structures and / or points.

[0164] This identification, as we have seen, can be done automatically or physically by the user using the appropriate software directly on the screen 10 of the computer 11.

[0165] The next step 27 allows us to determine the angle existing between a line X separating the outermost points A of the iris I and a horizontal line H (corresponding to the horizontal line of the image), and to perform a rotation to make these two lines X and H coincide.

[0166] This step is also illustrated by the. The angle between line 28 (X) separating points A (the outermost points of the iris I) and horizontal line 29 (H) is measured and a rotation (arrow 30) of the image of this angle (for example by a few degrees) is carried out to make lines 28 and 29 coincide.

[0167] A step 31 of detection of points of interest is then carried out comprising points A, B, D and E, the points A (approximately symmetrical with respect to the lens, formed by the outermost points of the iris I on the image having already been identified in the context of the rotation.

[0168] More specifically, and also with reference to the, we detect two points B substantially symmetrical with respect to the lens, formed by the innermost points on the image of the ciliary body C, two points D, substantially symmetrical with respect to the lens corresponding to the connection between the ciliary body C and its intraparietal portion M, a series of points E located on the anterior face F of the lens L.

[0169] The next step 32 (see also) consists of calculating from these points and by means of a polynomial function a reconstructed curve which corresponds to a section of the support surface of the implant in the eye, also called by extension reconstructed surface s1 on which the implant is planned to be placed.

[0170] We then repeat this calculation at least once (iteration 33) to construct at least a second reconstructed curve or surface s2.

[0171] On laon, two curves were thus represented corresponding to the reconstructed curves or surfaces s1 and s2 by fitting a polynomial function, for example of the 4th degree using the method of least squares, on the one hand from the points E and B (curve s1), and on the other hand from the points E and a point located in the middle of the segment drawn between the points A and D (curve s2).

[0172] The next step 34 allows the determination of the geometry or size of a first implant I1 from the optical power Po determined in step 25 and from the geometric characteristics of the implant I1 present in the Supplier implant database 13.

[0173] To do this (see also) we geometrically simulate (step 35) the image of the position of implant I1 on each of the said reconstructed curves s1, s2, etc.

[0174] On laon represented the simulation 36 of the implant on the s2 curve calculated during iteration 33.

[0175] From the image of these respective positions we calculate (step 37) the value Vti (reference 38 on the) of the theoretical Vault without compression for each reconstructed curve or surface s1, s2, ….

[0176] The theoretical or simulated uncompression Vtsc1 retained from implant I1 is the average of said Vti values ​​obtained for said reconstructed curves.

[0177] An example of calculating the theoretical Vault without compression is given below (TABLE 4).

[0178] [TABLE 4] - CALCULATION OF THE THEORETICAL VAULT WITHOUT COMPRESSION Recovery of the base Vault of the implant vault_base Determination of the vertical position y_base of the implant base on the simulated surface Determination of the vertical position of the implant optics in the eye y_optics = (y_base + vault_base) Determination of the y-coordinate of the most forward point of the lens y_lens simulated_vault = y_optics - y_lens

[0179] Then (step 39) we reiterate the calculations of theoretical Vault without compression Vtsci for each implant size Ii presenting the same optical power Po existing in said implant database.

[0180] For each implant Ii, we then calculate (step 40) the total theoretical compression CTi defined as the difference between the size of the implant Ii and the distance between the two points A.

[0181] In the embodiment of the invention more particularly described here, a step 41 of determining at least three risk categories is then carried out here, which will be used for the following step 42 of probability calculation, which will be detailed below.

[0182] Step 41 consists of separating the Patient database into three so-called risk categories Gi based on the extent of the differences observed between the theoretical Vault without compression and the Vault obtained after the patient's operation, namely a low risk category G1 defined by a total theoretical compression Cti less than a first threshold Sr1, has the smallest range of difference values, a high risk category G2 defined by a total theoretical compression Cti greater than a second threshold Sr2 has the widest range of difference values ​​and a moderate risk category G3 located between the two previous categories.

[0183] This dissociation will then be used, as we will see below, to estimate the highest probability value for each simulated implant size II.

