BONE IMPLANT

Abstract

Disclosed herein is an implant for use in a body, at least one portion of the surface of the implant being mutually engageable with at least one portion of at least one body part.

Also disclosed is a method of surgery comprising the steps of: forming an implant comprising at least one portion of the surface of the implant being mutually engageable with at least one portion of at least one body part, applying a layer of adhesive to the at least one portion of the surface, and engaging the at least one portion of the surface with the at least one portion of at least one body part.

Claims

1. An implant for use in a body, wherein at least one portion of the surface of the implant is mutually engageable with at least one portion of at least one body part.

2. The implant according to claim 1, wherein the at least one portion of the surface is coated with a layer of adhesive.

3. The implant according to claim 2, wherein the adhesive is curable by external input of a first energy, and the implant comprises at least one conduit whereby said first energy may be introduced into the adhesive, the at least one conduit extending through the body of the implant and opening at the at least one portion of the surface.

4. The implant according to claim 3, wherein the energy is selected from the group consisting of electro-magnetic radiation energy, ultrasound energy, and thermal energy.

5. The implant according to claim 4, wherein the electromagnetic radiation energy is UV light derived from a UV light source.

6. The implant according to claim 5, wherein the adhesive cures within 90 seconds under external input of energy.

7. The implant according to claim 1 wherein the implant comprises a flange extending around the periphery of the implant and protruding in a direction approximately at right angles away from the implant.

8. The implant according to claim 1, wherein the implant comprises one or more protrusions positioned away from the periphery of the implant and protruding in a direction approximately at right angles away from the implant.

9. The implant according to claim 1, wherein the body part is bone.

10. The implant according to claim 1, wherein the shape of the at least one portion of the surface is formed in accordance with a set of data derived from the body part in three dimensions.

11. The implant according to claim 10, wherein the set of data is obtained by computer tomography, ultrasound imaging, cone beam computed tomography (CBCT), and/or magnetic resonance imaging.

12. The implant according to claim 1, wherein the implant is manufactured by milling and/or grinding.

13. The implant according to claim 1, wherein the implant is made from titanium, titanium alloy, PEEK, and/or tissue/bone substitute.

14. The implant according to claim 1, wherein the material of the implant is opaque.

15. The implant according to claim 2, wherein the adhesive is deactivated by exposure to a second external energy, wherein the first and second energies are different.

16. A method of surgery comprising the steps of: forming an implant comprising at least one portion of the surface of the implant being mutually engageable with at least one portion of at least one body part; applying a layer of adhesive to the at least one portion of the surface; and engaging the at least one portion of the surface with the at least one portion of at least one body part.

17. The method according to claim 16, wherein prior to the application of adhesive to the implant, a bonding agent is applied to both the surface of the body part and the surface of the implant.

18. The method according to claim 16, wherein prior to en-gaging the implant with the body part, the at least one portion of at least one body part is coated with a layer of adhesive.

19. The method according to claim 16, wherein the adhesive is curable by external input of energy, and the implant comprises at least one conduit whereby said first energy may be introduced into the adhesive, the at least one conduit extending through the body of the implant and opening at the at least one portion of the surface, the method comprising the further step of exposing the adhesive in the region between the at least one portion of at least one body part and the at least one portion of the surface to external energy from an external energy source presented to the at least one conduit from the exterior of the implant.

20. The method according to claim 16, wherein the adhe-sive is deactivated by exposure to a second external energy, wherein the first and second energies are different.

21. The method according to claim 16, wherein the shape of the at least one portion of the surface is formed in accordance with a set of data derived from the body part in three dimensions.

22. The method according to claim 21, wherein the set of data is obtained by computer tomography, ultrasound imaging, cone beam computed tomography (CBCT) and/or magnetic resonance imaging.

23. The method according to claim 16, wherein the implant is manufactured by milling and/or grinding.

24. The method according to claim 16, wherein the implant is made from titanium, titanium alloy, PEEK and/or tissue/bone substitute.

25. A software model configured to design an implant for use in the body, the software model comprising; scanning a patient body part; generating a 3D model in silico of the patient body part from the scan; generating a 3D model in silico of the implant wherein the implant model has an abutment surface that is a negative of the surface of the body part model; predicting the loading pattern and/or fatigue stresses on the implant model using inputted patient data; and regenerating the model of the implant in silico until the loading pattern and/or fatigue stresses are within a threshold value.

26. The software model according to claim 25, wherein model provides data to validate the correct placement of the implant in-silico prior to surgical placement.

