Implant identification

Abstract

Implant identification involves obtaining a machine learning (ML) model trained to select an anatomical implant model of a physical implant for replacement of a target anatomical structure based on characteristics of the anatomical surface of an anatomical structure adjacent to the target anatomical structure to be replaced, obtaining image data of a patient's anatomical region, where the anatomical region has the target anatomical structure and other anatomical structures adjacent to the target anatomical structure, determining characteristics of the anatomical surface of the other anatomical structures, where the anatomical surface exists at each of the interfaces between the other anatomical structures and the target anatomical structure, applying the ML model using the determined characteristics of the anatomical surface, and obtaining an implant model selected by the ML model as a designation of a physical implant for at least partial replacement of the target anatomical structure as a surgical implant candidate within the patient.

Description

Background Art

[0001] Among anatomical injuries, there are some that are more difficult to heal than others. For example, for various bones, due to insufficient blood supply to promote rapid healing, patients may be unable to perform certain body functions if the healing period is prolonged. In some cases, especially for certain bones in the human body, the injury can be addressed by an implant device that replaces the bone or a part thereof. Due to the interaction of such bones with other anatomical structures, such as other bones in the case of joints, the accurate identification, selection, and creation of an appropriate implant for a specific patient are important.

Summary of the Invention

[0002] The provision of a computer-implemented method overcomes the disadvantages of the prior art and provides further advantages. The method obtains a machine learning model trained to select an anatomical implant model of a physical implant for partial or total replacement of a target anatomical structure based on the characteristics of the anatomical surface of the anatomical structure adjacent to the target anatomical structure to be replaced, in part or in whole. The method obtains image data of an anatomical region of a patient, the anatomical region having the patient's target anatomical structure and the patient's other anatomical structures, the other anatomical structures being adjacent to the patient's target anatomical structure. The method determines the characteristics of at least one anatomical surface of the other anatomical structures from the image data, the at least one anatomical surface being present at each of at least one interface between the patient's other anatomical structures and the target anatomical structure. The method utilizes the determined characteristics of the at least one anatomical surface, applies the machine learning model, and obtains the selected implant model, the selected implant model being selected by the machine learning model as a designation of a physical implant for at least partial surgical implantation in the patient as a partial or total replacement of the patient's target anatomical structure.

[0003] Furthermore, a computer system is provided, including memory and a processor that communicates with the memory, and the computer system is configured to perform the method. The method obtains a machine learning model trained to select an anatomical implant model for the partial or complete replacement of a target anatomical structure based on the characteristics of the anatomical surface of an anatomical structure adjacent to the target anatomical structure to be partially or completely replaced, the method obtains image data of an anatomical region of a patient, the anatomical region has the target anatomical structure of the patient and other anatomical structures of the patient, the other anatomical structures are adjacent to the target anatomical structure of the patient, the method determines the characteristics of at least one anatomical surface of the other anatomical structure from the image data, the at least one anatomical surface is present at each of at least one interfaces between the other anatomical structure of the patient and the target anatomical structure, the method utilizes the determined characteristics of the at least one anatomical surface, applies the machine learning model, obtains an implant model selected based on the application, the selected implant model is selected by the machine learning model as a designation of a physical implant for a candidate for surgical transplantation in the patient, at least partially, as a partial or complete replacement of the target anatomical structure of the patient.

[0004] Furthermore, a computer program product is provided for performing the method, which includes a computer-readable storage medium that is readable by a processing circuit and stores instructions for execution by the processing circuit. The method obtains a machine learning model trained to select an anatomical implant model for the partial or total replacement of a target anatomical structure based on the characteristics of the anatomical surface of an anatomical structure adjacent to the target anatomical structure to be partially or totaled. The method obtains image data of an anatomical region of a patient, the anatomical region having the target anatomical structure of the patient and other anatomical structures of the patient, the other anatomical structures being adjacent to the target anatomical structure of the patient. The method determines the characteristics of at least one anatomical surface of the other anatomical structure from the image data, the at least one anatomical surface being present at each of at least one interfaces between the other anatomical structure of the patient and the target anatomical structure. The method utilizes the determined characteristics of the at least one anatomical surface, applies the machine learning model, and obtains an implant model selected based on the application. The selected implant model is selected by the machine learning model as a designation of a physical implant for a candidate for surgical transplantation in the patient, at least partially, as a partial or total replacement of the target anatomical structure of the patient.

[0005] Additionally or alternatively, the method may include providing a model candidate designation module with selected implant models to designate model candidates for verification.

[0006] Additionally, alternatively, the method may include presenting a selected implant model to the user on a graphical user interface, receiving an operation on the selected implant model, the operation of modifying the selected implant model, generating implant model candidates for validation, where the implant model candidates specify one or more physical implants that differ in one or more physical properties compared to the physical implants specified by the selected implant model, and providing the implant model candidates to a validation module for validation, the validation of determining whether the physical implants specified by the implant model candidates are suitable for surgical implantation in a patient.

[0007] Additionally or alternatively, the operations performed on the selected implant model may include at least one operation specified by the user and / or at least one operation automatically determined by artificial intelligence.

[0008] Additionally or alternatively, based on verification that determines that a physical implant designated by an implant model candidate will not pass, the method may include repeating (i) receiving an operation and (ii) providing the implant model candidate to a verification module for verification, one or more times, in each iteration of the iteration, the implant model candidate that failed to pass is provided as the selected implant model for the next iteration of the iteration.

[0009] Additionally or alternatively, the method may include receiving manipulation on the properties of at least one anatomical surface of another anatomical structure, and the validation for determining whether a physical implant specified by a candidate implant model is suitable for surgical implantation in a patient may be at least partially based on the manipulated properties of at least one anatomical surface.

[0010] Additionally or alternatively, operations on the properties of at least one anatomical surface may include at least one operation specified by the user and / or at least one operation automatically determined by artificial intelligence.

[0011] Additionally or alternatively, the method may include receiving manipulation on the properties of at least one anatomical surface of another anatomical structure and providing a selected implant model as an implant model candidate for validation to a validation module for validation, the validation determining whether a physical implant specified by the implant model candidate would pass surgical implantation in a patient, and the validation being at least partially based on the manipulated properties of at least one anatomical surface.

[0012] Additionally or alternatively, based on (i) an operation on a selected implant model, wherein the operation modifies the selected implant model, generates implant model candidates for validation, and the different implant model candidates specify a physical implant having one or more different physical properties compared to the physical implant specified by the selected implant model, and / or (ii) an operation on the properties of at least one anatomical surface of another anatomical structure, the method may include providing the implant model candidates to a validation module for validation, the validation determining whether the physical implant specified by the implant model candidates is fit for surgical implantation in a patient, and based on the validation determining that the physical implant specified by the implant model candidates is fit for surgical implantation in a patient, the method may include displaying the implant model candidates in a training dataset as part of a training example relating the implant model candidates to at least one anatomical surface of another anatomical structure.

[0013] Additionally or alternatively, the method may include providing a selected implant model as an implant model candidate to a validation module for verification, which determines whether the physical implant specified by the implant model candidate is suitable for surgical implantation in a patient.

[0014] Additionally or alternatively, the implant model candidate may be the initial implant model candidate. Based on the verification that the physical implant specified by the initial implant model candidate is deemed unacceptable, the method may (a) receive an operation on the initial implant model candidate, wherein the operation modifies the initial implant model candidate, generates a different implant model candidate for verification, and the different implant model candidate specifies a physical implant having one or more different physical properties compared to the physical implant specified by the initial implant model candidate, and / or (ii) receive an operation on the properties of at least one anatomical surface of another anatomical structure, and (b) determine the next implant model candidate to provide to the verification module for verification, wherein, based on having received the operation on the initial implant model candidate, the next implant model candidate is a different implant model candidate generated from the operation on the initial implant model candidate. (c) Determining an initial implant model candidate as a next implant model candidate based on the fact that the initial implant model candidate has not received an operation on the initial implant model candidate, and having determined the plant model or received an operation on the properties of at least one anatomical surface of another anatomical structure; and (c) providing the next implant model candidate to a validation module for validation to determine whether the physical implant specified by the next implant model candidate will pass surgical transplantation in the patient, wherein the validation to determine whether the physical implant specified by the next implant model candidate will pass surgical transplantation in the patient is provided, which is at least partially based on at least one selected from a group having operations on the initial implant model candidate and operations on properties of at least one anatomical surface.

[0015] Additionally or alternatively, the method may include training a machine learning model to select anatomical implant models, training the machine learning model to select anatomical implant models from the library of implant models using samples from a library of implant models, and the selected implant models are selected by the machine learning model from the library of implant models.

[0016] Additionally or alternatively, the machine learning model may include a trained generator of a generative adversarial network (GAN), which is trained using samples from a library of implant models and trained to generate implant models, with the selected implant model being the one generated by the generator and output as the selected implant model.

[0017] Additionally or alternatively, the obtained image data may include three-dimensional digital model data representing the anatomical regions of the patient, and determining the properties of at least one anatomical surface of other anatomical structures may include processing the image data to present at least one anatomical surface as at least one digital three-dimensional surface, and converting at least one digital three-dimensional surface into at least one two-dimensional projection, wherein the determined properties of at least one anatomical surface are determined from at least one two-dimensional projection.

[0018] Additionally or alternatively, the method may include preprocessing image data to generate a three-dimensional digital model of the patient's anatomical regions from which the target anatomical structures have been omitted.

[0019] Additionally or alternatively, one or more anatomical surfaces may include one or more articular surfaces with which a physical implant will engage, based on being at least partially surgically implanted within the patient.

[0020] Additionally or alternatively, the anatomical region may include the patient's ankle, the anatomical structure of interest may have the talus, and at least one anatomical surface may have at least one articular surface of at least one bone adjacent to the talus.

[0021] Additionally or alternatively, determining the characteristics of at least one anatomical surface can be based on (i) manual indication by the user of at least one anatomical surface provided based on user input to a graphical user interface displaying a model including at least other anatomical structures of the patient, and / or (ii) automated analysis of image data to identify at least one anatomical surface.

[0022] Further features and advantages are realized by the concepts described herein.

[0023] The embodiments described herein are specifically pointed out and explicitly claimed in the claims at the end of this specification. The aforementioned and other purposes, features, and advantages of this disclosure are evident from the following detailed description in conjunction with the accompanying drawings. [Brief explanation of the drawing]

[0024] [Figure 1] A diagram illustrating an example environment for using the embodiments described herein. [Figure 2A] Figure showing an example of computed tomography (CT) images segmented according to the embodiments described herein. [Figure 2B] Figure showing an example of computed tomography (CT) images segmented according to the embodiments described herein. [Figure 2C] Figure showing an example of computed tomography (CT) images segmented according to the embodiments described herein. [Figure 3A] An illustrative ankle model with the talus bone omitted. [Figure 3B] An illustrative ankle model with the talus bone omitted. [Figure 4A] Figure showing an exemplary bone model depicting an articular surface specified by the aspects described herein [Figure 4B] Figure showing an exemplary bone model depicting an articular surface specified by the aspects described herein [Figure 5A] Figure showing characteristics of an articular surface separated from a patient's anatomical structure according to the aspects described herein [Figure 5B] Figure showing characteristics of an articular surface separated from a patient's anatomical structure according to the aspects described herein [Figure 6] Figure showing an exemplary interface for displaying and selecting a best-fit implant model according to the aspects described herein [Figure 7] Conceptual diagram of training and use of a machine learning model according to the aspects described herein [Figure 8] Figure showing an exemplary process flow for implant selection based on artificial intelligence according to the aspects described herein [Figure 9] Figure showing an exemplary graphical user interface of software for executing the aspects described herein [Figure 10] Figure showing an example of a command line interface for identifying model results [Figure 11] Figure showing an exemplary interface for displaying best-fit implant model candidates according to the aspects described herein [Figure 12] Figure showing an exemplary process for implant selection according to the aspects described herein [Figure 13] Figure showing an example of a computer system and related apparatus for incorporating and / or using the aspects described herein

Mode for Carrying Out the Invention

[0025] This specification describes methods for anatomical analysis and selection of optimal anatomical implant hardware. Embodiments can identify optimal anatomical implant models based on the existing anatomical structure of a patient. In certain embodiments, a machine learning (ML) model, such as a deep neural network, is trained to select at least one optimal or “best-fit” anatomical implant model and / or its characteristics based on one or more surfaces of surrounding patient anatomical structures that are expected to engage or interact with the corresponding physical implant once it is provided within the patient. A specific example provides artificial intelligence (AI)-assisted selection of optimal talus implant models, in which case the software identifies one or more best-fit models for a talus implant from a database of talus models. The identified implant models can, if available, indicate the selection of a physical embodiment of that model, or optionally, be used as a designation for the physical implant to be generated / manufactured. “Anatomical implant model” (hereinafter also interchangeably referred to as “implant model,” “anatomical structure model,” or “anatomical model”) means a digital (i.e., data-formed) model of an anatomical structure. This digital model can function as a specification for a physical implant, and the specification can be used for the manufacture of a physical implant and / or the selection of an existing, for example, commercially available physical implant (if such an implant has already been manufactured).