[0184] Step 42 consists of calculating the probability of obtaining a satisfactory Vault between two threshold values ​​S1 and S2 from the Patients database by considering the cases of patients in the database having a determined total theoretical compression CTi.

[0185] This is done for each implant II and from its theoretical or simulated Vault without compression Vtsci, its total theoretical compression CTi and the database of operated patients (Patients 18 database) for which the respective differences between their theoretical Vault without compression and the postoperative Vaults actually obtained with said respective patients, as well as their total theoretical compressions, were calculated.

[0186] An example (TABLE 5) of probability calculation is given below in pseudocode, as a non-limiting example.

[0187] [TABLE 5] - PROBABILITY CALCULATION Retrieving simulated_vault Retrieving comp_max Determining upper and lower bounds of the comp_max risk category Retrieving upper and lower bounds of the interval From the patient database, calculating the percentage of rows where vault_error is within [lower_bound - simulated_vault, upper_bound - simulated_vault] among those where comp_max is between lower and upper bounds

[0188] Then (final step 43) we choose the implant size with the highest probability of being in this range between these two threshold values ​​S1 and S2.

[0189] In the embodiment more particularly described here, and for the determination of the highest probability value, step 41 (which is an option) of risk determination is used as follows.

[0190] To estimate the highest probability value, for each simulated implant size Ii, we determine which risk category Gi we are in using the threshold values ​​of total theoretical compression Sr1 and Sr2, the risk category Gi being determined by calculation from the total theoretical compression of the implant Ii located between these threshold values, the thresholds S1 and S2 being the lower and upper limits respectively of the Vault of the desired target implant, which has a theoretical Vault without compression Vtsci, among the number N of patients (patient eyes) in the database that present the same risk category, we calculate the number of patients P who have a difference between the theoretical Vault without compression and the actual Vault obtained between the values ​​S1 - Vtsci and S2 - Vtsci, then we divide the value P by the number N of patients in the said category Gi concerned in the Patients database,which gives the highest estimated probability value P / N.

[0191] An example of screen rendering obtained with the method and device according to one embodiment of the invention has been shown.

[0192] Table 39 shows the size of the implant selected – here the size corresponding to the implant of the supplier listed 12.6 in its catalogue – (second column) for a simulated Vault (sim Vault of 334 microns), the risk category chosen being category G3 “moderate” and the probability value with only S1 and S2 being fixed at 250 and 750 (3rd line).

[0193] Lamontre the modeling assumptions of the positioning of the simulated implant 40, in the form of a flattened isosceles trapezoid showing the simulation of the approximate ends 41 in dashed line on the figure, forming the buffer zones capable of absorbing part of the compression, the simulated points 42 of application of the effective (non-absorbed) compression, the (spontaneous) Vault 43 of the implant outside the eye, subjected to no compression (vertical distance between the rear face of the optic and the line of the points of application of the compression), and the distance 44 between the points of application of the compression.

[0194] Options on the user interface can allow the surgeon to better explore the results by visualizing the position and contact points of each implant size in the eye, modifying the desired Vault window according to the case (we may want to avoid low or high Vaults depending on the patient's age, anatomy (iridocorneal angle...).

[0195] Similarly, it is possible to display the possible extent of post-operative Vaults with their probability to better guide the surgeon's choice (sometimes there is no perfect option and the most suitable compromise for the patient must be chosen).

[0196] By using one of the curves rather than the other, or both curves that model the posing surface according to the visibility of the structures behind the iris, and the type of anatomy, it is possible to explore different implantation orientations (if measurements are taken on all radial sections of the eye).

[0197] Ladonne, in a top view, a representation of a type 45 implant, as produced by market suppliers and forming part of its databases.

[0198] Its diameter 46 is that of the circle 47 in which it is inscribed. It has four end tabs 48 located at the four corners of a base of the implant forming substantially a rectangle with convex sides, tabs which will cooperate with the iris and ciliary bodies of the patient's eye, and a central circular zone 49, active for dioptric correction.

[0199] As can be seen in figure 12 representing two implants 50 and 51 of the same diameter, in cross-section one on top of the other, each implant has a different Vault 52, 53 and a progressive (or non-progressive) thickness e or e' in its center according to the particular design of the implant Supplier / manufacturer.