27. The software model according to claim 25, wherein the model validates the structural integrity of the implant given the eventual purpose of the implant.

28. The software model according to claim 25, wherein the model uses finite element analysis to predict the loading pattern on the body part.

29. The software model according to claim 25, further comprising inputting data from a database of manual iterations to further opti-mise the overall design of the implant.

30. The software model according to claim 25, further comprising automatic recognition of patient specific anatomical landmarks wherein anonymised patient specific landmarks augment and iteratively improve the generic model.

31. The software model according to claim 25, further comprising automatic initial design proposition of a patient specific implant using the difference between generic and patient specific anatomical landmarks as input parameters in an automatic parametric design process.

Description

DESCRIPTION OF THE DRAWINGS

[0051] The invention will now be described by way of example and/or illustration only with reference to the accompanying drawings in which:

[0052] FIG. 1 shows an embodiment of the implant according to the invention adhered to a vertebra;

[0053] FIG. 2 shows the results of experimental tests to demonstrate the tensile strength of the bonding achieved with and without the application of UV light to cure the adhesive;

[0054] FIG. 3 shows an embodiment of the implant according to the invention adhered to the orbital floor;

[0055] FIG. 4 shows an embodiment of the implant according to the invention adhered to the Malar region;

[0056] FIG. 5 shows an embodiment of the implant according to the invention in the form of a TMJ replacement for a canine;

[0057] FIG. 6 shows an embodiment of the implant according to the invention in the form of an elbow replacement;

[0058] FIG. 7 shows an embodiment of the implant according to the invention used as an S spine support; and

[0059] FIG. 8 shows an embodiment of the implant according to the invention in three dimensional view showing the flange.

[0060] FIG. 9 shows a flow chart for the application of the software model according to the invention in optimising the design of the implant.

[0061] FIG. 10 shows lateral, posterior and transparent views of an L4-S1 model.

[0062] FIG. 11 is a table showing the material properties of different spinal components.

[0063] FIG. 12 is a graph showing typical FE result for axial rotation.

[0064] FIG. 13 is a table showing the comparison between experimental results, literature results and FE modelling results.

[0065] FIG. 14 shows plots of von Mises stresses for typical loading conditions.

[0066] FIG. 15 is a table showing typical fatigue parameters to predict the fatigue strength of bone.

[0067] FIG. 16 shows plots of von Mises stresses in the high stress regions in the healthy lumbar human spine.

[0068] FIG. 17 shows plots of shear stress in the high stress region in the healthy lumbar spine.

[0069] FIG. 18 is a table showing the number of fatigue cycle for each loading condition of healthy lumbar spine.

[0070] FIG. 19 shows a model of a laminar defect after cut extrude.

[0071] FIG. 20 shows plots of von Mises stresses shown in the high stress region in the lumbar spine having a laminar defect.

[0072] FIG. 21 shows plots of shear stress in the high stress region in the lumbar spine with a laminar defect.

[0073] FIG. 22 is a table which shows the number of fatigue cycle for each loading condition of lumbar spine with a laminar defect.

[0074] FIG. 23 shows a flow chart describing the automatic design and learning program work flow.

DETAILED DESCRIPTION OF THE ILLUSTRATED EMBODIMENT

[0075] FIG. 1 shows an embodiment of the implant according to the invention. The implant is generally designated 1. The implant is shown attached to a human vertebra 2 which has a lamina defect 3 caused by a stress fracture in the bone. The defect in this case is malformation of the two halves of the vertebra so that there is no structural connection present. This can be a source of lower back pain. However having one of these conditions does not mean that back problems are a certainty. It does however present a higher risk compared to people that don't have the defect, towards developing pain. These conditions can cause mechanical pain, which is the kind of pain that come from within the moving parts of the spine. These conditions can also cause compressive pain, which is derived from pressure on the nerves in the lower back.

[0076] The implant 1 is custom made, which means that it is manufactured to specifically fit against the individual vertebra shown with the inside surface of the implant matching the surface of the bone. The implant has an approximately uniform thickness of between 2 and 3 mm inclusive. The greater the fracture area the greater the thickness of the implant. The implant is made from titanium alloy using a milling manufacturing process.

[0077] The implant 1 has two conduits 4 which extend through the whole thickness of the implant from the outward side (side shown) to the inward side (side in contact with the vertebra). Each conduit 4 is 2 mm in diameter, and in the example shown they are positioned either side of the defect. This is because placing a conduit over the defect would reduce the strength of the supporting implant in the region where is required the most.