[0026] Therefore, in some embodiments, a best-fit talus model identified and selected from a database or library of models can be presented for use in selecting or manufacturing a hardware implant with desired characteristics. For example, an implant model of an anatomical structure validated for a particular patient can serve as a basis for identifying existing implants or as a specification for an implant to be generated (e.g., manufactured). In the latter case, the identified anatomical implant model can be loaded from a file such as an .stl file, optionally manipulated, validated, and provided to a manufacturing device such as a 3D printer.

[0027] The search for the best-fit implant model may be based on the anatomical surfaces of anatomical structures adjacent to the target anatomical structure for replacement. For example, the anatomical surfaces may be articular surfaces of other anatomical structures with which the target anatomical structure for replacement engages.

[0028] This specification discusses and presents various aspects of the talus ("ankle"). The talus is one of a group of bones of the foot called the tarsals. The tarsals form the lower part of the ankle joint through articulation. The talus can transmit the entire body weight, or a significant portion of the body weight, to the foot. The talus is broadly covered with cartilage and, together with the calcaneus ("heel bone") and navicular bone, forms the talocalcaneal joint. The talus can be said to have three basic parts: (i) the head, (ii) the neck, and (iii) the trunk. Examples of such articular surfaces in talus replacement surgery include the calcaneal articular surface, the navicular articular surface, the tibial articular surface, and / or the fibular articular surface.

[0029] As mentioned above, various injuries can be difficult to heal. For example, because the talus does not receive sufficient blood supply, a fractured talus may make it impossible to walk without crutches for months. One approach is to treat the injury with a talus implant. To insert a talus implant as a substitute for the patient's natural talus, the ankle joint is first incised and exposed, and then the obstructing extensor digitorum longus tendon, anterior tibial artery, and extensor digitorum longus muscle are moved or removed. The talus may be fixed not only by the anterior talofibular ligament and superficial deltoid ligament, but also by syndesmosis and the anterior tibiofibular ligament. The exposed, damaged talus can be removed from the ankle joint and replaced by inserting a talus implant into the space previously occupied by the patient's natural talus. However, because the talus interacts with other bones in the ankle, precise talus replacement and transplantation methods / systems are desired for accurate selection, creation, and transplantation of the replacement talus in the patient. Existing talus replacement surgery employs a ground-up approach, designing a custom-made talus for each patient. This is time-consuming and expensive. In several aspects described herein, machine learning is used to assist in the selection and generation of the best-fitting anatomical implant.

[0030] The embodiments described herein are presented with reference to the human talus, but this is for illustrative purposes only. Those skilled in the art will recognize that the embodiments presented herein can be used to inform the selection / generation and other embodiments of implants for use in conjunction with other anatomical features in the human body and / or other anatomical structures, and furthermore, that the embodiments described herein can be applied to both whole and partial replacement of anatomical structures.

[0031] Therefore, by example rather than by limitation, several embodiments described herein utilize machine learning to support the selection and / or generation of best-fit implants, such as talus implants. Artificial intelligence may be used, for example, in the form of one or more trained ML models, to cognitively analyze specific anatomical parameters (e.g., articular surfaces) of a particular patient and from this analysis to select a best-fit anatomical structure implant model, such as a talus design. In some examples, the selection is made from an existing library / database of designs, such as a digital three-dimensional (3D) model (or other designation) of the talus. In some examples, the analysis of anatomical parameters and the selection of candidate best-fit models may signal operations on the model that result in new models to provide to the database. The ML models employed herein are self-learning and can be improved by use. Furthermore, the parameters used to select anatomical models, such as talus models for implants, can be adjusted by additional processing to select and generate best-fit solutions tailored to individual patients.

[0032] The program code can determine anatomical (e.g., joint) surface properties, such as dimensions, distances, profiles, and other characteristics, from images of a given patient's anatomical structure, and use these characteristics to select a best-fit anatomical implant model from an optional library, utilizing machine learning. The ML models used by the embodiments discussed herein may be of various types, including, for example, neural networks including recurrent neural networks and / or convolutional neural networks. If specific data is not available to help the program code select an existing best-fit model, the embodiment can generate a simulation of this missing data, and the simulation can be used to select (and optionally manufacture) an implant. For example, if a portion of the talus anatomical structure is missing, that portion can be simulated for the purpose of selection / manufacturing.

[0033] In an exemplary embodiment, program code running on one or more processors processes and analyzes data from images, such as digital imaging, communication, and medical (DICOM) images of a patient's anatomical structure, generates a digital model of the anatomical structure, and automatically determines / selects an implant model from a library of implant models based on the digital model of the anatomical structure. This selected model may be manipulated and become a candidate to replace the patient's anatomical structure. The candidate may be subject to validation for a particular patient. If validated, a corresponding physical implant is selected and, if available, for example, as an existing "off-the-shelf" product. Alternatively, the physical implant can be manufactured using manufacturing techniques such as 3D printing, for example, according to the specification of the anatomical structure implant model.

[0034] Therefore, in some embodiments, AI-implementing software can provide recommendations regarding appropriate anatomical models for potential replacement of a patient's anatomical structure, and thus, based on a particular patient's anatomical structure, can help identify implant solutions for selection or arbitrary manufacturing. Current approaches to producing implants are largely manual, potentially inaccurate, and customized, which can lead to high workloads and increased overall process costs. In contrast, embodiments discussed herein can adapt / design prototypes of custom implants for a specific set of conditions by automating parts of the implant design / selection process based on training and application of machine learning models, and by identifying the best existing designs for selecting / producing patient-specific implants.

[0035] Figure 1 shows an exemplary environment for incorporating and using the embodiments described herein. The environment shows an exemplary technical architecture and data flow that illustrates a particular general embodiment described in more detail herein. The environment includes a controller 110, such as a computer system, having processing circuitry 120 including one or more processors (as an example) and memory 130. Memory 130 may contain program code / instructions executable by the processing circuitry to cause the controller 110 to perform functions such as the processing described herein. The program code in memory 130 may contain a set of various codes / instructions configured to perform a particular activity. Figure 1 shows an example of such a set, called a module or component. These are shown as separate elements within memory 130 for conceptual purposes, but in other examples, these modules can be further separated and / or combined. The specific divisions between modules shown in Figure 1 are for illustrative purposes only.

[0036] The data processing module 140 acquires image data 102. In other examples, the image data is acquired from a data store that stores the image data, but in this example, the image data is acquired from an imaging device such as a medical imaging device. The image data can include images of anatomical regions and structures in any of a variety of digital formats. One such format is based on the Digital Imaging and Communications in Medicine (DICOM) standard. The exemplary image data is scan data (e.g., a set of images) generated from computed tomography (CT) or other imaging diagnostic procedures.

[0037] The data processing module 140 processes the obtained image data 102 to perform desired standardization, cleaning, or initial processing. For example, a CT scan generates a stack of two-dimensional (2D) images that divide anatomical structures into very thin "slice." For example, a CT scan of a foot yields 600 images. The program code of the data processing module 140 can segment the image data, which in this example means converting 2D images into a 3D model. Thus, the data processing module 140 can create a 3D model from the input image data 102.

[0038] The anatomical regions presented by the image data may include the patient's anatomical structures that may be replaced by implants (referred to herein as the "target anatomical structure"), and other patient anatomical structures that are at least partially adjacent to the target anatomical structure.

[0039] By specific example, Figures 2A to 2C show examples of computed tomography (CT) images loaded and segmented according to the embodiments described herein. Figure 2A shows a coronal-plan image of the patient's ankle, Figure 2B shows an axial-plan image of the patient's ankle, and Figure 2C shows a sagittal-plan image of the patient's ankle. The image data has been processed into bone segments that differ from the patient's anatomical structure. The images shown present different bones.

[0040] Given that the program code is ultimately used to select / specify an implant suitable for the target anatomical structure (the talus in these examples), the image data is processed by the data processing unit 140 and / or data analyzer 142 to remove image data of the specific target anatomical structure that the implant will replace. The resulting model can then be presented to the user for viewing. For example, Figures 3A and 3B show examples of anterolateral (Figure 3A) and posteromedial (Figure 3B) views of a digital model of a patient's ankle region with the talus omitted. Specifically, the tibia 302, fibula 304, calcaneus 306, and tarsal / metatarsal bones 308 are depicted surrounding the space 310 where the talus resides but is removed in the program.

[0041] The processed data from the data processing device 140 is provided to the data analyzer 142 for analysis. The data analyzer 142 is used to identify the characteristics of anatomical surfaces adjacent to the target anatomical structure. In this example, the anatomical surface is the articular surface, i.e., the surface of the anatomical structure into which the patient's talus engages (in this example), and therefore the surface into which a suitable talus implant is expected to make contact when placed in the patient. Thus, as part of the data analysis by the analyzer 142, the patient's articular surfaces surrounding the anatomical structure are identified. In some embodiments, the articular surfaces are identified in whole or in part manually by a user (such as a physician or an engineer specializing in virtual surgical planning) who identifies the boundaries and other characteristics of these articular surfaces using a computer system, display, and input device. Additionally or alternatively, machine learning-based AI may perform the identification by taking image data as input and using a model trained to identify the articular surfaces of adjacent anatomical structures presented in the image data. For example, the program code of the data analyzer 142 can apply a machine learning algorithm utilizing a neural network to detect and determine the characteristics of articular surfaces presented in the image. In the context of the embodiments presented herein, the articular surface is the surface of at least several other bones with which the talus contacts, such as the navicular bone, calcaneus, fibula and / or tibia.

[0042] Figures 4A to 4B show examples of the patient bone models from Figures 3A to 3B, with the three aforementioned articular surfaces highlighted. Models with highlighted surfaces can be rendered for user viewing. In Figures 4A to 4B, the fibula 404, calcaneus 406, and navicular bone 412 are shown in their relative positions to each other, leaving space 410 for a talus implant. 414a shows the calcaneal articular surface between the calcaneus and the talus, i.e., the portion of the calcaneus surface that the talus physically contacts / engages with and articulates with; 414b shows the navicular articular surface between the navicular bone and the talus, i.e., the portion of the navicular bone surface that the talus physically contacts / engages with; and 414c shows the fibular articular surface between the fibula and the talus, i.e., the portion of the fibula base surface that the talus physically contacts / articulates with. Based on user input, the software can determine and / or understand various properties of these surfaces (size, shape, perimeter, area, surface contour, etc.) and use this information to assist in model selection / generation processes as described.

