Bone reconstruction and orthopedic implants

Abstract

To provide systems and methods for bone and tissue reconstruction. [Solution] A method for constructing a patient-specific orthopedic implant, comprising the steps of: (a) comparing a patient-specific abnormal bone model derived from the actual biological structure of the patient's abnormal bone with a reconstructed patient-specific bone model also derived from the biological structure of the patient's bone, wherein the reconstructed patient-specific bone model reflects the normalized biological structure of the patient's bone, the patient-specific abnormal bone model reflects the actual biological structure of the patient's bone including at least one of incomplete bone, deformed bone, and crushed bone, the patient-specific abnormal bone model comprising at least one of a patient-specific abnormal point cloud and a patient-specific abnormal bone surface model, and the reconstructed patient-specific bone model comprising at least one of a reconstructed patient-specific point cloud and a reconstructed patient-specific bone surface model.

Description

[Technical field]

[0001] (CROSS REFERENCE TO RELATED APPLICATIONS) This application claims the benefit of U.S. Provisional Patent Application No. 61 / 891,047, filed October 15, 2013, entitled "CRANIUM AND POSTCRANIAL BONE AND SOFT TISSUE RECONSTRUCTION," the disclosure of which is incorporated herein by reference. Summary of the Invention [Means for solving the problem]

[0002] (Related Technology) FIELD OF THEINVENTION The present disclosure is directed to various aspects of orthopaedics including bone and tissue reconstruction, patient-specific mass-customized orthopaedic implants, gender and ethnicity specific orthopaedic implants, cutting guides, trauma plates, bone graft cutting and placement guides, patient-specific instruments, the use of inertial measurement units in anatomical tracking for kinematics and pathology, and the use of inertial measurement units for navigation during orthopaedic procedures. (Preface to the invention)

[0003] A first aspect of the present invention provides a method for constructing a patient-specific orthopedic implant, the method comprising the steps of: (a) comparing a patient-specific abnormal bone model derived from an actual anatomy of the patient's abnormal bone with a reconstructed patient-specific bone model also derived from the anatomy of the patient's bone, the reconstructed patient-specific bone model reflecting a normalized anatomy of the patient's bone, the patient-specific abnormal bone model reflecting an actual anatomy of the patient's bone including at least one of an incomplete bone, a deformed bone, and a fractured bone, the patient-specific abnormal bone model comprising at least one of a patient-specific abnormal point cloud and a patient-specific abnormal bone surface model, and the reconstructed patient-specific bone model comprising at least one of a reconstructed patient-specific point cloud and a reconstructed patient-specific bone surface model; (b) optimizing one or more parameters using data output from the comparison of the patient-specific abnormal bone model with the reconstructed patient-specific bone model such that the patient-specific orthopedic implant is fitted to the patient's abnormal bone; and (c) generating an electronic design file for the patient-specific orthopedic implant taking into account the one or more parameters.

[0004] In a further detailed embodiment of the first aspect, the method further comprises manufacturing the patient-specific implant using the electronic design file. In yet another further detailed embodiment, the method further comprises comparing the patient-specific abnormal bone model with the reconstructed patient-specific bone model to identify missing or deformed bones from the patient-specific abnormal bone model, and localizing the missing or deformed bones onto the reconstructed patient-specific bone model. In a further detailed embodiment, the method further comprises generating a patient-specific abnormal bone model from data representative of the patient's abnormal bones, and generating a reconstructed patient-specific bone model from data representative of the patient's abnormal bones and data from a statistical atlas, the statistical atlas data comprising at least one of a point cloud and a surface model of a normal bone similar to the patient's abnormal bone. In yet a further detailed embodiment, the data representative of the patient's abnormal bones comprises at least one of a magnetic resonance image, a computed tomography image, an x-ray image, and an ultrasound image. In a further detailed embodiment, the statistical atlas data is derived from at least one of a magnetic resonance image, a computed tomography image, an x-ray image, and an ultrasound image of a normal bone. In a further detailed embodiment, the identified missing or deformed bone comprises a set of boundary points, and said step of localizing the missing or deformed bone onto the reconstructed patient-specific bone model comprises associating the set of boundary points with the reconstructed patient-specific bone model. In another further detailed embodiment, said step of comparing the patient-specific abnormal bone model and the reconstructed patient-specific bone model to identify missing or deformed bones from the patient-specific abnormal bone model comprises outputting at least two lists of data, the at least two lists of data comprising a first list identifying missing or deformed bones and a second list identifying bones in common between the patient-specific abnormal bone model and the reconstructed patient-specific bone model. In yet another further detailed embodiment, the first list comprises vertices belonging to missing or deformed bones from the patient-specific abnormal bone model, and the second list comprises vertices belonging to bones in common between the patient-specific abnormal bone model and the reconstructed patient-specific bone model.In yet another more detailed embodiment, the method further includes determining one or more patient-specific orthopedic implant fixation locations using data from the patient-specific abnormal bone model and data from the reconstructed patient-specific bone model.