[0078] Between the implant 1 and the vertebra is a layer of medical grade adhesive which is applied to the implant 1 before the implant is applied to the vertebra. The adhesive is a UV curable adhesive.

[0079] All around the edge of the implant 1 there is flange 30, which may also be described as a protruding lip, which protrudes inwards towards the vertebra. For the sake of clarity, inwards in the context of the present invention means towards the body part in use.

[0080] In use, prior to the application of adhesive the surface of the implant is coated in a layer of bonding agent. The adhesive is then applied to the under surface 32 of the implant 1. See FIG. 8 for detail of the under surface 32 and the flange 30. This is done using a conventional dispenser or by hand mixing the adhesive for 20 seconds and then pasting on with a rigid applicator. Prior to applying the implant in place against the bone the bone surface is prepared firstly by brushing the surface of the bone to remove the thin layers of soft tissue surrounding the bone. This reveals the native bone surface and roughens it microscopically to improve bond strength. Then the bone is coated in a layer of bonding agent. As the agent is liquid, a 3d printed silicone barrier dam is created and applied around the bonding area to prevent the liquid from damaging the surrounding tissues. The dam may be manufactured in substantially the same way as the implant is manufactured.

[0081] The implant 1 is then pressed into place against the vertebra 2. Maximum force can be applied to the implant 1 to ensure maximum adherence without risk of squeezing out all the adhesive because the flange around the edge contacts the vertebra and controls the separation of the vertebra with the under surface of the implant.

[0082] The inventors have also found a surprising effect of the combination of the at least one conduit and this flange. Effectively it means that when the user applies the implant to the body part, and applies pressure, the only place for the adhesive to escape is out of the conduit. Surprisingly, this actually increases the bond strength of the adhesive. It is thought that this is because of the increased surface areas generated by the conduits in combination with a vacuum effect.

[0083] A UV light pipe attached to a UV source is then engaged with each of the conduits 4 and UV light is delivered to each of the conduits simultaneously. The light pipe is a light conducting rod that delivers UV light to the UV activated adhesive. The adhesive is UV transparent and is able to transmit the UV through it. Furthermore, the inside surface of the implant is reflective. The combination of these features means that the UV light diffuses outwards from the conduit into the adhesive gap between the implant and the bone. It takes approximately 90 seconds for the adhesive to completely cure across the whole surface of the underside of the implant. The paths of light diffusion are represented by the arrows shown on the figure. The implant 1 is thereby set in place in contact with the vertebra and provides a supporting function within 90 seconds. This greatly reduces the time the patient spends in theatre. Using conventional methods it can take 24 hours for the adhesive to fully cure.

[0084] In order to manufacture the implant 1 so that the face of the implant is mutually engageable with the bone, firstly the body part is scanned. This can be done before surgery. Most modern methods of scanning will be suitable. In this case ultrasound, CT and/or MRI scanning are used. The scans produce a series of slices through the bone. The slices are combined to create a three dimensional image of the bone in detail. A three dimensional model of the implant is then created having a bone abutment surface that is a negative of the above three dimensional image of the bone surface. The model, which is present as a data file, is processed by a milling machine in order to then manufacture the implant. It will be appreciated that an additive process may also be used such as selected laser sintering for example. In the milling process a 3 or 5 axis miller progressively grinds away material from a block of material (in the same way that a sculptor will cut from a block of stone to create a bust) in accordance with the model to reveal the implant. Whichever process is used, the general process is the same, which involves the collection of medical data, the definition of the layers for the creation of the face of the implant, determining the optimal positions for the conduits and any other parts attached to the implant, determination of the depth of the flange and eventual manufacture.

[0085] It is recognised that due to the inaccuracy of even most modern scanning techniques the implant may not be an exact fit over the body part of interest. It may be, for example that there are peaks and troughs in the body part of interest, which are not replicated in the implant. Without the flange 30 around the rim of the implant, then it is likely that when the implant is applied to the body part of interest the implant would contact the bone at various points. In such places, the layer of adhesive is thinner and therefore weaker. The flange 30 of 80 m, guarantees that the face of the implant always stood off from even the highest unaccounted for peaks in the bone's surface. Thus the amount of adhesive between the implant and bone is controlled and therefore the mechanical forces involved after curing are controlled too. Furthermore, as there are no bone peaks that contact the implant, there is an unhindered pathway for the UV light to travel through during curing. This means that UV light travels further in the adhesive gap and the adhesive can be fully cured within 90 seconds.