[0043] Figures 5A and 5B show the characteristics (dimensions, circumference, and relative position, etc.) of the articular surfaces of Figures 4A and 4B, where surfaces 414a, 414b, and 414c are separated from the anatomical structures of other patients, according to the embodiments described herein. The embodiments may, if necessary, display these figures on a display for the user to see.

[0044] In some situations, anatomical structures that identify the articular surface may be absent. In such cases, the program code can infer the missing anatomical structure from the available image data 102. For example, if the program code is determining which implant to use on the patient's left side, and the desired articular surface is not present in image data 102, the program code may infer the missing surface based on the available data.

[0045] Returning to Figure 1, once the articular surface is identified by the data analyzer 142, the controller's processing can proceed to the model candidate designation module 144. In some embodiments, the model candidate selection module 144 can use the articular surface identified from 142 to provide limits (as distance, perimeter, and other parameters) for selecting implant model candidates to fill the implant space. Module 144 can leverage the model selection module 146 to select / identify the best-fit anatomical implant model from the model library 148, which is a library / database of anatomical implant models. The models in the model library 148 may include models of the actual human talus, and / or models at least partially specified / manipulated by the user, as described below. In the example, the model selection module 146 uses the ML models described herein to find the nearest point in an n-dimensional latent space to select the best-fit implant model present in the library 148.

[0046] Optionally, the operation can be applied to a selected implant model as described herein, and such operation is facilitated by the model operation module 152.

[0047] In either case, the model candidate designation module is used to specify implant model candidates to be validated. Validation determines whether the candidate is appropriate for use with the specific patient from whom the image data 102 was taken. For example, the user selects a best-fit model identified by 146 as a candidate, and this selected candidate is provided for validation. Optionally, the selected best-fit model returned by 146 is presented to the user in an interface for viewing, and the user can select such a best-fit model as the first selected implant model. Alternatively, a best-fit model is automatically selected (e.g., by 146 or 144) and optionally confirmed by the user. Feedback in the form of user selection or confirmation regarding the best-fit design for a given input can, if desired, be used by the program code to update the machine learning algorithm used to select the best-fit design from the model library.

[0048] Once a model candidate is specified, it is provided to the model validation module 150. The validation module 150 receives the model candidate and attempts to validate it. Validation may be manual, automated, or a combination of both. In certain examples, the user performs at least part of the validation by digitally identifying, for example, using a controller, whether the implant fit is appropriate or if any adjustments are needed. Additionally or alternatively, in some embodiments, validation is automated at least partially (e.g., using a machine learning model) to validate whether the selected best-fit implant model (which may also be subject to manipulation) has an acceptable fit. Validation can determine whether the physical implant specified by the implant model candidate will pass surgical implantation in the patient. Validation may include one or more desired validation processes, tests, checks, etc., performed with respect to the selected implant model or collection of candidate selections.

[0049] In one example, validation may involve an assessment (fully manual, fully automated, or a combination thereof) of the distance between a selected implant and various patient anatomical structures when the implant is in the surgical position. For example, it may be desirable to examine the average size of the gap between the implant surface and the corresponding articular surface of the patient's anatomical structure that forms an interface with the implant surface, based on surface mapping. In this example, the desired gap size may be 0–3 millimeters, or more specifically, 1–2 millimeters. Generally, spatial and / or volumetric analysis may be useful in determining the fit between the implant and the surrounding anatomical structure. In another example, validation may utilize motion simulations in which the implant model is placed within the patient to simulate and evaluate how the fit is revealed by the patient's movements. For example, a gait cycle or other motion simulation can simulate the movement, shifting, etc., of the implant based on the simulated user's movements. This can provide insight into the suitability of a selected implant for a particular patient.

[0050] In particular, if a patient has contracted an infection or removed an anatomical structure due to other events before a replacement plan is developed, changes in the surrounding anatomical structure (such as soft tissue) that alter the characteristics of the space where the implant will be placed may be considered before the implant selection is finalized. For example, if soft tissue begins to grow into the space left by the removed anatomical structure, joint distraction may be desired, even if not to the extent that a talus implant sized to match the patient's natural anatomical structure would be appropriate. The talus implant may be selected to be smaller than the original talus but larger than the space currently to be occupied, in order to achieve the desired level of joint distraction. Verification in this context can confirm whether the implant is likely to achieve the desired amount of distraction.

[0051] Validation may incorporate the opinions of physicians and other medical professionals. For example, validation may be based on responses from a collection after presenting selected implant models in combination with specific patient anatomical structures / structures. In one embodiment, a physician's collection may be used to vote on which implant-anatomical structure pairs are appropriate. In a specific example, a collection of implant models is presented (e.g., in grid form) as candidates for one or more corresponding patient anatomical structures, and physicians are asked to indicate which ones show appropriate fit. Responses regarding the appropriate fit of implants can be collected in the form of crowdsourced input. The proposed implant-anatomical structure combinations may be presented in any desired way, including cross-sectional or other static diagrams, and / or as the motion simulations described above. In some examples, the responses are used to train a machine learning model to validate the selected implant model against a given anatomical structure. Additionally or alternatively, the responses may inform of statistical "rules" (tolerance ranges, mean values, etc.) useful for implant validation.

[0052] If a candidate model is validated, the physical implant model selector / generator 154 is used to select an existing physical implant (e.g., a commercially available product) if its manufacturer / model can be identified from the specifications, or to start model generation / manufacturing by passing the implant specifications to the implant fabrication device 160 in order to manufacture the implant. These specifications can instruct the creation of the implant. For example, in some embodiments, specifications in digital format can be provided to the implant fabrication device 160, which may be, for example, a 3D printer or other additive manufacturing equipment or its support and control device. The implant fabrication device 160 is controlled by program code of a controller and / or other computer system and can automatically create implants.

[0053] If a candidate is not validated, the process can return to the model candidate designation module 144 via the manipulation module 152. The manipulation module 152 allows for arbitrary manipulation of the model as part of the designation of a model candidate. For example, the user can optionally manipulate a model candidate, such as a model selected from a model library, and / or a model candidate that failed validation. Manipulation may be performed if the user and / or process identifies that a particular model candidate is not sufficient for a particular patient. For example, the user or process may observe that the model candidate is unacceptably small for the space it occupies (e.g., exceeding one or more thresholds), or that the area of ​​the implant that contacts the articular surfaces of surrounding anatomical structures is not satisfactorily aligned with those surfaces. Additionally or alternatively, the user may want to induce anatomical changes, such as joint separation, by selecting an implant. Therefore, the manipulation module can be used for the user to manually adjust the size, shape, and other characteristics of the implant model to better fit the model to the patient. Optionally, such adjustments may be performed automatically in some embodiments.

[0054] Instead of, or in addition to, manipulation of the implant model, the manipulation module 152 may manipulate the size, location, distances between features, and / or other properties of the patient's anatomical structure as reflected by the input data 102, thereby being used to modify the patient's anatomical structure on which model selection and / or validation are based. This includes changes to the properties of the articular surfaces (size, location, distances between features, etc.) identified from 142. By modifying the patient's anatomical structure, better implant model candidates may be identified, whether they are implant model candidates presented as implant candidates to the validation module 150 and which failed validation in the validation module 150, or implant model candidates identified after updating and searching the model library 148 via the selection module 146. Furthermore, implant candidates that initially failed validation may pass without further manipulation of the implant model itself if appropriate manipulation of the patient's anatomical structure is applied. Since the patient's anatomical structure may naturally adapt to the implant, partial manipulation of the patient's anatomical structure may be acceptable in appropriate circumstances.

[0055] In some examples, the library model selection module 146 selects several implant models from the library that are the “best-fitting” models for a given patient’s anatomical structure. The selection module 146 may rank these multiple best-fit models by confidence, or by an indication of how well the ML models used by the selection module 146 will fit them. The “best” model in such a ranking can be selected automatically or manually by the user as the initial selected implant model. This initial selected model can then be sent directly to the validation module 150 for validation, effectively serving as the initial implant model candidate to attempt validation of the selected implant model. If this candidate is rejected, the process may return to the model candidate designation module 144 to select the next best-fit model in the ranking, as an operation on the next candidate and / or implant model candidate, the patient’s anatomical structure, or both. Automated or user operations via 152 may or may not be performed in each iteration, and either one or both may be optional. A concrete example of the process is to return to 144 without modifying the candidate and select the next ranked model from among the multiple models selected by the library model selection module 146 as the next candidate to try. However, as mentioned above, the user may apply operations to the selected / implant model candidate in any iteration or when the next model is selected. The user may have criteria for understanding whether an operation on an anatomical structure or a model candidate that failed to validate might lead to a successful validation, or whether it would be better to select a different model from the library as the next model instead. AI may also be used in this decision-making, for example, to identify situations where working with the current model and applying operations to the current model and / or anatomical structure might be better than switching to the next selected model selected by, for example, library 148 to 146.

[0056] The model candidate designation module 144 is used to designate model candidates for validation. Candidates for validation may remain in the library 148 and / or may be manipulated automatically or by the user before being sent for validation. If a model candidate that passes validation (i) does not exist in library 148, or (ii) exists in the library but has not been previously validated for the anatomical structure corresponding to the anatomical structure of the validated patient, this can be useful training feedback for the system. For example, a designation of a validated implant model that is not yet in library 148 can be added to the library (151). The manipulated model may not represent the corresponding natural anatomical structure from the patient, but nevertheless, it has been found to be valid for a particular patient's anatomical structure and validated, so it can be added to library 148 as a valid implant model. Next, or if an implant model was previously known in the library but was not related to the anatomical structure of that patient or the anatomical structure corresponding to the procedure applied thereto, this can be provided as further training data for training an ML model used by the library model selection module 146 to identify the best-fit implant model for a given patient's anatomical input (e.g., articular surface characteristics).

[0057] In some cases, the initial model candidate is selected from the library as the best fit, and the user or system may manipulate the model as an initial adjustment before attempting validation. In either case, the process according to Figure 1 can be repeated for candidates and / or manipulations until a valid model is identified, after which a physical implant is obtained / manufactured and used. In this way, the process can be iterated until a suitable and effective implant model candidate is identified. This process may include manipulations of the patient's anatomical structure as reflected by the input data, and / or implant models placed for validation. Manipulations of implant models may result in the definition of new implant models that are not in the library. In this case, assuming the new model has been validated against a patient's anatomical structure, the new model can be saved in the library for later selection. Alternatively, the model can be provided as an example for further training, along with the characteristics of the corresponding articular surfaces that were validated.

[0058] In the example, a graphical user interface (GUI) is provided, which the user invokes the processes described herein, including the selection and specification of implant models based on one or more best-fit options selected by program code (e.g., Library Model Selection Module 146) that applies one or more trained machine learning algorithms. The GUI may allow the user to upload digital files constituting image data (e.g., Figure 1, 102) and specify specific search criteria and / or search methods to use. The GUI can invoke the processes described herein to select and present best-fit models to the user. Based on these search results, the user can select a predetermined result as a selected implant model to provide to Model Candidate Designation Module 144, and / or the next selected implant model to provide to Model Candidate Designation Module 144 may be automatically selected. The user may have the option to perform operations on the anatomical structures of the model and / or patient (e.g., including articular surfaces).

[0059] The program code can search based on one or more articular surfaces and display the best-fitting model itself, from which the user can select the next implant model to offer as a candidate. In other examples, the program code retrieves image data and automatically searches for / selects a model.

