Computer-implemented method, data processing system and computer readable medium for creating a functional form

Abstract

The invention relates to a computer-implemented method for creating a functional form (50), in particular as a basis for individually adapting an orthosis or prosthesis to a first patient, comprising the steps of receiving anatomic structure data (ASD) of the first patient and applying a conversion algorithm (30) to generate functional form data (FFD) for the first patient, wherein the resulting functional form data (FFD) forms a basis for manufacturing a personalized orthosis or prosthesis. Furthermore, the invention relates to a data processing system and to a computer-readable medium.

Description

TECHNICAL FIELD

[0001] The present invention relates to a computer-implemented method for creating a functional form, in particular as a basis for individually adapting an orthosis or prosthesis to a first patient. Furthermore, the present invention relates to a data processing system and to a computer-readable medium. BACKGROUND

[0002] Generally, orthoses or prostheses are manufactured and adjusted individually for a patient. For this purpose, a functional form can be created which replaces the patient's limb for adapting the orthosis or prosthesis. In particular, a certified prosthetist / orthotist and / or an orthotist adjusts the orthosis or prosthesis manually based on the patient's own limb, e.g. a lower leg, or based on the functional form thereof. In this regard, a visualized image of a digital representation of the patient's scan data / anatomical data or a three-dimensional structure of the respective functional form can further assist the certified prosthetist / orthotist and / or the orthotist.

[0003] However, such manual preparation and adjustment of a patient-specific orthosis or prosthesis not only requires time (thus resulting in high costs) but also demands a certified prosthetist / orthotist and / or an orthotist to have a rich educational and background knowledge, not only with regard to professional knowledge but also with regard to handling by appropriate software. Therefore, currently, a certified prosthetist / orthotist and / or an orthotist has to undergo additional training courses in order to handle the regular workflow, in particular with regard to different types of software programs for individually creating adapted orthoses or prostheses in a digital manner. SUMMARY

[0004] It is an object of the present invention to provide an improved method for laying the foundation for individually adapting an orthosis or prosthesis which is adapted and manufactured individually for a respective patient, wherein the process is both time- and cost-saving, easy to handle and reduces the background knowledge and educational requirements necessary to provide such individual orthoses or prostheses. Furthermore, it is an object of the present invention to provide a data processing system and to provide a computer-readable medium.

[0005] These problems are solved by providing a computer-implemented method according to claim 1, a data processing system according to independent claim 13 and a computer-readable medium according to claim 15. Further preferred embodiments of the present invention are described by the dependent claims, respectively.

[0006] According to the present invention, a computer-implemented method for creating a functional form, in particular as a basis for individually adapting an orthosis or prosthesis to a first patient, is provided, the method comprising the following steps:

[0007] a) receiving anatomical data of a first patient;

[0008] b) applying a conversion algorithm to generate functional form data for the first patient;

[0009] wherein the resulting functional form data forms the basis for manufacturing a personalized orthosis or prosthesis.

[0010] The present application is based on the idea to provide a functional form for a patient's anatomy, in particular a limb such as a lower leg, as a basis for adapting an orthosis or prosthesis prior to manufacturing it. Thus, by utilizing a conversion algorithm which is capable of converting scan data of a patient to finally obtain a functional form, a patient-specific functional form can be created automatically.

[0011] By automating the process of obtaining at least a first rough functional form, time and financial resources can be saved. Furthermore, the additional background knowledge required by a respective certified prosthetist / orthotist and / or orthotist to create such a functional form, in particular in terms of software handling, can be reduced.

[0012] For the present application, a functional form represents an adapted or optimized geometry of a patient's anatomy, such as a lower leg, in order to properly fit an orthosis or prosthesis. Preferably, the functional form or functional form data and the scan data / anatomy data of the first patient are not identical, but differ in order to properly adapt the resulting orthosis or prosthesis.

[0013] Furthermore, according to the present application, the conversion algorithm is capable of converting data which can be further processed to finally obtain functional form data. Thus, the conversion algorithm can provide data in any suitable format, in particular structural data which can be converted into functional form data to provide a functional form.

[0014] In a preferred embodiment, step b) comprises encoding the anatomy data, preferably by a first statistical shape model.

[0015] According to another preferred embodiment, the first statistical shape model encodes the anatomy data in the form of a deviation between the anatomy data of the first patient and anatomy data of a limited number of different patients, in particular in the form of encoded anatomy data of the limited number of different patients as comprised by the first statistical shape model.