[0086] Using the method described above, the inventors completed a number of experimental tests to demonstrate the tensile strength of the bonding achieved. An implant was manufactured according to the method described above and bonded to a section of bone. UV light curing was used to cure the adhesive according to the method described above for 90 seconds. Successive retractive forces were applied and the displacement of the implant from the bone measured. The results are shown in FIG. 2. The results show that the adhesive can resist a force of up to 750N with a displacement of only approximately 0.8 mm, before it breaks and the implant separates from the bone. A comparative test without the use of UV curing exhibited a substantially greater displacement when a force in excess of 500N is applied.

[0087] FIG. 3 shows an embodiment of the implant 1 according to the invention adhered to the orbital floor in the skull 10.

[0088] FIG. 4 shows an embodiment of the implant 1 according to the invention adhered to the malar region of the skull 10.

[0089] FIG. 5 shows an embodiment of the implant 1 according to the invention in the form of a temporomandibular joint (TMJ) replacement for a canine. The left figure shows the view from the rear and the right figure shows the view from the front. The skull 12 and mandible parts 14 are shown joined by the new TMJ joint.

[0090] FIG. 6 shows an embodiment of the implant according to the invention in the form of an elbow replacement. In this example the implant has been fixed into the central part of the existing bones and is in two halves which are joined by a hinge joint.

[0091] FIG. 7 shows an embodiment of the implant 1 according to the invention used to fix a spine support. As is exemplified in this figure, the implants 1 have super structures 20 fixed to them. These super structures are linked by a brace 22. The brace 22 can have tension applied to it to straighten the spine, or can be used to stabilise damaged joints.

[0092] FIG. 8 shows an embodiment of the implant 1 according to the invention in three dimensional view showing the flange 30. The flange 30 extends around the whole periphery of the implant and is 80 m deep. In use the adhesive is spread on the under surface 32 of the implant 1. The implant is applied to the body part with the under surface 32 downward. An adhesive gap is thereby created as the flange 30 is in contact with the body part around the edge of the implant, but the under surface 32 is held away from the body part. Only the under surface 32 is mutually engageable with the body part.

[0093] Example Application of the Software Model According to the Invention in Optimising the Design of the Implant.

[0094] FIG. 9 shows a flow chart for the application of the software model according to the invention in optimising the design of the implant. The process for optimising the design of the implant begins with a patent specific CT scan. This scan is then segmented and a patient specific implant is designed, substantially as described above. However, in order to optimise the design, the design is converted into a computer model which allows the application of simulation programs and computing procedures to produce the optimum result. If the optimum result is not achieved initially then further modelling is carried out. Inputs into the model include but are not limited to weight of the patient, walking gait data and moment of rotation. After successive rounds of optimisation a final design is produced, which is implemented into the construction of the implant.

[0095] Example of the Software Model and Its Use in Response Analysis of the Lumbar Spine During Regular Training of an Athlete.

[0096] This example involved creating a finite element (FE) model of the lumbar spine including cortical bone, spongial bone, annulus, nucleus, and facet joint. The regions L4-S1 human lumbar spine were studied and the FE model simulated the loading pattern of a sport athlete. The patient data inputted included parameters of extra extension, flexion and axial rotation. Stress analysis of the lumbar spine was studied and used to predict further fatigue fracture risk in healthy individuals.

[0097] FIG. 10 shows a three dimensional aspect of a nonlinear finite element model of a L4-S1 human lumbar motion segment 100 which was developed using finite element code ABAQUS 6.13. The lumbar spine model was modelled with solid tetrahedral linear elements (types C3D4). There are total 205422 elements in the model. The S1 is the final vertebrae which acts as a rigid body in motion with little deformation. The sacral slope was oriented to 66 mimicking the degree of sacral slope in athletes. A displacement of constraint Encaster was used to define inferior vertebral of S1 to prevent any movement when loading is applied. Different loading conditions were used to describe the body movement including compressive force and torsional moment. The compressive force is representing the body weight of the upper body following by the torsional moment loading such as flexion, extension or axial rotation.