[0060] Figure 6 shows an exemplary interface presented by such a GUI for displaying and selecting best-fit implant models, according to the embodiments described herein. Element 602 presents a 2D rendering of the articular surface of the scaphoid bone of the patient's existing anatomical structure, i.e., the portion of the patient's scaphoid bone surface that is expected to physically engage with the talus implant. The properties of this surface may be input to the ML model of the library model selection module 146, or may be utilized by the library model selection module 146. The library model selection module 146 determines one or more (five in this example) best-fit models, indicated by 604, which present 2D renderings of the scaphoid engagement surfaces of the top five best-fit models.

[0061] Figure 6 shows a 2D rendering of a surface, which is one example of how to display the results of selecting input and library models. In this regard, when using 3D model data, neural networks and other ML models for AI may not perform at their best. In such cases, the 3D surfaces representing articular surfaces used as input to the model can be provided as 2D projections. Figure 602 is a 2D projection of the 3D scaphoid articular surface of a patient. Using different colors, shades, numerical values, or other data, a third "depth" dimension can be represented where each point corresponds to the distance between the point on the articular surface and the plane behind the surface. In the examples of Figure 602 and 604, the points "closest" to the 2D plane are shown with the darkest shadows, and the points furthest from the 2D plane are shown with the lightest shadows. By representing 3D articular surfaces as 2D data, AI can handle the properties of the articular surfaces more efficiently. This 2D rendering may be optional, as other existing or future AIs may be able to handle 3D data directly.

[0062] As mentioned above, 604 presents the top five best-fit models, represented by the surface that engages with the patient's navicular bone. The user can select one of these results, and the GUI can display the selected implant along with the surrounding anatomical structures. Figure 606 shows the anterior-lateral and posterior-medial views of the patient's ankle incorporating the selected implant model 608.

[0063] Therefore, the program code in Figure 1 can identify the articular surfaces of the patient's anatomical structures that will engage with the implant. Based on the characteristics of these articular surfaces, the library model selection module 146 can determine the best-fit model of the implant selected from the library. To select the best-fit model (e.g., one expected to engage most appropriately with the identified articular surfaces of the patient's anatomical structures), the library model selection module 146 can utilize classifiers, including image classifiers such as artificial neural networks (ANNs), convolutional neural networks (CNNs) (e.g., Mask-RCNNs), autoencoder neural networks (AEs), deep convolutional networks (DCNs), and / or other image classifiers and / or segmentation models, as well as combinations thereof. In the example using a CNN, the CNN can be constructed using an AI instruction set (e.g., a native AI instruction set or an appropriate AI instruction set). In the specific example herein, the classifier used is a deep learning model. The nodes and connections of the deep learning model can be trained and retrained without redesigning their number, arrangement, interface with image input, etc. In some examples, these nodes collectively form a neural network. In certain embodiments, the classifier nodes do not have a layer structure. To construct a neural network, program code can, for example, connect the layers of the network, define skip connections between network layers, set coefficients (e.g., convolution coefficients) to learned values, set the filter length, and determine the patch size (e.g., an n x n pixel region of an image, where n is 3, 4, etc.).

[0064] In some embodiments, the program code utilizes a neural network to select a model from a model library in real time, allowing the user providing input to a GUI to obtain results almost instantly. In some examples, the program code includes a pre-trained CNN configured to classify aspects of images obtained by the program code. In some examples, the CNN may be MobileNet or MobileNet2 and may be available in Keras, TensorFlow, and / or other types of libraries. The classifier of the model selection module can, in embodiments, be generated by the program code from a pre-trained CNN by, for example, deleting, modifying, and / or replacing layers of neurons (e.g., input and output layers).

[0065] In accordance with the embodiments described herein, program code such as the program code for the library model selection module 146 can implement an ML model trained to learn key features of anatomical structures, particularly articular surfaces, and to identify one or more best-fit designs from a model library. In some examples, the program code includes a deep neural network, which is trained to learn the features of designs in a model library 148, including articular surfaces that form corresponding pairs, in order to appropriately classify the input.

[0066] Figure 7 shows a conceptual diagram of training and using a machine learning model according to the embodiments described herein. The training dataset 702 is used to train the ML model 704. The training dataset 702 may include models of exemplary adjacent anatomical structures (e.g., articular surfaces) and / or their properties, along with correlated anatomical implant models that "fit" to their adjacent anatomical structures. Thus, using the examples described herein, the training dataset may include models of talus bones (which may be models of actual human talus bones) and models (e.g., 3D and / or 2D) of the articular surfaces to which those talus bones are paired. The properties of the articular surfaces and the talus bones paired with them can be determined, for example, by preprocessing image data of actual patients.

[0067] ML model 704 can be any type of ML model, one example being an autoencoder, a type of neural network. By training ML model 704, it learns to identify a talus model that fits a given input articular surface. Such a talus model can correspond to an existing implant or an implant to be manufactured. ML model 704 can be trained with any amount of training data, but generally, a larger number of training examples leads to higher accuracy than a smaller number of training examples. In a particular example, ML model 704 may be trained to obtain acceptable results based on approximately 200 training cases (articular surface-bone model pairs).

[0068] Generally, training involves training the ML model 704 to identify the most relevant patterns in the data so that, when the characteristics of the articular surfaces adjacent to a given target anatomical structure for replacement are input to the model, the model returns an appropriate implant model for the anatomical replacement. Dimensionality reduction to a “lower-dimensional space” picks up the most representative patterns in the anatomical structure as features to look for when the input 706 is provided. In the example, the input data may be a 3D model of the anatomical structure, which is converted to 2D data (e.g., projection as described above) and supplied to the ML model 704 as input 706. Model 704 is trained to identify implant models from a model library 708 that fit a given input. The model library 708 may contain implant models corresponding to examples in the training dataset, and / or other implant models. The ML model 704 provides zero or more selected models as output 710. If multiple models are identified as “best fit” implants for a given input, they may or may not be ranked.

[0069] The models in model library 708 and any "hits" based on the input can be represented by an "n-dimensional latent space" 720. While the example in 720 shows three dimensions, in many real-world examples the latent space is n-dimensional, where n is greater than 3. Each point in the latent space represents a possible output label, for example, implant models in these examples. Thus, each talus model may be represented as a specific point in the latent space. In this example, the two dots 721 and 722 represent two "hits," best-fit implant models, based on a hypothetical input surface fed into the trained ML model.

[0070] In some embodiments, the ML model may be provided as a 3D Generative Adversarial Network (GAN) structure. In the GAN approach, a discriminator is trained to distinguish between real / actual samples and fake / generated samples generated by the generator. The generator, on the other hand, is trained to generate better samples. Once the generator is sufficiently trained, it can be used as a classifier / ML model for input anatomical features. In the context of the embodiments described herein, the generator can thereby provide a method for selecting an implant model, either by selecting a best-fit model from an existing library or by attrition to generating an implant model (e.g., a whole talus model) from scratch. The generator effectively learns from patterns present in the input, with the goal of generating an anatomical structure (e.g., a talus) that a human would naturally generate for a given anatomical structure. The results generated by the generator are a point cloud. The point cloud can be processed using various filtering and interpolation techniques to generate closed surfaces and ultimately a continuous volume representing a complete anatomical structure such as a talus.

[0071] Figure 8 shows an exemplary process flow for artificial intelligence-based implant selection according to the embodiments described herein. All or part of this process can be performed by a computer system executing program code. This process identifies a first portion (e.g., the talus) of an anatomical structure in a first anatomical region (e.g., the ankle / lower limb) to be replaced by an implant (802). Optionally, the process may determine (e.g., based on queries or other processing) that the contralateral side of the patient (e.g., having a second portion of the anatomical structure in a second anatomical region) alone does not provide sufficient image data to form the basis for the desired implant that will replace the first portion of the anatomical structure (804). In this case, the process acquires image data of the first portion of the anatomical structure and the first anatomical region (806) and generates a 3D model incorporating the first portion of the anatomical structure and the first anatomical region (808). This process involves manipulating a 3D model to hide a first portion of an anatomical structure relative to a first anatomical region (810), then identifying and isolating the articular surfaces of components of the first anatomical region (e.g., tibia, fibula, talonavicular joint, calcaneus) with which the first portion of the anatomical structure interacts (812). The process can generate a partial volume based on the articular surfaces (814), i.e., the articular surfaces indicate a portion of the periphery of the volume / implant to occupy space. The partial volume represents limits with respect to at least some limits / dimensions of the implant. The process then selects one or more implant models of the first portion of the anatomical structure that best fit the partial volume from a library of models (816). The process can then fit one or more of the thus selected models for desired engagement with one or more components of the first anatomical region (818), these components may be articular surfaces and / or other components of the patient's anatomical structure.Ultimately, the model specification may be validated, at which point the process generates an implant specification file based on the validated model specification (820), sends the implant specification file to a 3D printer or other implant manufacturing device (822), and the implant manufacturing device produces the implant (824). Alternatively, the validated model specification may be selected and obtained from an existing physical implant that it corresponds to. In any case, once the physical implant is obtained, it can be surgically implanted into the patient.

[0072] Figure 9 shows an example of a graphical user interface (GUI) for software to perform the embodiments described herein. The GUI can be used, for example, as part of at least the library model selection module 146 in Figure 1 to identify and present best-fit models to the user for selection. Interface 900 is displayed when the software is loaded. The TOP-RESULTS section 902 includes options 904 for the user to select how many outputs (e.g., best-fit models) the ML model should select and provide as possible implant models. Options 904 can be configured with a default, for example, 5 as shown. In this example, option 904 is implemented as a number picker with an up / down selector.

[0073] Interface 900 also includes an anatomical structure type selection unit 906 that allows the user to select an anatomical surface (in this case, an articular surface) that the software should use when comparing models in the library with anatomical structure models loaded by the file input unit 908. In this example, the anatomical structure type selection unit 906 is implemented as three radio buttons corresponding to the calcaneal, navicular, or fibular articular surfaces, and the user is to select one of the three. In embodiments where multiple articular surfaces may be considered in the selection of the best-fit model, the surface type selection unit 906 may be implemented using different interface element types to allow the user to indicate multiple articular surfaces. In the example in Figure 9, the user has selected the calcaneal articular surface (and / or the program uses it by default).

[0074] The file input unit 908 allows the user to provide files to process. In this example, the software takes in .stl files of anatomical structures. The user selects the search button and browses the file system for a suitable .stl file containing a patient's anatomical structure model. Once the user selects a file, that file may be opened / loaded.

[0075] The user also selects a search method in the interface unit 910, which is implemented as three radio buttons corresponding to pre-trained, custom, and histogram search methods. In the pre-trained search method, the search utilizes an ML model (e.g., a convolutional neural network) trained on the dataset, and then a final layer is added with classes / labels corresponding to available implant models in the database. The training may be based on a desired anatomical structure, specific to an anatomical region (e.g., shoulder, ankle, knee), or more general (e.g., whole-body CT or scan). The custom search method may train the ML model in a more customized way for specific symptoms, deformities, etc. (i.e., a less general way than the pre-trained method). The histogram search method may instead be based on statistics and statistical modeling that compare the selected anatomical structure to "normal" (informed by the mean) anatomical structures, measured values, curvature, etc., stored in a database of normal anatomical structures. This allows us to understand the degree of correlation between the actual patient's anatomical structure and the average / typical anatomical structure, and to identify the desired characteristics of the patient's anatomical structure, and consequently, the characteristics of the implant that will help in selecting an appropriate implant.

[0076] Once these parameters are specified, the user selects the OK button, which causes the system to start processing and display the best-fit results. In this embodiment, the results can be output to a command-line interface, as shown in Figure 10, in this example identifying five results (#0 to #4) by the name of the implant model or other identifiers. Additionally or alternatively, the interface presents a 2D graphical rendering of the input articular surfaces and a 2D rendering of the identified articular surface profiles corresponding to the five best-fit results, as shown in Figure 11.