[0200] We will now describe, in particular with reference to the, a method of performing the calculation of the Vault without compression.

[0201] After taking dioptric measurements on the patient's eye to determine the correction to be made in optical power, and making one or more horizontal sections of the eye to be corrected, the user of the process reconstructs the support surface of the implant in section (diagram 60 corresponding to the) using in particular the polynomial function f(x) – curve 61.

[0202] For each curve 61 the actual surface section used 62 (see diagram 63) to simulate the position is the curve formed by each value of x of the maximum between the polynomial function f(x) and the highest point of the ciliary body (when it exists in x).

[0203] It is then assumed that the implant 64 (see diagram 65) is horizontally centered with respect to the vertical central axis 66 of the lens.

[0204] Knowing the distance separating the two points x1 and x2 of application of the compression, we determine the coordinates y (f(x)) of the support surface section for the positions x1 and x2 corresponding to each of the application points centered horizontally by the central axis of the crystalline.

[0205] We have represented on the cross-sectional diagrams of the eye the parameters allowing to measure and / or take into account the iridocorneal angle of a patient to avoid poor drainage of aqueous humor as described above.

[0206] More specifically, diagrams 70, 71 and 72 of a cross-section of the eye 73 show the iris 74 forming an angle with the cornea 77 and its lens 75 fixed on the ciliary processes 76.

[0207] Diagram 70 thus shows the pre-operative iridocorneal alpha angle 78, dependent on the meridian.

[0208] This angle is either entered by the user (parameter provided or measurable on the imaging machine known under the references OCT (initials of Optical Coherence Tomography) (Heidelberg, TowardPi, Intalight, CSO), or UBM (initials of Ultrasound Bio Microscopy) (Quantel Medical, ArcScan), etc.), or measured on the image by image analysis method.

[0209] The lowest value is advantageously taken, as it can indeed vary from one meridian to another in the eye.

[0210] We then determine, after rotation if necessary to define a horizontal plane 79 of the eye, the angle beta 80, which the iris 74 makes with the horizontal plane 79.

[0211] For this we retain one of the two visible segments of the iris 74 defined by the two most extreme points O and P, on the right O and below and on the left and below P.

[0212] From the maximum delta angle 81 (see following diagram referenced 71) and the original position of the iris, we determine the maximum acceptable position 74' of the iris pushed forward, corresponding to the minimum acceptable angle alpha 78'.

[0213] From this position we can then calculate the maximum Vault value 82, for implant 83 (see diagram 72) allowing us to obtain this position.

[0214] It is noted here that the geometry of the implant is thus taken into account because in general the edge of the optic 84 is the point which will determine the point of application 85 on the iris 74' in its maximum position of the forward push by the implant 83. The position of the edge of the optic depends on the size of the optical zone 86 which varies with the power.

[0215] As is self-evident and as follows from the foregoing, the present invention is not limited to the embodiments described in more detail. On the contrary, it encompasses all variants thereof, and in particular those in which several databases of operated patients are used (after anonymization).