[0098] Different ranges of compressive forces and torsional moments were used to simulate a realistic load scheme of body movement. In this work, 800N of compressive force followed by 10 Nm of torsional moment was applied to act as a safety factor and take into account unexpected variation. Another reason for these amounts were to simulate a maximum loading value which is a prediction of the further failure risk of the bone. Through the ABAQUS load module, two types of load were created (concentrated force and Moment) and applied to the superior vertebral endplate of L4. As shown in FIG. 10(b), red colour meshes represent cortical bone and white is spongial bone. The intervertebral regions were modelled with annulus (white) and nucleus (Green). The material properties had been assigned individually in the program and then assembled together according to the original coordinates or position of the human body. The annulus and nucleus was modelled as a low modulus nearly incompressible body with Poisson's ratio of 0.499. FIG. 11 shows the material properties of each component.

[0099] In this example, all the elements were assumed to be fully integrated. The surfaces were simulated by imposing a rough friction behaviour to the tie-contact pair to prevent extraction from each other. FIG. 12 shows typical FE results for axial rotational angle of spine lumbar over torsional moment. Four loading types have been simulated here as shown in FIG. 13. The table in FIG. 13 shows the comparison between literature and FE model results, which were very close to each other. The literature loading angle of torsional moment loading was used as a guide to validate the model.

[0100] Qualitative investigation of the stress features was used to predict the fracture according to the bone strength. von Mises stresses were used to study the load transfer and stress concentration. The general form of the von Mises stress is

v2=1 2[(1122)2+(2223)2+(1133)2+6(232+312+122)](1)(Eq. 1)

[0101] Where 11, 22 and 33 is represent x,y and z principle stress, 23, 31 and 12 is represent yz,zx and xy shear stress.

[0102] Looking at the same lumbar portion (L4-S1) 100, as shown in FIG. 14, different colour contours represent the von Mises stress level starting from minimum in blue to maximum in red, as compressive load and torsional moment were applied to the L4 superior endplate. External compression and shear loading were transmitted from L4 superior endplate to L5 and L5 to S1 through the resistance of the intervertebral discs and the facet joint. Due to the fact that S1 was fixed and L4 is transmitting all the loading toward L5, L5 is sustaining all the loading transmitted from L4 and restricted movement by S1 superior articular process. Such a loading condition generates twisting moment of the pedicles region. As shown in FIG. 14(a), an axial rotation (to the left) has been simulated and result shows that a high stress field is located by the right side of the laminar region of L5. Similar results were obtained for axial rotation (to the right) as shown in FIG. 14(b). High stress occurred by the left par region of L5. FIG. 14(c) is the extension loading scheme and the stresses are equal under compression on the par region of L5. For the flexion motion see FIG. 14(d) which shows that the laminar region of L5 is under tension by the laminar region of L5. The results are sensible and show every single motion and different stress distributions on different regions of vertebrae.

[0103] Stress distributions are important to investigate and understand as the clinical consequence of excessive fatigue damage in bone. The stress fracture of the bone can occur when the material is continuously loaded to respective yield point. The higher the stress level, the more the bone behaves like a brittle material. The fatigue strength for compression is about 150 Mpa. The fatigue behaviour of bone has been investigated in the form of Wohler curves (S-N curves) as represented by Equation 2 below.

S=S0+srlog (Nf)(Eq.2)

[0104] S0=149.8 Mpa and Sr=17.96 Mpa.

[0105] where S is the fatigue strength, Nf is number of cycles to failure, S0 is maximum fatigue strength and Sr is slope of the fatigue curve. FIG. 15 shows a table of typical fatigue parameters to predict the fatigue strength of bone and is one of the sample group taken from donors with the age of 53 and provides some experimental results for fatigue strength of the bone. Equation 2 predicts the fatigue strength of the simulation results. The S0 is a maximum strength in a single cycle failure where if the stress over or equal to the S0 then fail immediate within one cycle. FIG. 14 is a plot of von Mises stress with colour contour to show the difference between stress levels of the spine lumbar by the laminar region.

[0106] FIG. 16 shows similar plots of von Mises stresses for the inner vertebral arch 110 of the lumbar spine. The label on the left of each of the figures shows the values of stress and what each colour correspond to. FIG. 14(a), shows a typical loading type (Axial rotation (To the left)) with higher von Mises stress by the right side of the laminar (19.29 Mpa) and inner pedicle of L5. For the axial rotation (To the right), stress is concentrated within left side of L5 between laminar and par region and is about 17.14 Mpa as shown in FIG. 14(b). From the inner view (FIG. 16(b)) stress is occurred by the bottom left side of pedicle region of L5 (close to 30 Mpa). Extension of the spine may cause compression on the L5 and FIG. 14(c) shows the laminar region of L5 in both sides with the maximum stress of 19.29 Mpa. Similar phenomena and stress (19.29 Mpa) is shown for flexion loading (FIG. 14(d)), but this time it is in tension stress due to force and moment bending downward and stop by the superior articular facet of S1 to form pulling effect. For FIG. 16(c & d) the maximum stress is close to 30 Mpa and occurred by the inner pedicle. This suggest that the inner pedicle region is major stress concentration region and not the laminar region.