[0077] Thus, based on the search results selected by the program code after the search has been performed, the software can display one or more best-fit results. In some examples, the user selects a predetermined result. One or more of these results can be used in combination with other embodiments described herein, such as the embodiment in Figure 1, like the model candidate designation module 144. In embodiments in which a search is performed based on multiple articular surfaces, different sets of best-fit results may be provided, each corresponding to one of the articular surfaces. Additionally or alternatively, the software may also display a full model from which the user can select.

[0078] Figure 12 shows an example of the process for implant selection according to the embodiments described herein. In some examples, this process is performed by one or more computer systems as described herein.

[0079] This process obtains a machine learning (ML) model trained to select an anatomical implant model of a physical implant for partial or complete replacement of a target anatomical structure based on the anatomical surface characteristics of the anatomical structures adjacent to the target anatomical structure to be partially or completely replaced (1202). The method also obtains image data of an anatomical region of a particular patient (1204). The anatomical region includes the patient's target anatomical structure and other anatomical structures of the patient, the other anatomical structures being adjacent to the patient's target anatomical structure. In the example, the target anatomical structure is the patient's talus, and the other anatomical structures are the navicular bone, calcaneus, tibia and / or fibula.

[0080] This process proceeds by determining the properties of at least one anatomical surface of another anatomical structure from the image data (1206). The at least one anatomical surface is located at at least one interface between the patient's other anatomical structure and the anatomical structure of interest. In an example, the anatomical surface is an articular surface. The anatomical surface may include articular surfaces into which a physical implant will engage, based on being at least partially surgically implanted within the patient. In a particular embodiment, the anatomical region has the patient's ankle, the anatomical structure of interest has the talus, and the at least one anatomical surface has at least one articular surface of at least one bone adjacent to the talus, such as the articular surfaces of the navicular, calcaneus, tibia and / or fibula of each bone surrounding the patient's talus.

[0081] The user may manually identify the surface, and / or a process, for example, to which a machine learning model is applied, may automatically identify the surface. Furthermore, the user may optionally refine the identified surface. Thus, determining the characteristics of at least one anatomical surface (1206) can be based on (i) manual indication by the user of at least one anatomical surface provided based on user input to a graphical user interface displaying a model including at least other anatomical structures of the patient, and / or (ii) automated analysis of image data to confirm at least one anatomical surface.

[0082] In one embodiment, preprocessing is performed to convert 3D data to 2D data. Thus, the obtained image data may include 3D digital model data representing an anatomical region of the patient, and determining the properties of at least one anatomical surface of other anatomical structures (1206) may include processing the image data to present at least one anatomical surface as at least one digital 3D surface, and converting at least one digital 3D surface to at least one 2D projection, where the determined properties of at least one anatomical surface are determined from at least one 2D projection.

[0083] Referring to Figure 12, the process involves applying an ML model using the determined properties of at least one anatomical surface (1208) and obtaining a selected implant model based on the application (1210). The selected implant model is chosen by the ML model as a designation for a physical implant for at least partial surgical transplantation candidates within the patient, as a partial or total replacement of the patient's target anatomical structure.

[0084] ML models can be trained using samples from a library of implant models. This training allows the ML model to select anatomical implant models from the implant model library, and thus the selected implant model can be selected by the ML model from the implant model library.

[0085] Alternatively, the ML model could include a trained generator of a generative adversarial network, which is trained using samples from a library of implant models and trained to generate implant models. In this case, the selected implant models could include those generated by the generator and selected by the generator to be output as the selected implant models.

[0086] The selected implant model may or may not be appropriate for the patient. Therefore, the process proceeds by providing the selected implant model to a model candidate designation module to specify a model candidate for validation (1212). The model candidate designation module can process the selected model for, for example, possible operations and / or other tasks. In this regard, the selected implant model may or may not be the one initially presented for validation. Furthermore, the selected implant model can be presented to the user on a graphical user interface. Optionally, the image data can be preprocessed to create a three-dimensional digital model of the patient's anatomical region with the target anatomical structures omitted, and this can be presented to the user for display.

[0087] This process also optionally includes receiving an operation (1214). The user may wish to operate on the selected model at this point (or after validation has failed), which will result in a different model specification and, therefore, a different specification of the physical implant that the model indicates. Thus, in one embodiment, receiving an operation (1214) may include receiving an operation on the selected implant model, the operation modifying the selected implant model and generating implant model candidates for validation. Based on such an operation, the implant model candidates specify a physical implant with one or more physical properties different from the physical implant specified by the selected implant model initially selected and provided prior to the user's operation. As an example, the operation on the selected implant model may include an operation specified by the user and / or an operation automatically determined by artificial intelligence and optionally applied automatically.

[0088] Additionally or alternatively, the user may wish to apply an operation to a representation (e.g., surface) of a patient's anatomical structure. Thus, receiving an operation (1214) may include receiving an operation on the properties of at least one anatomical surface of other anatomical structures. For example, the user may change the shape or other properties of a surface, and / or the distance between surfaces. When the user moves a bone or other anatomical structure, the position of the anatomical surface of that bone may change. An operation on the properties of at least one anatomical surface may include operations specified by the user and / or operations that are automatically determined by artificial intelligence and optionally applied automatically.

[0089] Regardless of whether the operation has been applied, at this point, the implant model candidate is obtained, as obtained and provided by 1210 / 1212, or after the operation. This process proceeds by providing the implant model candidate to the validation module for validation (1216). Validation determines whether the physical implant specified from the implant model candidate will pass for surgical implantation in the patient. Since the validation is performed on the presented candidate, if an operation has been performed, the operation will be taken into consideration. Thus, the validation determining whether the physical implant specified by the implant model candidate will pass for surgical implantation in the patient will be performed, at least in part, on any manipulated properties of the anatomical surface and / or on the operation performed on the implant model itself.

[0090] The process determines at 1218 whether the implant model candidate passes validation, i.e., whether the physical implant specified by the implant model candidate is suitable for surgical implantation in the patient. If not (1218, N), the process can return to 1214, where the candidate implant selection module may again receive the candidate that failed validation and / or the operation on the patient's anatomical structure. This provides information for the next candidate, which may be a modified or unmodified version of the first failed candidate, and this next candidate is provided for validation at 1216. In this respect, the failing candidates are essentially the next iteration of the “selected implant model” initially attached to the implant candidate designation module. For each iteration, the failed implant model candidate is provided as the “selected implant model” for the next iteration of the iteration, the operation is optionally received, and the next candidate is acquired and provided to the validation module for validation. The iteration may be repeated any number of times until a candidate is validated.

[0091] If an implant model candidate fails (1218, N), instead of returning to 1214, the process returns to 1220 to obtain the (next) selected implant model, and work can begin from that next model. This is especially possible if the ML model has identified multiple best-fit implant models. Additionally or alternatively, if the user has manipulated the patient's anatomical structure as described above, it may be desirable to reapply the ML model (returning to 1208) to see if another best-fit implant from the library fits better.

[0092] In specific cases where the presented implant model is sent directly for validation (e.g., when no operation is created, and / or when the selected implant model chosen by ML is provided directly for validation as an initial task), the selected implant model is provided to the validation module as an implant model candidate for validation. In this case, the implant model candidate is the initial implant model candidate. Based on the validation that the physical implant specified by the initial implant model candidate does not pass (1218, N), the process returns to 1214. Next, the process receives an operation on the initial implant model candidate at 1214, which modifies the initial implant model candidate and generates a different implant model candidate for validation, the different implant model candidate specifying a physical implant having one or more different physical properties compared to the physical implant specified by the initial implant model candidate and / or (ii) an operation on the properties of at least one anatomical surface of another anatomical structure. This process determines the next implant model candidate to provide to the validation module for validation as part of providing implant model candidate 1216. Based on the receipt of the operation on the initial implant model candidate, the next implant model candidate to provide is determined to be a different implant model generated from the operation on the initial implant model candidate. However, based on receiving manipulation on the properties of at least one anatomical surface of other anatomical structures and not receiving manipulation on the initial implant model candidate, the next implant model candidate is determined to be the initial implant model candidate. Thus, the next implant model candidate may be a failed initial implant model candidate or not. As described, this next implant model candidate is provided to a validation module for validation to determine whether the physical implant specified by the next implant model candidate will pass surgical implantation in the patient (1216), and the validation is at least partially based on the manipulation received on the initial implant model candidate and / or on the properties of at least one anatomical surface.

[0093] Alternatively, in 1218, if the candidate is validated (1218, Y), the process provides the implant model candidate as the designation to be used by the physical implant (1220). At that point, the physical implant may be obtained if available, and may be manufactured if desired.

[0094] If an implant model candidate is an manipulated version of a selected implant model chosen from the library, and that manipulated version does not exist separately in the library, it can be added to the library as a valid implant model. Additionally or alternatively, an implant model candidate can be presented in the training dataset as part of a training example that associates the implant model candidate with at least one anatomical surface of another anatomical structure, in order to associate the implant model candidate with an anatomical structure as a suitable model for that structure. Furthermore, even if an implant model candidate already exists in the library, but in the training example it is not associated with the anatomical structure of a particular patient that was validated (e.g., the patient's anatomical structure is not reflected in the existing training example, and / or the user manipulates the actual patient's anatomical structure to generate a virtual patient's anatomical structure that is not yet reflected), the implant model candidate may still be presented in the training dataset as part of a training example that associates the implant model candidate with the patient's anatomical structure, in order to associate the implant model candidate with an anatomical structure as a suitable model for that structure.

[0095] In this specification, terms such as “connection,” “linking,” “contact,” and “joining” are defined broadly to encompass a variety of arrangement and assembly techniques. These arrangements and techniques include, but are not limited to, (1) direct joining of one component to another without intervening components (e.g., parts are in direct physical contact) and (2) joining of one component to another with one or more components intervening, provided that one component that “connects,” “contacts,” or “joins” the other component is operationally linked to the other component in any way (e.g., electrically, fluidly, physically, optically, etc.) (despite the presence of one or more additional components between them). It should be understood that some components that are in direct physical contact with each other may or may not be in electrical and / or fluid contact with each other. Furthermore, two electrically connected, electrically coupled, optically connected, optically coupled, fluidly connected, or fluidly coupled components may or may not be in direct physical contact, and one or more other components may be positioned between them.

[0096] As used herein, the terms “contains” and “possess” mean the same thing.

[0097] The terms “substantially,” “about,” “approximately,” “relatively,” or other similar terms that may be used throughout this disclosure, including the claims, are used to describe and explain small fluctuations from a standard or parameter, such as variations in processing. Such small fluctuations include zero fluctuations from the standard or parameter. For example, they may refer to ±10% or less, e.g., ±5% or less, e.g., ±2% or less, e.g., ±1% or less, e.g., ±0.5% or less, e.g., ±0.2% or less, e.g., ±0.1% or less, e.g., ±0.05% or less. As used herein, terms such as “substantially,” “about,” “approximately,” and “relatively” may also mean no variation.

[0098] As used herein, “electrically coupled” refers to the transfer of electrical energy between any combination of power sources, electrodes, conductive surfaces, droplets, conductive traces, wires, waveguides, nanostructures, and other circuit segments. The term electrically coupled may also be used in relation to direct or indirect connections, which may be mediated through various media such as fluid media or air gaps.