[0005] In yet another more detailed embodiment of the first aspect, the step of determining one or more patient-specific orthopaedic implant fixation locations includes excluding any locations where missing or deformed bone is identified. In yet another more detailed embodiment, the step of optimizing one or more parameters for the patient-specific orthopaedic implant includes using an implant parameterization template to define generic parameters that are subsequently optimized using a reconstructed patient-specific bone model. In a further detailed embodiment, the parameters include at least one of an angle parameter, a depth parameter, a curvature parameter, and a fixation device position parameter. In a still further more detailed embodiment, the method further comprises the step of constructing an initial iteration of a surface model of the patient-specific orthopaedic implant. In a further detailed embodiment, the step of constructing an initial iteration of a surface model includes combining contours from a patient-specific abnormal bone model and contours from a reconstructed patient-specific bone model. In a further detailed embodiment, the step of constructing an initial iteration of a surface model includes considering an intended implantation location for the patient-specific orthopaedic implant. In another further detailed embodiment, the method further comprises the step of constructing a subsequent iteration of a surface model of the patient-specific orthopaedic implant. In yet another more detailed embodiment, the step of configuring a subsequent iteration of the surface model of the patient-specific orthopaedic implant includes manual review of the subsequent iteration of the surface model and the reconstructed patient-specific bone model to discern whether a further iteration of the surface model is required. In yet another more detailed embodiment, the electronic design file includes at least one of a computer-aided design file, a computer numerical control file, and a rapid manufacturing instruction file.

[0006] In a further detailed embodiment of the first aspect, the method further comprises generating an electronic design file for the patient-specific implant placement guide using the one or more parameters optimized for the patient-specific orthopaedic implant. In yet another detailed embodiment, the method further comprises manufacturing the patient-specific implant placement guide using the electronic design file for the patient-specific implant placement guide. In a further detailed embodiment, the one or more parameters optimized for the patient-specific orthopaedic implant include at least one of a size parameter, a shape parameter, and a contour parameter. In an even more detailed embodiment, the at least one contour parameter is common between the patient-specific orthopaedic implant and the patient-specific implant placement guide. In a further detailed embodiment, the method further comprises designing the patient-specific implant placement guide to include a surface shape that is a negative of a surface shape of the patient's bone in which the patient-specific implant placement guide is intended to reside. In a more detailed embodiment, the patient-specific abnormal bone model comprises at least one of a patient-specific abnormal femur model and a patient-specific abnormal pelvic bone model derived from the actual anatomy of the patient's abnormal hip joint, the reconstructed patient-specific bone model comprises at least one of a reconstructed patient-specific femur model and a reconstructed patient-specific pelvic bone model derived from the anatomy of the patient's hip joint, the reconstructed patient-specific model reflects normalized anatomy from the patient's hip joint, and the patient-specific abnormal bone model reflects actual anatomy from the patient's hip joint. In another more detailed embodiment, the patient-specific abnormal bone model comprises a patient-specific abnormal femur model, the reconstructed patient-specific bone model comprises a reconstructed patient-specific femur model, the reconstructed patient-specific model reflects normalized anatomy from the patient's proximal femur, the patient-specific abnormal bone model reflects actual anatomy from the patient's proximal femur, and the patient-specific orthopaedic implant comprises a femoral stem implant.

[0007] In further detailed embodiments of the first aspect, the patient-specific abnormal bone model comprises a patient-specific abnormal pelvic bone model, the reconstructed patient-specific bone model comprises a reconstructed patient-specific pelvic bone model, the reconstructed patient-specific model reflects normalized anatomy from the patient's pelvis, the patient-specific abnormal bone model reflects actual anatomy from the patient's pelvis, and the patient-specific orthopedic implant comprises an acetabular cup implant. In yet other further detailed embodiments, the electronic design files for the patient-specific orthopedic implant include at least one of a computer-aided design file, a computer numerical control file, and a rapid manufacturing instruction file.

[0008] A second aspect of the present invention provides a method for generating an electronic reconstructed bone model of an abnormal bone, the method comprising: (a) using at least one of a point cloud and a surface model of the abnormal bone to at least one of identify a bone from a statistical atlas that is similar to the abnormal bone, align the bone from the statistical atlas to the abnormal bone, and morph surface points on the reconstructed model of the abnormal bone to at least one of the point cloud and the surface model of the abnormal bone, where the abnormal bone includes at least one of an incomplete bone, a deformed bone, and a fractured bone; and (b) generating a reconstructed model of the abnormal bone.