[0016] For the present application, data processed by a statistical shape model can refer to statistical deviations and the like, in particular to compress the amount of data stored and / or processed. By applying a statistical shape model, different approaches can be used to reduce or compress data, in particular scan data / anatomy data. For example, a statistical deviation / difference of the data of the first patient compared to data of a limited number of patients can be utilized. In particular, such a statistical deviation can be focused on a specific number of feature landmarks of the first patient's limb.

[0017] In one embodiment, the statistical shape model comprises the average position of the landmarks and has a plurality of parameters which control the main modes of variation found in the data of the limited number of different patients. Each spatial axis of the three-dimensional data provides a mode of variation to describe in which way the landmarks tend to move together as the shape varies. Thus, the mode of variation of the data of a single patient refers to the difference from the average of all data of the limited number of different patients.

[0018] The first statistical shape model provides the average position and variation of the anatomical data of different patients, whereas the second statistical shape model provides the average position and variation of the landmarks of the functional form data of different patients. Thus, for the present invention, a statistical shape model refers to the statistical deviation between the data of patients, in particular the statistical deviation between the different anatomical data and / or the landmarks of the structural data of different patients.

[0019] In particular, the first statistical shape model has / comprises the anatomical data of a plurality of different patients to provide the average and deviation in the form of the encoded anatomical data. The second statistical shape model has / comprises the functional form data of a plurality of different patients to provide the average and deviation in the form of the structural data. Thus, the first statistical shape model and the second statistical shape model comprise pairs of data, in particular pairs of data of a limited number of different patients.

[0020] In this regard, the encoded anatomical data of the first patient can be provided by the first statistical shape model, i.e. as an object which preferably comprises the average position of the landmarks according to the anatomical data of the limited number of different patients and the mode of variation of the first patient. Furthermore, any number of feature landmarks can be used, which are manually or automatically defined on the anatomical data of the limited number of different patients and the anatomical data of the first patient, e.g. up to 30, 25, 20, 15, 10, 5 or 3 landmarks.

[0021] By transforming the anatomical data using the statistical shape model to obtain the functional form data, compressed data is provided. Thus, by reducing the amount of data, in particular the amount of data of relevant and / or important information, the processing of these encoded / compressed anatomical data by the conversion algorithm is optimized.

[0022] Furthermore, since there is a correlation between the increasing amount of feature parameters and the increasing amount of necessary data of other different patients as comprised by the statistical shape model, it is highly advantageous to use as little data as possible, e.g. as few landmarks as possible. Thus, e.g. a large number of landmarks would require more data of different patients to train / adjust the conversion algorithm, thus leading to a longer processing time.

[0023] In one embodiment of the application, step b) further comprises decoding the structure data provided by the conversion algorithm, preferably by the second statistical shape model, to obtain the functional form data.

[0024] In another embodiment, the second statistical shape model decodes the structure data of the first patient, which structure data represents a deviation from the functional form data, in particular in the form of the structure data, of a limited number of different patients as comprised by the second statistical shape model, thereby providing the functional form data for the first patient.

[0025] The conversion algorithm can provide the structure data in the form of a second object having an average position as well as a variation pattern. Thus, the data as comprised by the second statistical shape model refers to a functional form, in particular to a variation pattern and an average position of landmarks defined according to the functional form.

[0026] In particular, the process of decoding the structure data can be considered as some kind of inverse operation of the encoding process by the first statistical shape model to provide the appropriate functional form data / the appropriate functional form.

[0027] The second statistical shape model decodes the structure data provided by the conversion algorithm to obtain the functional form data. This decoding process of the second statistical shape model can also be considered as a decompression / regeneration of data information compared to the data compression process by the first statistical shape model.

[0028] According to another embodiment, step b) further comprises:

[0029] applying a conversion algorithm, in particular using a neural network, to the anatomical structure data of the first patient as comprised by the first statistical shape model, in particular to the encoded anatomical structure data of the first patient.

[0030] Thus, the conversion algorithm also processes the individual data of the patient as encoded and provided by the first statistical shape model. I.e. the compressed / encoded anatomical structure data of the first patient is converted / translated by the conversion algorithm to obtain the structure data provided to the second statistical shape model.

[0031] By using the first statistical shape model and the second statistical shape model for data compression and decompression, the process of the conversion algorithm for transforming / translating the data to finally obtain the functional form of the first patient is in particular optimized in terms of data volume and processing time.

[0032] In another preferred embodiment, the conversion algorithm is modified and / or trained based on different patient's anatomical data and functional data as comprised by the first statistical shape model and the second statistical shape model, in particular in form of encoded anatomical data and functional data of different patients, such that the conversion algorithm, in particular using a neural network, is capable of generating the functional data for the first patient provided to the second statistical shape model.