[0107] FIG. 17(a-d) shows the shear stress of every loading condition at the inner vertebral arch 110 and the highest shear stress is about 11 Mpa for axial rotation (To the left), as shown in FIG. 17(a). FIG. 18, is a table which shows the number of fatigue cycles for each loading condition of healthy lumbar spine based on Eq. 2. Although no single cycle failures were observed in any loading conditions, stresses that correspond to compression stress failure after 1 million cycles of repetitive axial rotation conditions case were observed. For extension and flexion loading conditions, fatigue failure only occur over 10 million cycles.

[0108] Example of the software model and its use in response analysis of the lumbar spine which has been modelled with laminar defects.

[0109] In this example, a 3D model of the lumbar spine with a laminar defect of L5 was developed and investigated for the influence of the defects in a substantially similar way as described above. A laminar defect was created through ABAQUS cut-extruded to dictate the area of fracture. To understand the stress distribution in L5 with one side laminar defect, FE modelling was used to mimic the laminar fracture by Cut Extrude (blue circle) on right side of the L5 laminar region 120 as shown in FIG. 19. The purpose of this study was to find out the further risk of fracture in the laminar defect on right side of L5 vertebrae. The mesh C3D4 tetrahedron element type was used in this model. Similar material properties and interaction properties were used in this model to assess the comparability results.

[0110] Laminar facture may lead to back pain and serious illness which can be explained in biomechanics behaviour. The modelled laminar fracture of L5 120 is again shown in FIG. 20. FIG. 20 shows the axial rotation (To the left) with relatively large stress concentration within the pedicle of L5. Similar results for axial rotation (To the right) with 45 Mpa of stress occurs in the pedicle of L5 in the left (FIG. 20(b)). There was little difference for extension compared with healthy lumbar spine (FIG. 20(c)), but 15 Mpa extra for flexion loading as shown in FIG. 20(d). In addition, a laminar defect can increase the chance of fracture to the vertebrae in different regions. As shown in FIG. 21(a), shear stress occurs by the right side of pedicle. For FIG. 21(b) shear stress increased to 12 Mpa. Both are the axial rotation loading conditions. No or little increment of shear stress for extension and flexion was shown (FIG. 21(c & d).

[0111] FIG. 22 shows a table showing the overall result of number fatigue cycles prediction. For axial rotation (To the right), the compression stress is about 45 Mpa and has reduced the number of fatigue cycles to less than 1 million cycles. This is because of the unconnected right laminar (fracture) which is unable to share the load from one side to another, therefore stress is concentrated on one side and no longer fully transmitting to S1 to equivalent the stress distribution. Undoubtedly, the discs are the major spinal components for the absorption of impact energy, particularly during the loading period. Through deformation of the discs, loads can be transmitted gently in order to prevent spinal injury. This suggests that with the laminar defect on right side of the L5 there is a risk of causing another fracture in the left side. Thus there is a risk of serious illness (spondylolysis or fatigue fracture) if no further treatment or implantation taken.

[0112] In another example of the software according to the invention the software further comprises inputs from an artificial intelligence (AI) program. The AI Program comes with a generic 3D patient model with anatomical landmarks. As described above a patient specific CT scan is performed and from that a three dimensional model is created. The AI program then personalises the generic 3D model with patient specific data. Then the AI program with a pre-programmed database is used to automatically recognise surface landmarks on the model. Then, anonomised patient specific landmarks augment and iteratively improve the generic model. The difference between generic and patient specific anatomical landmarks gives parameters for parametric automatic initial design proposition for the implant. A technician or surgeon will review the initial design and edit it within certain limits/parameters pre-verified by FE analysis on the generic model. The technician or surgeon is then allowed to manually change the design when required. Every time the technician or surgeon manually changes or modifies the design, the AI program will capture and store the manual changes into the database so that it can be called on for the next job or patient. In this way the software learns at the same time it improves the design for the next job.

[0113] If the design is confirmed it will then be sent to FE analysis for verification within the boundary conditions. This includes the ability for remodelling, if the optimum design is not reached. FIG. 23 shows a flow chart describing this exemplified automatic design and learning program work flow.