[0009] In a further detailed embodiment of the second aspect, the step of utilizing at least one of the point cloud and the surface model of the abnormal bone comprises identifying a statistical atlas bone that is most similar to the abnormal bone. In yet another further detailed embodiment, the step of utilizing at least one of the point cloud and the surface model of the abnormal bone comprises registering the statistical atlas to the abnormal bone. In a further detailed embodiment, the step of utilizing at least one of the point cloud and the surface model of the abnormal bone comprises morphing surface points on the reconstructed model of the abnormal bone to at least one of the point cloud and the surface model of the abnormal bone. In yet a further detailed embodiment, the step of identifying a statistical atlas bone that is most similar to the abnormal bone comprises using one or more similarity measures to identify the statistical atlas bone. In a further detailed embodiment, the statistical atlas comprises a plurality of mathematical representations, where each of the plurality of mathematical representations represents a bone. In a further detailed embodiment, the statistical atlas comprises a plurality of virtual models, where each of the plurality of virtual models represents a bone. In yet another more detailed embodiment, the method further comprises registering at least one of the point cloud and surface model of the abnormal bone to an identified bone from the statistical atlas that resembles the abnormal bone. In yet another more detailed embodiment, the method further comprises augmenting shape parameters between (a) at least one of the point cloud and surface model of the abnormal bone and (b) the identified bone from the statistical atlas that resembles the abnormal bone. In yet another more detailed embodiment, augmenting shape parameters comprises interpolating between (a) at least one of the point cloud and surface model of the abnormal bone and (b) the identified bone from said statistical atlas that resembles the abnormal bone to identify missing or deformed bones in at least one of the point cloud and surface model of the abnormal bone.

[0010] In yet another more detailed embodiment of the second aspect, the augmentation of the shape parameters results in the generation of surface points corresponding to the missing or deformed bone. In yet another more detailed embodiment, the method further comprises morphing surface points interpolated from a bone from a statistical atlas similar to the abnormal bone with at least one of the point cloud and the surface model of the abnormal bone to generate a reconstructed model of the abnormal bone. In a further detailed embodiment, the abnormal bone comprises at least one of a deformed pelvic portion, a crushed pelvic portion, and an incomplete pelvic portion lacking a bone, and the reconstructed model of the abnormal bone comprises at least a full pelvic model portion in which at least one of the bone deformation in the deformed pelvic portion, the crushed bone comprising a portion of the crushed pelvic portion, and the bone missing from the incomplete pelvic portion has been restored. In yet a further more detailed embodiment, the full pelvic model portion comprises an acetabular cup anatomy. In a further detailed embodiment, the abnormal bone comprises at least one of a deformed femoral portion, a crushed femoral portion, and an incomplete femoral portion lacking bone, and the reconstruction model of the abnormal bone comprises at least a complete femoral model portion in which at least one of the bone deformation in the deformed femoral portion, the crushed bone comprising a portion of the crushed femoral portion, and the bone defect from the incomplete femoral portion has been repaired. In a further detailed embodiment, the complete femoral model portion comprises a proximal femur having a neck and a ball anatomy.In yet another detailed embodiment, the abnormal bone comprises at least one of a deformed humerus portion, a crushed humerus portion, and an incomplete humerus portion lacking bone, a deformed ulna, a crushed ulna, an incomplete ulna lacking bone, a deformed radius portion, a crushed radius portion, an incomplete radius portion lacking bone, a deformed cranial portion, a crushed cranial portion, an incomplete cranial portion lacking bone, a deformed vertebral portion, a crushed vertebral portion, and an incomplete vertebral portion lacking bone, and the reconstructed model of the abnormal bone comprises at least one of a deformed ulnar portion, a crushed ulnar portion, and an incomplete vertebral portion lacking bone, and at least one of a complete humerus model part, a complete ulna model part, a complete radius model part, a complete skull model part, and a complete vertebral model part in which at least one of a comminuted bone comprising a part of a comminuted radius part, a bone defect from an incomplete radius part, a bone defect in a deformed cranial part, a comminuted bone comprising a part of a comminuted skull part, a bone defect from an incomplete skull part, a bone defect in a deformed vertebral part, a comminuted bone comprising a part of a comminuted vertebral part, and a bone defect from an incomplete vertebral part has been repaired.