Claims

A method for determining the size of a phakic corrective lens implant for a patient's eye by a user, wherein a raw image of a horizontal section (6) of the anterior segment of the patient's eye (3) is obtained from said raw image on the one hand and a previously determined optical power of the implant Po (24) on the other hand, the structures corresponding to the iris I, the lens L, and the intraparietal portion M of the ciliary body C are identified (26) on the image, the angle existing between a line X (28) separating the outermost points A of the iris I and a horizontal line H (29) is determined, a rotation (30) is performed to make these two lines X and H coincide, four sets of points are detected, namely the two points A, substantially symmetrical with respect to the lens, formed by the outermost points of the iris I on The image, two points B, approximately symmetrical with respect to the lens,formed by the innermost points on the image of the ciliary body C, two points D, substantially symmetrical with respect to the lens corresponding to the connection between the ciliary body C and its intraparietal portion M, a series of points E located on the anterior face F of the lens L, from these points we calculate (32) by means of a polynomial function f(x) of degree n with n greater than 4 at least two reconstructed curves (s1, s2, ...si) which correspond to a section of the support surface of the implant in the eye along the plane of the image section, we determine (34) the geometry or size of a first implant I1 from the determined optical power and from the geometric characteristics of the implant I1 present in a determined implant database, hereafter also referred to as the Supplier database (13), we geometrically simulate (35) the respective positions of the implant I1 on the said reconstructed curves s1 and s2,From the image of these respective positions, we calculate (37) the value Vti of the theoretical uncompressed Vault for each reconstructed curve s1, s2,...si, the theoretical or simulated uncompressed Vtsc1 retained for implant I1 being the average of said Vti values ​​obtained for said reconstructed curves, we repeat (39) the calculations of the theoretical uncompressed Vtsci for each implant size Ii having the same optical power Po existing in said implant database, for each implant Ii we calculate (40) the total theoretical compression CTi defined as the difference between the size of implant Ii and the distance between the two points A, for each implant Ii and from its theoretical or simulated uncompressed Vtsci, its total theoretical compression Cti and a database of operated patients (18, 18',...)(hereinafter referred to as the Patient database) for which the difference between the theoretical Vault without compression concerning them and the actual postoperative Vault obtained and the total theoretical compression concerning them has been calculated, we calculate (42) the probability of obtaining a satisfactory Vault between two threshold values ​​S1 and S2 and we choose (43) the implant size with the highest probability of being in this range between S1 and S2. A method according to claim 1, characterized in that an Artificial Intelligence (AI), trained by the method described in steps (i), (ii) and (iii), is used, inputting the raw images as well as the size of the implants to predict the theoretical Vault without compression, from a number of images greater than 500, for each of the five existing implant sizes for each image, by calculating a basic Vault specific to each size (Vt) independent of the power and corresponding to the distance between the posterior face of the implant and a flat surface on which it would be placed, the added Vault (Vp) specific to the power p is calculated, the simulated theoretical Vault for a given size and power being Vtp = Vt + Vp = Vtsci for the implant concerned. Method according to claim 2, characterized in that it further improves the speed of obtaining results by taking into account from the outset the compression by predicting the theoretical Vault without compression readjusted with the data of a number of eyes greater than 500, actually operated and for which the implanted size and the actual post-operative Vault obtained are known. A method according to any one of the preceding claims characterized in that the Patient database is separated (41) into three risk categories Gi based on the extent of the differences observed between the theoretical Vault without compression and the Vault obtained after the patient's operation, namely a low risk category G1 defined by a total theoretical compression Cti less than a first threshold Sr1, has the smallest range of difference values, a high risk category G2 defined by a total theoretical compression Cti greater than a second threshold Sr2 has the widest range of difference values ​​and a moderate risk category G3 situated between the two previous categories. A method, according to claim 4, characterized in that, to estimate the highest probability value, for each simulated implant size Ii, the risk category Gi is determined using the threshold values ​​of total theoretical compression Sr1 and Sr2, the risk category Gi being determined by calculation from the total theoretical compression of the implant Ii located between these threshold values, the thresholds S1 and S2 being the lower and upper limits respectively of Vault of the desired target implant whose simulated Vault without compression is Vtsci, among the number N of patients (patient eyes) in the database that have the same risk category, the number of patients P that have a difference between the theoretical Vault without compression and the actually obtained Vault between the values ​​S1 - Vtsci and S2 - Vtsci is calculated.Then we divide the value P by the number N of patients in the relevant category Gi in the Patients database, which gives the highest estimated probability value P / N. A method according to any one of the preceding claims, characterized in that S1 is a fixed value > 250 microns and S2 is a fixed value < 750 microns, advantageously S1 > 300 microns and S2 < 600 microns. Method according to any one of the preceding claims, characterized in that the detection of the series of points is carried out manually on the image (9) by the user (5). A