[0033] In particular, the different patient's anatomical data and functional data form a data pair comprised by the first statistical shape model and the second statistical shape model. On this basis, the conversion algorithm can be trained in order to provide individual functional data for the first patient based on the scan image of the first patient, in particular based on the anatomical data of the first patient.

[0034] According to another embodiment, the conversion algorithm uses a neural network to generate the functional data, in particular the functional data decoded by the second statistical shape model, to provide the functional data.

[0035] Preferably, the conversion algorithm uses a neural network. Thus, the conversion algorithm can make use of the information provided by the various different patient's data pairs. Furthermore, additional information extracted from other data combinations and statistical combinations can be utilized by using a neural network. Thus, the results of the conversion algorithm can advantageously be improved by using a neural network.

[0036] Furthermore, the conversion algorithm, preferably using a neural network, can be provided as a self-learning algorithm or self-learning system. Thus, the efficiency of the modification / learning / training process of the conversion algorithm can be additionally increased.

[0037] In one embodiment, step a) comprises scanning a limb of the first patient, in particular a lower leg of the first patient, to provide the anatomical data of the first patient. Preferably, a three-dimensional scan is performed to acquire the anatomical data of the first patient, in particular the anatomical data of the relevant limb of the first patient.

[0038] According to one embodiment, the method further comprises the following steps:

[0039] - a visualization image illustrating at least a first three-dimensional structure of the anatomical data of the first patient, and / or

[0040] - a visualization image illustrating at least a second three-dimensional structure of the functional data of the first patient.

[0041] In particular, the anatomical data of the first patient as well as the generated functional data can be visualized as a three-dimensional structure and thus illustrated, for example, by a display or the like.

[0042] Thus, a visual assessment of the patient's limb based on the anatomical data and the resulting functional form data provided by the process comprising preferably the first statistical shape model, the conversion algorithm and the second statistical shape model can be performed, e.g. by the first patient and / or the certified prosthetist / orthotist and / or the orthotist.

[0043] In another embodiment of the present application, the visualization image of the first three-dimensional structure and the visualization image of the second three-dimensional structure are movable relative to each other such that the first three-dimensional structure and the second three-dimensional structure do not overlap each other or at least partially overlap each other.

[0044] Thus, the two resulting models can be visually compared and overlaid, e.g. for identifying necessary modifications to the provided functional form data.

[0045] According to one embodiment, after visualizing the visualization image of the first three-dimensional structure and / or the visualization image of the three-dimensional structure of the functional form data, a user input for modifying the visualization image of the first three-dimensional structure of the anatomical data and / or the visualization image of the three-dimensional structure of the functional form data is received.

[0046] In particular, the first three-dimensional structure of the anatomical data and / or the second three-dimensional structure of the functional form data itself can be manually modified. Thus, the certified prosthetist / orthotist and / or the orthotist can additionally manually, individually adjust the automatically provided functional form data.

[0047] Another aspect of the present application relates to a data processing system comprising means for performing the steps of the method according to the present application. In one preferred embodiment, at least one client and at least one server are provided, wherein the client is able to send anatomical data of a first patient to the server and to receive functional form data of the first patient from the server, and wherein the at least one server is able to:

[0048] - receive anatomical data of a first patient from a client,

[0049] - process the anatomical data of the first patient by at least a conversion algorithm, preferably by a first statistical shape model, a conversion algorithm and a second statistical shape model, to generate functional form data of the first patient,

[0050] - provide the functional form data of the first patient to the client.

[0051] The client can be used, for example, at the local site of the respective certified prosthetist / orthotist and / or orthotist, to collect the anatomical data and to view and / or manually modify the generated functional form data. In contrast thereto, the extensive workload for processing the anatomical data to provide the functional form data for the first patient can be outsourced to an external server, which has sufficient processing power and storage space for processing the data.

[0052] In another aspect, the present application relates to a computer readable medium comprising instructions which, when executed by a computer, cause the computer to perform the steps of the method according to the present application. BRIEF DESCRIPTION OF DRAWINGS

[0053] The present application will be described in more detail in the following with reference to the enclosed drawings. However, other examples of the present application that can be envisaged, are not to be excluded thereby.

[0054] The enclosed drawings illustrate schematically:

[0055] Figure 1 Exemplary flow chart of a computer implemented method for creating a functional form. DETAILED DESCRIPTION

[0056] According to Figure 1 The first method step comprises scanning the patient 10, in particular a process of scanning a limb, such as a lower leg, of the first patient, thereby collecting anatomical data ASD of the first patient. This step can be handled by a client, for example located at the site of a certified prosthetist / orthotist and / or orthotist.