[0099] As used herein, “neural network” refers to a biology-inspired programming paradigm that enables computers to learn from observational data. This learning is called deep learning and is a set of learning techniques in neural networks. Neural networks, including modular neural networks, enable pattern recognition with speed, accuracy, and efficiency in situations where datasets are multiple and wide-ranging, including the entire distributed network of the technological environment. Modern neural networks are nonlinear statistical data modeling tools. They are typically used to model complex relationships between inputs and outputs or to identify patterns in data (i.e., neural networks are tools for nonlinear statistical data modeling or decision-making). Generally, program code that utilizes neural networks can model complex relationships between inputs and outputs and identify patterns in data. Due to the speed and efficiency of neural networks, especially when analyzing multiple complex datasets, neural networks and deep learning offer solutions to many problems in image recognition, speech recognition, and natural language processing. Neural networks can model complex relationships between inputs and outputs for classification and identify patterns in data, including images. Due to their high speed and efficiency, particularly when analyzing multiple complex datasets, neural networks and deep learning offer solutions to many problems in image recognition that would otherwise be impossible. As will be discussed later, in some embodiments of the present invention, neural networks are used to learn various features of talus implant design, including but not limited to focusing on articular surfaces.

[0100] As used herein, “Convolutional Neural Networks” (CNNs) are a class of neural networks. CNNs utilize feedforward artificial neural networks and are most commonly applied to the analysis of visual images. CNNs are so named because they utilize convolutional layers that apply a convolution operation (a mathematical operation performed on two functions to generate a third function representing how the shape of one function is modified by the other) to the input and pass the result to the next layer. The convolution mimics the response of individual neurons to visual stimuli. Each convolutional neuron processes only the data in its receptive field. Processing images using general (i.e., fully connected feedforward) neural networks is impractical because the input size associated with images is so large that it requires a very large number of neurons. This problem can be addressed by using CNNs because, to support the learning process, CNNs fine-tune a large number of parameters and a vast pre-labeled dataset, allowing a consistent number of learnable parameters to be available regardless of image size. This reduces the number of free parameters and allows the network to be deeper with fewer parameters. CNNs solve the problems of vanishing gradients and explosions in training traditional multilayer neural networks by using backpropagation. Therefore, CNNs can be used in large-scale (image) recognition systems, yielding state-of-the-art results in segmentation, object detection, and object retrieval. While CNNs can be multidimensional, most existing CNNs are two-dimensional and process single images. These images contain pixels in a two-dimensional (2D) space (height, width) and are processed through a series of two-dimensional filters to understand the set of pixels that best corresponds to the final output classification. Three-dimensional CNNs (3D-CNNs) are an extension of more traditional two-dimensional CNNs and are commonly used in problems related to video classification. 3D-CNNs accept multiple images, often consecutive image frames from a video, and use 3D filters to understand a three-dimensional set of presented pixels. In the current context, as described herein, the images provided to the CNN include anatomical images of patients.

[0101] As used herein, the “classifier” comprises various cognitive algorithms, artificial intelligence (AI) instruction sets, and / or machine learning algorithms. Classifiers include, but are not limited to, deep learning models (e.g., neural networks with many layers) and random forest models. The classifier categorizes items (data, metadata, objects, etc.) into groups based on the relationships between data elements contained in metadata from the records. In some embodiments of the present invention, program code can identify and filter false positives by utilizing the frequency of feature occurrences in mutual information. Generally, program code uses the classifier to create a boundary between first-quality data and second-quality data. The accuracy of the classifier improves through continuous use and testing. When training the classifier, in some examples, program code feeds one or more cognitive analysis algorithms to be trained an existing set of features describing metadata and / or data features. The program code trains the classifier to categorize records based on the presence or absence of predetermined conditions known before tuning. The presence or absence of conditions is not explicitly stated in the records of the dataset. When classifying sources as providing data for a given condition (based on metadata), a classifier can be used so that program code indicates the probability of a given condition on a scale between 0 and 1, where 1 indicates certain presence. The classification does not have to be binary and may be a value on an established scale. As disclosed herein, in some examples, a classifier is used to select the optimal talus from a database based on program code that cognitively analyzes the anatomical structure of a patient.

[0102] In this specification, the term "deep learning model" refers to a type of classifier. Deep learning models can be implemented in various forms, such as neural networks (e.g., convolutional neural networks). In some examples, a deep learning model includes multiple layers, each layer having multiple processing nodes. In some examples, the layers are processed sequentially, with nodes in layers closer to the model input layer being processed before nodes in layers closer to the model output layer. In this way, the layers feed the next layer. Internal nodes are often "hidden" nodes in the sense that their input and output values ​​are not visible from outside the model.

[0103] As used herein, the term “Conditional Generative Adversarial Network” (“cGAN”) refers to a Generative Adversarial Network (GAN), a machine learning framework used to train generative models. Specifically, cGANs are used to conditionally generate images. A GAN relies on a generator that learns to produce new images and a discriminator that learns to distinguish between synthetic and real images. In a cGAN, conditional setting is applied, meaning both the generator and the discriminator are conditioned on auxiliary information from other modalities. In this way, a cGAN can learn a multimodal mapping from input to output by being supplied with different contextual information. In the examples herein, the contextual information can include both anatomical data from patients and a library of existing talus implants.

[0104] As used herein, the term “processor” means a hardware and / or software device capable of executing computer instructions, including, but not limited to, one or more software processors, hardware processors, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), and / or programmable logic devices (PLDs).

[0105] While various examples are provided, they can be modified without departing from the spirit of the claims.

[0106] The processes described herein may be carried out individually or collectively by one or more computer systems. Figure 13 shows an example of such a computer system and associated apparatus for incorporating and / or using the embodiments described herein. A computer system may be referred to herein as a data processing device / system, a computing device / system / node, or simply a computer. A computer system may be based on one or more of various system architectures and / or instruction set architectures, such as those provided by Intel Corporation (Santa Clara, California, USA) or ARM Holdings plc (Cambridge, England, UK).

[0107] Figure 13 shows a computer system 1300 communicating with an external device 1312. The computer system 1300 includes one or more processors 1302, such as a central processing unit (CPU). A processor may include functional components used to execute instructions, such as functional components that invoke program instructions from locations such as cache or main memory, functional components that decode program instructions, functional components that execute program instructions, functional components that access memory for instruction execution, and functional components that write the results of executed instructions. The processor 1302 may also include registers used by one or more functional components. The computer system 1300 also includes memory 1304, input / output (I / O) devices 1308, and I / O interfaces 1310, which may be coupled to the processor 1302 and each other via one or more buses and / or other connections. Bus connections represent one or more of several types of bus structures, including memory buses or memory controllers, peripheral buses, accelerated graphics ports, and processor or local buses using any of the various bus architectures. Examples of such architectures include the Industry Standard Architecture (ISA), Micro Channel Architecture (MCA), Enhanced ISA (EISA), Video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnect (PCI).

[0108] Memory 1304 may be or include main memory or system memory (e.g., random access memory), storage devices such as hard drives, flash media, or optical media, and / or cache memory, for example, used for executing program instructions. Memory 1304 may include caches such as a shared cache that can be coupled to the local cache of processor 1302 (e.g., L1 cache, L2 cache, etc.). Furthermore, memory 1304 may be or include at least one computer program product having a set (e.g., at least one) of program modules, instructions, code, etc., configured to perform the functions of the embodiments described herein when executed by one or more processors.

[0109] Memory 1304 can store the operating system 1305 and other computer programs 1306, such as one or more computer programs / applications, which are executed to perform the embodiments described herein. Specifically, the programs / applications may include computer-readable program instructions, which may be configured to perform the functions of embodiments of the embodiments described herein.

[0110] Examples of I / O devices 1308 include, but are not limited to, microphones, speakers, Global Positioning System (GPS) devices, cameras, lights, accelerometers, gyroscopes, magnetometers, sensor devices configured to sense light, proximity, heart rate, body temperature and / or ambient temperature, blood pressure and / or skin resistance, and activity monitors. The I / O devices may be integrated into the computer system as shown in the illustration, but in some embodiments, the I / O devices may be considered as external devices (1312) coupled to the computer system via one or more I / O interfaces 1310.

[0111] The computer system 1300 may communicate with one or more external devices 1312 via one or more I / O interfaces 1310. Exemplary external devices include keyboards, pointing devices, displays, and / or any other devices that enable a user to interact with the computer system 1300. Other exemplary external devices include any devices that enable the computer system 1300 to communicate with one or more other computing systems or peripheral devices such as printers. Network interfaces / adapters are exemplary I / O interfaces that enable the computer system 1300 to communicate with one or more networks, such as local area networks (LANs), common wide area networks (WANs), and / or public networks (e.g., the Internet), and provide communication with other computing devices or systems, storage devices, etc. Ethernet-based (such as Wi-Fi) interfaces and Bluetooth® adapters are examples of currently available network adapters used in computer systems (Bluetooth® is a registered trademark of Bluetooth SIG, Inc., Kirkland, Washington, USA).

[0112] Communication between the I / O interface 1310 and the external device 1312 may occur via a wired and / or wireless communication link 1311, such as an Ethernet-based wired or wireless connection. Examples of wireless connections include cellular, Wi-Fi, Bluetooth®, proximity, short-range wireless, or other wireless connections. More conceptually, the communication link 1311 may be any suitable wireless communication link and / or wired communication link for transmitting data.

[0113] A specific external device 1312 may include one or more data storage devices that may store one or more programs, one or more computer-readable program instructions, and / or data. The computer system 1300 may include and / or be coupled with a removable / non-removable, volatile / non-volatile computer system storage medium, and / or communicate with the computer system storage medium (for example, as an external device of the computer system). For example, it may include and / or be coupled with a magnetic disk drive for reading from and writing to a non-removable non-volatile magnetic medium (typically called a “hard drive”), a removable non-volatile magnetic disk (e.g., a “floppy disk”), and / or an optical disk drive for reading from and writing to a removable non-volatile optical disk such as a CD-ROM, DVD-ROM, or other optical medium.

[0114] Computer system 1300 may operate in a number of other general-purpose or special-purpose computing system environments or configurations. Computer system 1300 can take any of the following forms, well known examples include, but are not limited to, personal computer (PC) systems, server computer systems such as messaging servers, thin clients, thick clients, workstations, laptops, handheld devices, smartphones, tablets and wearable devices and other mobile devices / computers, multiprocessor systems, microprocessor-based systems, telephony devices, network appliances (such as edge appliances), virtualization devices, storage controllers, set-top boxes, programmable home appliances, network PCs, minicomputer systems, mainframe computer systems and any of the above systems or devices, etc.

[0115] Aspects of the present invention may be systems, methods, and / or computer program products, any of which may be configured to perform or facilitate the aspects described herein.

[0116] In some embodiments, aspects of the present invention may take the form of a computer program product which may be implemented as a computer-readable medium. A computer-readable medium may be a tangible storage device / medium storing computer-readable program code / instructions. Examples of computer-readable media include, but are not limited to, electronic, magnetic, optical, or semiconductor storage devices or systems, or combinations thereof. Exemplary embodiments of a computer-readable medium include hard drives or other mass storage devices, electrical connections with wires, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory such as EPROM or flash memory, optical fibers, portable computer disks / diskettes such as compact disk read-only memory (CD-ROM) or digital versatile disk (DVD), optical storage devices, magnetic storage devices, or any combination thereof. A computer-readable medium may be readable by a processor or processing unit, etc., to retrieve data (e.g., instructions) from the medium for execution. In certain examples, a computer program product is or includes one or more computer-readable media containing / storing computer-readable program code in order to provide and facilitate one or more embodiments described herein.

[0117] As described above, program instructions contained in or stored on a computer-readable medium can be retrieved and executed by any of the various appropriate components of the computer system, such as the processor, in order to operate and function the computer system in a particular manner. Such program instructions for performing actions to execute, achieve, or facilitate the embodiments described herein may be written in any desired programming language or compiled from code written in any desired programming language. In some embodiments, such programming languages ​​include object-oriented programming languages ​​and / or procedural programming languages ​​such as C, C++, C#, and Java.