method according to any one of the preceding claims, characterized in that the method is used for calculating an “add-on” implant on an artificial lens. A method according to any one of the preceding claims, characterized in that the polynomial function is a 4th degree polynomial f(x) whose coefficients vary from one image to another and are adjusted by the least squares method to obtain the curve corresponding to a section of the desired surface. A method according to any one of the preceding claims, characterized in that the method is repeated on other radial sections of the eye to obtain a 3D map of the implantation space. A method according to any one of the preceding claims, characterized in that one or more Patient databases (18, 18', ...) are accessed via an internet portal (22). A method of any one of the preceding claims, characterized in that the patient's iridocorneal angle is taken into account by adapting the maximum value S2 to take into account the patient's pre-operative angle by adopting the following steps: the post-operative angle is considered to be less than a determined value v0, between 10° and 35°, from the pre-operative angle a maximum angle delta is determined, the value S2' of Vault is calculated, this maximum delta would correspond to, S2 is replaced by this value S2' before calculating the probability. A device for determining the size of a phakic corrective lens implant for the vision of a patient's eye (3) from a raw image (9) of a horizontal section (6) of the anterior segment of the eye, on the one hand, and a previously determined optical power of the implant Po, on the other hand, characterized in that it comprises means (8) for identifying on the image (9) the structures corresponding to the iris I, the lens L, and the interparietal portion M of the ciliary bodies C; means (12) for calculating the angle between a line X separating the outermost points A of the iris I and a horizontal line H; means for rotating the image to make these two lines X and H coincide; and means (12) for detecting four sets of points, namely, said two points A, substantially symmetrical with respect to the lens, formed by the outermost points of the iris I on the image, and two points B, substantially symmetrical with respect to the crystalline,formed by the innermost points on the image of the ciliary body C, two points D, substantially symmetrical with respect to the lens corresponding to the connection between the ciliary body C and its intraparietal portion M, a series of points E located on the anterior face F of the lens L, means (12) for calculating, by means of a polynomial function f(x) of degree n with n greater than or equal to 4, at least two reconstructed curves (s1, s2, ...si) which correspond to a section of the implant support surface in the eye along the plane of the image section, means (12) for calculating the geometry of a first implant I1 from the determined optical power Po and from the geometric characteristics of the implant I1 present in a determined implant database called the Supplier database (13, 13'...), means (14) for geometric simulation of the respective positions of the implant I1 on the aforementioned reconstructed curves s1, s2,... if, means (15) for calculating, from the image of these positions, the value Vti of the theoretical Vaults without compression, for each reconstructed curve s1, s2,... if, the theoretical or simulated Vault without compression Vtsc1 retained for the implant I1 being the average of said Vti values ​​obtained for said reconstructed curves, and this for each prospective implant size Ii presenting the same optical power Po existing in said implant database, means (16) for calculating the total theoretical compression CTi defined as the difference between the size of the implant Ii and the distance between the two points A, and for each implant Ii and from its theoretical or simulated Vault without compression Vtsc1,of its total theoretical compression Cti and a database of operated patients (hereinafter also referred to as the Patients database) for which the difference between the theoretical Vault without compression concerning them and the actual postoperative Vault obtained and the total theoretical compression has been calculated, means (17) of calculation arranged to calculate the probability of obtaining a satisfactory Vault between two threshold values ​​S1 and S2 from the Patients database by considering the cases of patients in the database having a determined total theoretical compression CTi, allowing the size of the implant (21) to be retained to be chosen as the one having the highest probability value (19) (20) of being located in this range between S1 and S2., Device according to claim 13, characterized in that it comprises an Artificial Intelligence (AI), trained by the method described in steps (i), (ii) and (iii) to predict the theoretical Vault without compression, from a number of images greater than 500, for each of the five existing implant sizes for each image, said AI comprising means for calculating a basic Vault specific to each size (Vt) independent of the power and corresponding to the distance between the posterior face of the implant and a flat surface on which it would be placed, and means for calculating the superimposed Vault (Vp) specific to the power p, the simulated theoretical Vault for a given size and power being Vtp = Vt + Vp = Vtsci for the implant concerned. Device according to claim 14, characterized in that it further comprises calculation means allowing to take into account from the outset the compression by predicting the theoretical Vault without compression readjusted with the data of a number of eyes greater than 500, actually operated and for which the implanted size and the actual post-operative Vault obtained are known. Device according to any one of claims 13 to 15, characterized in that it further comprises means for taking into account the iridocorneal angle of the patient arranged to adapt the maximum value S2 comprising calculation means in which, from a determined maximum value v0 of postoperative angle, a minimum angle delta dm is determined, it is calculated to what value S2' of Vault, this minimum delta would correspond and S2 is replaced by this value S2' before calculating the probability.

Images