[0057] Preferably, the anatomical data ASD is provided to a server comprising a first statistical shape model (SSM1) 20. Further, the first statistical shape model 20 has anatomical data of a limited number of different patients, in particular comprising encoded anatomical data of a limited number of different patients.

[0058] The first statistical shape model 20 encodes the anatomical data ASD of the first patient to provide encoded anatomical data eASD. This encoded anatomical data eASD represents compressed data of the anatomical data ASD of the first patient, preferably comprising a deviation of the anatomical data ASD of the first patient from the anatomical data of the limited number of different patients. Thus, the data volume is reduced to characteristic information, for example a statistical deviation of specific landmarks of the relevant limb.

[0059] Subsequently, the encoded anatomical data eASD is provided to the conversion algorithm 30 and processed by it, preferably using a neural network, in order to convert the encoded anatomical data eASD of the first patient as included by the first statistical shape model 20 into structural data SD. The structural data SD is provided to the second statistical shape model 40.

[0060] The second statistical shape model 40 comprises the structural data SD of the first patient as processed by the conversion algorithm 30, and a limited number of functionally form data, preferably in the form of structural data, of different patients. Thus, the first statistical shape model 20 and the second statistical shape model 40 provide a limited number of pairs of data of different patients in order to appropriately modify and / or train the conversion algorithm 30. On this basis, the conversion algorithm 30, preferably using a neural network, is able to convert the encoded anatomical data eASD of the first statistical shape model 30 into the structural data SD of the first patient provided to the second statistical shape model 40.

[0061] In a next step, the second statistical shape model 40 decodes the structural data SD of the first patient in order to obtain functionally form data FFD of the first patient. Thus, the conversion algorithm 30 is able to convert anatomical data ASD, in particular encoded anatomical data eASD, of the first patient into functionally form data FFD, in particular structural data SD, thereby forming the basis for the individual adaptation of the orthosis or prosthesis.

[0062] Preferably, the functionally form data FFD is sent back by the server to the local client of the certified prosthetist / orthotist and / or the orthotist. Thus, large workloads can be centrally processed on the server, while the scanning process 10 for collecting the data and the individual processing of the automatically obtained functionally form (FF) 50 can be implemented at the local location of the patient and / or the certified prosthetist / orthotist and / or the orthotist.

[0063] The steps according to Figure 1 may be summarized as follows. In a first step, a patient scan 10 is performed in order to obtain anatomical data ASD of a first patient. These anatomical data ASD are preferably sent by the local client to the server.

[0064] In a next step, the first statistical shape model 20 processes / encodes the anatomical data ASD of the first patient on the basis of anatomical data of other patients as included by the first statistical shape model 20 to obtain encoded anatomical data eASD. This encoding process can be understood as a data compression, thereby reducing the anatomical data ASD to statistical deviations etc.

[0065] Subsequently, the conversion algorithm 30 processes the encoded anatomic structure data eASD of the first patient to provide the structure data SD. This transformation is processed by the conversion algorithm 30, preferably using a neural network. In particular, the conversion algorithm 30, preferably using a neural network, can be trained to appropriately transform the encoded anatomic structure data eASD of the first patient into the structure data SD based on data pairs, i.e. anatomic structure data and structure data, of a limited number of patients as provided to the first statistical shape model 20 and the second statistical shape model 40 and compressed into the encoded anatomic structure data and the structure data.

[0066] In a further step, the second statistical shape model 40 decodes the structure data SD of the first patient as received from the conversion algorithm 30 in order to finally obtain the functional form data FFD. This decoding process by the second statistical shape model 40 can be understood as regenerating the data information by transforming the statistical deviations as represented by the structure data SD into the complete data set as represented by the functional form data FFD.

[0067] Preferably, the anatomic structure data ASD is received and processed by the server, wherein the resulting functional form data FFD is transmitted back to the client by the server.

[0068] Thus, after providing the functional form data FFD, preferably, the three-dimensional structure of the anatomic structure data ASD and the three-dimensional structure of the functional form data FFD / functional form 50 can be visualized and illustrated for the first patient and / or the certificated prosthetist / orthotist and / or the orthotist by the client.

[0069] In short, the present invention provides an option to automatically generate a personalized functional form 50 based on anatomic structure data ASD of a patient as provided by a routine scan of a limb. This individual functional form 50 can be used to appropriately fit an orthosis or a prosthesis to the first patient.