[0118] The program code may include one or more program instructions obtained for execution by one or more processors. The computer program instructions may, for example, be provided to one or more processors of one or more computer systems, and when executed by one or more processors, the program instructions manufacture a machine such that they perform, achieve, or facilitate aspects of the invention, such as operations or functions described in the flowcharts and / or block diagrams described herein. Therefore, each block, or combination of blocks, in the flowcharts and / or block diagrams depicted and described herein can be implemented by computer program instructions in some embodiments.

[0119] Although various embodiments have been described above, these are merely examples.

[0120] The terms used herein are for the sole purpose of describing specific embodiments and are not intended to be limiting. In this specification, the singular forms “a”, “an”, and “the” are intended to include the plural form unless the context clearly indicates otherwise. It will be further understood that, as used herein, the terms “have” and / or “possess” identify the presence of a described feature, integer, step, operation, element, and / or component, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0121] All structures, materials, actions, and equivalents (if any) corresponding to all means-plus-function or step-plus-function elements in the following claims are intended to include any structures, materials, and actions for performing the function in combination with other elements specifically claimed. Descriptions of one or more embodiments are presented for illustrative and explanatory purposes, but are not intended to be exhaustive or restrictive in the disclosed form. Many modifications and variations will be apparent to those skilled in the art. The embodiments have been selected and described to best illustrate various aspects and practical applications, and to enable those skilled in the art to understand various modifications suitable for specific intended uses.

Claims

1. A method that is carried out by computer, The computer obtains a machine learning model trained to select an anatomical implant model for the partial or complete replacement of the target anatomical structure based on the characteristics of the anatomical surface of the anatomical structure adjacent to the target anatomical structure to be partially or completely replaced. The computer acquires image data of an anatomical region of a patient, wherein the anatomical region includes the patient's target anatomical structure and other anatomical structures of the patient, and the other anatomical structures are adjacent to the patient's target anatomical structure. The computer determines the properties of at least one anatomical surface of the other anatomical structure from the image data, wherein the at least one anatomical surface is adjacent to the target anatomical structure of the patient, and the properties of the at least one anatomical surface are determined to be adjacent to the target anatomical structure of the patient. The computer utilizes the determined properties of the at least one anatomical surface, applies the machine learning model, and obtains an implant model selected based on the application, wherein the selected implant model is selected by the machine learning model as a designation of a physical implant for a candidate for surgical transplantation within the patient, at least partially, as a partial or total replacement of the target anatomical structure of the patient. Includes, The aforementioned method, The computer further includes training the machine learning model so that the machine learning model selects an anatomical implant model. Training involves using samples from a library of implant models and training the machine learning model to select anatomical implant models from the said library of implant models. The selected implant model is selected by the machine learning model from the library of implant models. method.

2. The method according to claim 1, further comprising the computer providing a selected implant model to a model candidate designation module in order to designate a model candidate for verification.

3. The computer presents the implant model selected by the user on a graphical user interface, The computer receives an operation on the selected implant model, the operation being to modify the selected implant model, generate implant model candidates for verification, and the implant model candidates specify one or more physical implants that differ in one or more physical characteristics from the physical implants specified by the selected implant model. The method according to claim 2, further comprising the computer providing an implant model candidate to a verification module for verification, the verification determining whether the physical implant specified by the implant model candidate is suitable for surgical implantation in a patient.

4. The procedure for the selected implant model is: At least one operation specified by the user, and The method according to claim 3, comprising at least one selected from a group including at least one operation automatically determined by artificial intelligence.

5. The computer further comprises, based on a verification that determines that a physical implant specified by an implant model candidate does not pass, (i) receiving an operation and (ii) providing the implant model candidate to the verification module for verification, one or more times, The method according to claim 3, wherein in each iteration, an implant model candidate that did not pass is provided as a selected implant model for the next iteration.

6. The computer further includes receiving operations on the properties of the at least one anatomical surface of the other anatomical structure, The method according to claim 3, 4, or 5, wherein the verification for determining whether a physical implant designated by an implant model candidate is suitable for surgical implantation in a patient is at least partially based on the manipulated properties of the at least one anatomical surface.

7. The manipulation of the properties of the at least one anatomical surface is, At least one operation specified by the user, and The method according to claim 6, comprising at least one selected from a group including at least one operation automatically determined by artificial intelligence.

8. The computer receives operations on the properties of the at least one anatomical surface of the other anatomical structure, The computer further includes providing the selected implant model to a verification module for verification as a candidate implant model for verification, The validation determines whether the physical implant designated by the implant model candidate will pass surgical implantation in the patient. The method according to claim 2, wherein the verification is at least partially based on the manipulated properties of the at least one anatomical surface.

9. The method is such that the computer (i) An operation on a selected implant model, wherein the operation modifies the selected implant model, generates implant model candidates for verification, and specifies a physical implant having one or more different physical characteristics compared to the physical implant specified by the selected implant model, and (ii) Based on identifying implant model candidates after receiving at least one selected from the group including manipulation of the properties of the at least one anatomical surface of the other anatomical structure, Further including providing implant model candidates to a validation module for verification, the validation determines whether the physical implant specified by the implant model candidate will pass surgical implantation in a patient. The above method involves the computer, The method according to claim 2, further comprising displaying the implant model candidate in a training dataset as part of a training example of associating the implant model candidate with the at least one anatomical surface of the other anatomical structure, based on verification that determines that the physical implant specified by the implant model candidate will pass surgical implantation in a patient.

10. The computer further provides the selected implant model as an implant model candidate to a verification module for verification, The method according to claim 1, wherein verification determines whether a physical implant designated by an implant model candidate is suitable for surgical implantation in a patient.

11. The implant model candidate is an initial implant model candidate, The aforementioned method, Based on the verification that the aforementioned computer determined the physical implant specified by the initial implant model candidate does not pass, An operation on an initial implant model candidate, wherein the operation modifies the initial implant model candidate, generates a different implant model candidate for verification, and the different implant model candidate specifies a physical implant having one or more different physical characteristics compared to the physical implant specified by the initial implant model candidate, and Operations on the properties of the at least one anatomical surface of the other anatomical structure, To receive at least one selected from a group having, The computer determines the next candidate implant model to provide to the verification module for verification, Based on the operation received on the initial implant model candidate, it is determined that the next implant model candidate is a different implant model generated from the operation on the initial implant model candidate, or Based on the fact that the properties of at least one anatomical surface of the other anatomical structures have been manipulated and the initial implant model candidate has not been manipulated, the next implant model candidate is determined to be the initial implant model candidate. The method according to claim 10, wherein the computer provides the next implant model candidate to the verification module for verification to determine whether the physical implant specified by the next implant model candidate is suitable for surgical implantation in a patient, the verification for determining whether the physical implant specified by the next implant model candidate is suitable for surgical implantation in a patient is provided, which is at least in part based on receiving at least one selected from a group having operations on the initial implant model candidate and operations on the properties of the at least one anatomical surface.

12. The aforementioned machine learning model has a trained generator of a Generative Adversarial Network (GAN), The generator is trained using samples from a library of implant models to generate implant models. The method according to any one of claims 1, 2, 3, 4, 5, 8, 9, 10, or 11, wherein the selected implant model is generated by the generator and has an implant model selected by the generator to be output as the selected implant model.

13. The obtained image data includes three-dimensional digital model data representing the anatomical regions of the patient. Determining the properties of the at least one anatomical surface of the other anatomical structure is The computer processes the image data to present the at least one anatomical surface as at least one digital three-dimensional surface, The computer includes converting at least one digital three-dimensional surface into at least one two-dimensional projection, The method according to any one of claims 1, 2, 3, 4, 5, 8, 9, 10, or 11, wherein the determined properties of the at least one anatomical surface are determined from the at least one two-dimensional projection.

14. The method according to claim 13, further comprising the computer preprocessing the image data to generate a three-dimensional digital model therefrom of the anatomical region of the patient with the target anatomical structure omitted.

15. The method according to any one of claims 1, 2, 3, 4, 5, 8, 9, 10, or 11, wherein one or more anatomical surfaces have one or more articular surfaces with which a physical implant will engage based on being surgically implanted at least partially within a patient.

16. The aforementioned anatomical region includes the patient's ankle. The aforementioned anatomical structure has a talus, The method according to any one of claims 1, 2, 3, 4, 5, 8, 9, 10, or 11, wherein the at least one anatomical surface has at least one articular surface of at least one bone adjacent to the talus.

17. Determining the properties of the at least one anatomical surface is, Manual indication by the user of at least one anatomical surface provided based on user input to a graphical user interface that displays a model including at least the other anatomical structures of the patient, The method according to any one of claims 1, 2, 3, 4, 5, 8, 9, 10, or 11, based on at least one selected from the group having an automated analysis of the image data for confirming the at least one anatomical surface.

18. Memory and A processor that communicates with the memory, Configured to perform the method, The aforementioned method, Obtaining a machine learning model trained to select an anatomical implant model for a physical implant for the partial or complete replacement of a target anatomical structure, based on the anatomical surface characteristics of an anatomical structure adjacent to the target anatomical structure to be partially or completely replaced, Acquiring image data of an anatomical region of a patient, wherein the anatomical region includes the patient's target anatomical structure and other anatomical structures of the patient, and the other anatomical structures are adjacent to the patient's target anatomical structure. The method involves determining the characteristics of at least one anatomical surface of the other anatomical structure from the image data, wherein the at least one anatomical surface is adjacent to the target anatomical structure of the patient, and determining its characteristics. The process involves utilizing the determined properties of the at least one anatomical surface, applying the machine learning model, and obtaining an implant model selected based on the application, wherein the selected implant model is selected by the machine learning model as a designation of a physical implant for a candidate for surgical transplantation within the patient, at least partially, as a partial or total replacement of the target anatomical structure in the patient. The method further includes training the machine learning model to select an anatomical implant model. Training involves using samples from a library of implant models and training the machine learning model to select anatomical implant models from the said library of implant models. The selected implant model is selected by the machine learning model from the library of implant models. Computer system.

19. The computer system according to claim 18, further comprising providing a selected implant model to a model candidate designation module in order to designate a model candidate for verification.

20. The aforementioned method, Presenting the implant model selected by the user on a graphical user interface, The operation involves receiving an operation on a selected implant model, the operation of modifying the selected implant model, generating implant model candidates for verification, and the implant model candidates specifying one or more physical implants that differ in one or more physical properties compared to the physical implants specified by the selected implant model. The computer system according to claim 19, further comprising providing an implant model candidate to a verification module for verification, the verification of determining whether the physical implant specified by the implant model candidate is suitable for surgical implantation in a patient.

21. The procedure for the selected implant model is: At least one operation specified by the user, and The computer system according to claim 20, having at least one selected from a group including at least one operation automatically determined by artificial intelligence.

22. Based on verification that determines that the physical implant specified by the implant model candidate does not pass, the method further includes (i) receiving an operation and (ii) providing the implant model candidate to the verification module for verification, one or more times. The computer system according to claim 20, wherein in each iteration, implant model candidates that did not pass are provided as selected implant models for the next iteration.

23. The method further includes receiving an operation on the properties of the at least one anatomical surface of the other anatomical structure, The computer system according to claim 20, 21, or 22, wherein the verification for determining whether a physical implant designated by an implant model candidate is suitable for surgical implantation in a patient is at least partially based on the manipulated properties of the at least one anatomical surface.

24. The manipulation of the properties of the at least one anatomical surface is, At least one operation specified by the user, and The computer system according to claim 23, having at least one selected from a group including at least one operation automatically determined by artificial intelligence.

25. The aforementioned method, To receive an operation on the properties of at least one anatomical surface of the other anatomical structure, This further includes providing the selected implant model to a validation module as a candidate implant model for validation, The validation determines whether the physical implant designated by the implant model candidate will pass surgical implantation in the patient. The computer system according to claim 19, wherein the verification is at least partially based on the manipulated properties of the at least one anatomical surface.

26. (i) An operation on a selected implant model, wherein the operation modifies the selected implant model, generates implant model candidates for verification, and specifies a physical implant having one or more different physical characteristics compared to the physical implant specified by the selected implant model, and (ii) Based on identifying implant model candidates after receiving at least one selected from the group including manipulation of the properties of the at least one anatomical surface of the other anatomical structure, The method further includes providing a candidate implant model to a validation module for verification, The validation determines whether the physical implant designated by the implant model candidate will pass surgical implantation in the patient. The computer system according to claim 19, wherein, based on verification that a physical implant designated by an implant model candidate is deemed to pass surgical implantation in a patient, the method further comprises displaying the implant model candidate in a training dataset as part of an example of training that associates the implant model candidate with the at least one anatomical surface of the other anatomical structure.

27. The method further includes providing the selected implant model as an implant model candidate to a validation module for verification, The computer system according to claim 18, wherein verification determines whether a physical implant designated by an implant model candidate is suitable for surgical implantation in a patient.

28. The implant model candidate is an initial implant model candidate, The aforementioned method, Based on verification that determined the physical implant specified by the initial implant model candidate was not acceptable, An operation on an initial implant model candidate, wherein the operation modifies the initial implant model candidate, generates a different implant model candidate for verification, and the different implant model candidate specifies a physical implant having one or more different physical characteristics compared to the physical implant specified by the initial implant model candidate, and Operations on the properties of the at least one anatomical surface of the other anatomical structure, To receive at least one selected from a group having, To determine the next candidate implant model to be provided to the verification module for verification, Based on the operation received on the initial implant model candidate, it is determined that the next implant model candidate is a different implant model generated from the operation on the initial implant model candidate, or Based on the fact that the properties of at least one anatomical surface of the other anatomical structures have been manipulated and the initial implant model candidate has not been manipulated, the next implant model candidate is determined to be the initial implant model candidate. The computer system according to claim 27, comprising providing a next implant model candidate to the verification module for verification to determine whether the physical implant specified by the next implant model candidate is suitable for surgical implantation in a patient, wherein the verification to determine whether the physical implant specified by the next implant model candidate is suitable for surgical implantation in a patient is at least in part based on receiving at least one selected from a group having operations on the initial implant model candidate and operations on the properties of at least one anatomical surface.

29. The aforementioned machine learning model has a trained generator of a Generative Adversarial Network (GAN), The generator is trained using samples from a library of implant models to generate implant models. The computer system according to any one of claims 18, 19, 20, 21, 22, 25, 26, 27, or 28, wherein the selected implant model is generated by the generator and has an implant model selected by the generator to be output as the selected implant model.

30. The obtained image data includes three-dimensional digital model data representing the anatomical regions of the patient. Determining the properties of the at least one anatomical surface of the other anatomical structure is Processing the image data to present the at least one anatomical surface as at least one digital three-dimensional surface, This includes converting at least one digital three-dimensional surface into at least one two-dimensional projection, The computer system according to any one of claims 18, 19, 20, 21, 22, 25, 26, 27, or 28, wherein the determined properties of the at least one anatomical surface are determined from the at least one two-dimensional projection.

31. The computer system according to claim 30, further comprising preprocessing the image data to generate a three-dimensional digital model of the anatomical region of the patient from which the target anatomical structure has been omitted.

32. The computer system according to any one of claims 18, 19, 20, 21, 22, 25, 26, 27, or 28, wherein one or more anatomical surfaces have one or more articular surfaces with which a physical implant will engage based on being surgically implanted at least partially within a patient.

33. The aforementioned anatomical region includes the patient's ankle. The aforementioned anatomical structure has a talus, The computer system according to any one of claims 18, 19, 20, 21, 22, 25, 26, 27, or 28, wherein the at least one anatomical surface has at least one articular surface of at least one bone adjacent to the talus.

34. Determining the properties of the at least one anatomical surface is, Manual indication by the user of at least one anatomical surface provided based on user input to a graphical user interface that displays a model including at least the other anatomical structures of the patient, A computer system according to any one of claims 18, 19, 20, 21, 22, 25, 26, 27, or 28, based on at least one selected from the group having automatic analysis of the image data for identifying the at least one anatomical surface.

35. A computer-readable storage medium is provided which is readable by a processing circuit and stores instructions for execution by the processing circuit to perform a method, The aforementioned method, Obtaining a machine learning model trained to select an anatomical implant model for a physical implant for the partial or complete replacement of a target anatomical structure, based on the anatomical surface characteristics of an anatomical structure adjacent to the target anatomical structure to be partially or completely replaced, Acquiring image data of an anatomical region of a patient, wherein the anatomical region includes the patient's target anatomical structure and other anatomical structures of the patient, and the other anatomical structures are adjacent to the patient's target anatomical structure. The method involves determining the characteristics of at least one anatomical surface of the other anatomical structure from the image data, wherein the at least one anatomical surface is adjacent to the target anatomical structure of the patient, and determining its characteristics. The process involves utilizing the determined properties of the at least one anatomical surface, applying the machine learning model, and obtaining an implant model selected based on the application, wherein the selected implant model is selected by the machine learning model as a designation of a physical implant for a candidate for surgical transplantation within the patient, at least partially, as a partial or total replacement of the target anatomical structure in the patient. Includes, The method further includes training the machine learning model to select an anatomical implant model. Training involves using samples from a library of implant models and training the machine learning model to select anatomical implant models from the said library of implant models. The selected implant model is a computer program product selected by the machine learning model from the library of implant models.

36. The computer program product according to claim 35, further comprising providing a selected implant model to a model candidate designation module in order to designate a model candidate for verification.

37. The aforementioned method, Presenting the implant model selected by the user on a graphical user interface, The operation involves receiving an operation on a selected implant model, the operation of modifying the selected implant model, generating implant model candidates for verification, and the implant model candidates specifying one or more physical implants that differ in one or more physical properties compared to the physical implants specified by the selected implant model. The computer program product according to claim 36, further comprising providing an implant model candidate to a verification module for verification, the verification determining whether the physical implant specified by the implant model candidate is suitable for surgical implantation in a patient.

38. The procedure for the selected implant model is: At least one operation specified by the user, and The computer program product according to claim 37, having at least one selected from a group including at least one operation automatically determined by artificial intelligence.

39. Based on verification that determines that the physical implant specified by the implant model candidate does not pass, the method further includes (i) receiving an operation and (ii) providing the implant model candidate to the verification module for verification, one or more times. The computer program product according to claim 37, wherein in each iteration, implant model candidates that did not pass are provided as selected implant models for the next iteration.

40. The method further includes receiving an operation on the properties of the at least one anatomical surface of the other anatomical structure, The computer program product according to claim 37, 38, or 39, wherein the verification for determining whether a physical implant designated by an implant model candidate is suitable for surgical implantation in a patient is at least partially based on the manipulated properties of the at least one anatomical surface.

41. The manipulation of the properties of the at least one anatomical surface is, At least one operation specified by the user, and The computer program product according to claim 40, having at least one selected from a group including at least one operation automatically determined by artificial intelligence.

42. The aforementioned method, To receive an operation on the properties of at least one anatomical surface of the other anatomical structure, This further includes providing the selected implant model to a validation module as a candidate implant model for validation, The validation determines whether the physical implant designated by the implant model candidate will pass surgical implantation in the patient. The computer program product according to claim 36, wherein the verification is at least partially based on the manipulated properties of the at least one anatomical surface.

43. (i) An operation on a selected implant model, wherein the operation modifies the selected implant model, generates implant model candidates for verification, and specifies a physical implant having one or more different physical characteristics compared to the physical implant specified by the selected implant model, and (ii) Based on identifying implant model candidates after receiving at least one selected from the group including manipulation of the properties of the at least one anatomical surface of the other anatomical structure, The method further includes providing a candidate implant model to a validation module for verification, The validation determines whether the physical implant designated by the implant model candidate will pass surgical implantation in the patient. The computer program product according to claim 36, wherein, based on verification that a physical implant designated by an implant model candidate is deemed to pass surgical implantation in a patient, the method further comprises displaying the implant model candidate in a training dataset as part of an example of training that associates the implant model candidate with the at least one anatomical surface of the other anatomical structure.

44. The method further includes providing the selected implant model as an implant model candidate to a validation module for verification, The computer program product according to claim 35, wherein verification determines whether a physical implant designated by an implant model candidate is suitable for surgical implantation in a patient.

45. The implant model candidate is an initial implant model candidate, The aforementioned method, Based on verification that determined the physical implant specified by the initial implant model candidate was not acceptable, An operation on an initial implant model candidate, wherein the operation modifies the initial implant model candidate, generates a different implant model candidate for verification, and the different implant model candidate specifies a physical implant having one or more different physical characteristics compared to the physical implant specified by the initial implant model candidate, and Operations on the properties of the at least one anatomical surface of the other anatomical structure, To receive at least one selected from a group having, To determine the next candidate implant model to be provided to the verification module for verification, Based on the operation received on the initial implant model candidate, it is determined that the next implant model candidate is a different implant model generated from the operation on the initial implant model candidate, or Based on the fact that the properties of at least one anatomical surface of the other anatomical structures have been manipulated and the initial implant model candidate has not been manipulated, the next implant model candidate is determined to be the initial implant model candidate. The computer program product according to claim 44, which includes providing a next implant model candidate to the verification module for verification to determine whether the physical implant specified by the next implant model candidate is suitable for surgical implantation in a patient, wherein the verification to determine whether the physical implant specified by the next implant model candidate is suitable for surgical implantation in a patient is at least in part based on receiving at least one selected from a group having operations on the initial implant model candidate and operations on the properties of at least one anatomical surface.

46. The aforementioned machine learning model has a trained generator of a Generative Adversarial Network (GAN), The generator is trained using samples from a library of implant models to generate implant models. The computer program product according to any one of claims 35, 36, 37, 38, 39, 42, 43, 44, or 45, wherein the selected implant model is generated by the generator and has an implant model selected by the generator to be output as the selected implant model.

47. The obtained image data includes three-dimensional digital model data representing the anatomical regions of the patient. Determining the properties of the at least one anatomical surface of the other anatomical structure is Processing the image data to present the at least one anatomical surface as at least one digital three-dimensional surface, This includes converting at least one digital three-dimensional surface into at least one two-dimensional projection, The computer program product according to any one of claims 35, 36, 37, 38, 39, 42, 43, 44, or 45, wherein the determined properties of the at least one anatomical surface are determined from the at least one two-dimensional projection.

48. The computer program product according to claim 47, further comprising preprocessing the image data to generate a three-dimensional digital model of the anatomical region of the patient from which the target anatomical structure has been omitted.

49. The computer program product according to any one of claims 35, 36, 37, 38, 39, 42, 43, 44, or 45, wherein one or more anatomical surfaces have one or more articular surfaces into which a physical implant will engage based on being surgically implanted at least partially within a patient.

50. The aforementioned anatomical region includes the patient's ankle. The aforementioned anatomical structure has a talus, The computer program product according to any one of claims 35, 36, 37, 38, 39, 42, 43, 44, or 45, wherein the at least one anatomical surface has at least one articular surface of at least one bone adjacent to the talus.

51. Determining the properties of the at least one anatomical surface is, Manual indication by the user of at least one anatomical surface provided based on user input to a graphical user interface that displays a model including at least the other anatomical structures of the patient, A computer program product according to any one of claims 35, 36, 37, 38, 39, 42, 43, 44, or 45, based on at least one selected from the group having, automatic analysis of the image data for identifying the at least one anatomical surface.

Images