[0070] Furthermore, by intermediate encoding and decoding of the data by the statistical shape model 20 and the statistical shape model 40, the amount of data to be processed can be reduced to a significant feature data set, which is then regenerated without loss of relevant information, resulting in a suitable personalized functional form 50.

[0071] Moreover, by processing the data by the conversion algorithm 30, preferably using a neural network, the process efficiency can be advantageously increased as well as the accuracy of the resulting functional form data FFD.

[0072] List of signs

[0073] 10 scan

[0074] 20 first statistical shape model (SSM1)

[0075] 30 conversion algorithm

[0076] 40 second statistical shape model (SSM2)

[0077] 50 functional form (FF)

[0078] ASD anatomical structure data

[0079] eASD encoded anatomical structure data

[0080] FFD functional form data

[0081] SD structural data

Claims

1. Computer-implemented method for creating a functional form (50) representing an adapted geometry of an anatomical structure of a patient in order to properly fit an orthosis or a prosthesis, comprising the steps of: a) receiving anatomical structure data ASD of a first patient from a patient scan (10); b) applying a conversion algorithm (30) to generate functional form data FFD for said first patient; wherein applying said conversion algorithm (30) comprises: bl) encoding said anatomical structure data ASD by means of a first statistical shape model (20) to provide encoded structure data eASD, b2) converting said encoded anatomical structure data eASD into structure data SD using a neural network, b3) decoding said structure data SD by means of a second statistical shape model (40) to obtain functional form data FFD, wherein said obtained functional form data FFD forms the basis for manufacturing a personalized orthosis or prosthesis.

2. Method according to claim 1, characterized in that said first statistical shape model (20) encodes said anatomical structure data ASD of said first patient in the form of deviations between said anatomical structure data ASD of said first patient and anatomical structure data of a finite number of different patients as comprised by said first statistical shape model (20).

3. Method according to claim 1 or 2, characterized in that said second statistical shape model (40) decodes said structure data SD of said first patient, said structure data representing deviations from functional form data of a finite number of different patients as comprised by said second statistical shape model (40), thereby providing functional form data FFD for said first patient.

4. Method according to claim 1 or 2, characterized in that said conversion algorithm (30) is modified and / or trained based on said anatomical structure data ASD and functional form data of different patients as comprised by said first statistical shape model (20) and said second statistical shape model (40) such that said conversion algorithm (30) is capable of generating structure data SD of said first patient provided to said second statistical shape model (40).

5. Method according to claim 1 or 2, characterized in that step a) comprises scanning a limb of said first patient to provide said anatomical structure data ASD of said first patient.

6. Method according to claim 5, characterized in that said limb is a lower leg.

7. Method according to claim 1 or 2, characterized in that said method further comprises the steps of: - at least a visualization image illustrating a first three-dimensional structure of said anatomical structure data ASD of said first patient, and / or - at least a visualization image illustrating a second three-dimensional structure of said functional form data FFD of said first patient.

8. Method according to claim 7, characterized in that said visualization image of said first three-dimensional structure and said visualization image of said second three-dimensional structure are movable relative to each other such that said first three-dimensional structure and said second three-dimensional structure do not overlap each other or at least partially overlap each other.

9. Method according to claim 7, characterized in that ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ After the visualization image of the first three-dimensional structure of the anatomical structure data ASD and / or the visualization image of the second three-dimensional structure of the functional form data FFD is illustrated, a user input modifying the visualization image of the first three-dimensional structure of the anatomical structure data ASD and / or the visualization image of the three-dimensional structure of the functional form data FFD is received.

10. A data processing system comprising means for carrying out the steps of the method according to any one of claims 1 to 9.

11. The data processing system according to claim 10, characterized in that at least one client and at least one server are provided, wherein the client is able to send the anatomical structure data ASD of the first patient to the server and to receive the functional form data FFD of the first patient from the server, and wherein the at least one server is able to: - receive the anatomical structure data ASD of the first patient from the client, - process the anatomical structure data ASD of the first patient by at least the conversion algorithm (30) to generate the functional form data FFD of the first patient, - provide the functional form data FFD of the first patient to the client.

12. The data processing system according to claim 11, characterized in that - the anatomical structure data ASD of the first patient is processed by the first statistical shape model (20), the conversion algorithm (30) and the second statistical shape model (40) to generate the functional form data FFD of the first patient.

13. A computer readable medium comprising instructions which, when executed by a computer, cause the computer to carry out the steps of the method according to any one of claims 1 to 9.

Images