[0011] A third aspect of the present invention provides a method for constructing a mass-customized orthopedic implant, the method comprising the steps of: (a) identifying features across a statistical atlas population of bones, the features comprising at least one of landmarks and shape features; (b) generating descriptors associated with an implant design using the features identified across the statistical atlas population of bones; (c) classifying at least a portion of the descriptors into groups having similar descriptors; (d) parameterizing the groups to extract parameters from the groups; and (e) generating an electronic design file for the mass-customized orthopedic implant.

[0012] In a more detailed embodiment of the third aspect, the method further comprises manufacturing the mass-customized orthopedic implant using the electronic design file. In yet another more detailed embodiment, the step of identifying features is performed automatically by a software program configured to calculate landmarks over the entire statistical atlas population of bones using position parameters incorporated in the computational logic. In a further detailed embodiment, the step of identifying features is performed automatically by a software program configured to calculate shape features over the entire statistical atlas population of bones using position parameters incorporated in the computational logic. In a still further detailed embodiment, the descriptors comprise mathematical descriptors calculated over the entire statistical atlas population of bones. In a further detailed embodiment, the step of classifying at least some of the descriptors into groups having similar descriptors comprises using statistical analysis to define the groups. In a further detailed embodiment, the descriptors extracted from the groups comprise design parameters for the shape of the mass-customized orthopedic implant. In another further detailed embodiment, the descriptors comprise mathematical descriptors, and the step of parameterizing the groups to extract the descriptors from the groups comprises converting the mathematical descriptors into surface descriptors. In yet another detailed embodiment, the electronic design file for the mass-customized orthopaedic implant includes a virtual three-dimensional model of the mass-customized orthopaedic implant. In yet another more detailed embodiment, parameterizing the groups to extract descriptors from the groups includes generating a virtual three-dimensional model of the mass-customized orthopaedic implant.

[0013] In yet another more detailed embodiment of the third aspect, the method further comprises extracting three-dimensional cancellous bone features across the statistical atlas population of bones and generating a three-dimensional bone model for each bone in the statistical atlas population of bones incorporating the extracted cancellous bone features specific to that bone. In yet another more detailed embodiment, the method further comprises performing a porosity assessment for each bone in the statistical atlas population of bones to determine cancellous bone size and pore size. In a further detailed embodiment, the method further comprises performing a stress testing process combining the cancellous bone size data, pore size data, and surface descriptor parameters to generate an electronic design file for the mass-customized orthopaedic implant. In a still further more detailed embodiment, the electronic design file comprises at least one of a computer-aided design file, a computer numerical control file, and a rapid manufacturing instruction file. In a further detailed embodiment, the method further comprises generating an electronic design file for the mass-customized implant placement guide using at least one of the extracted parameters. In further detailed embodiments, the method further comprises manufacturing the mass-customized implant placement guide using the electronic design file for the mass-customized implant placement guide. In other further detailed embodiments, the statistical atlas population of bones is ethnicity specific. In yet other further detailed embodiments, the statistical atlas population of bones is gender specific. In yet other further detailed embodiments, the statistical atlas population of bones comprises at least a segment of the femur. In yet other further detailed embodiments, the statistical atlas population of bones comprises at least a segment of the pelvic bone.

[0014] A fourth aspect of the present invention provides a method for constructing a mass customized trauma plate, the method comprising: (a) defining a virtual boundary for the mass customized trauma plate relative to a virtual three-dimensional bone model template; (b) selecting a plurality of surface points within the virtual boundary that correspond to surface locations on the virtual three-dimensional bone model template; (c) propagating the plurality of surface points throughout a statistical atlas including a plurality of virtual three-dimensional bone models; (d) constructing a virtual three-dimensional bone plate that is fitted to each of the plurality of virtual three-dimensional bone models using the plurality of surface points propagated to that particular bone model; (e) extracting a plurality of curvatures that represent each of the formed virtual three-dimensional bone plates; (f) statistically analyzing the extracted plurality of curvatures to estimate shape parameters for the mass customized trauma plate; and (g) generating an electronic design file for the mass customized trauma plate using the shape parameters.

[0015] A fifth aspect of the present invention provides a method of constructing a patient-specific cutting guide for preparing bone for an orthopedic implant, the method comprising the steps of: (a) processing a patient-specific bone contour to determine a size of an orthopedic implant to be attached to the patient's bone and a position of the implant when attached to the patient's bone; (b) designing a patient-specific cutting guide using the size of the orthopedic implant and the position of the implant when attached to the patient's bone; and (c) manufacturing the patient-specific cutting guide including a shape that is a negative of the shape of the patient's bone to which the cutting guide is configured to be attached. [Brief description of the drawings]

[0016] [Figure 1] FIG. 1 is a schematic diagram of the overall process of generating a mass-customized patient-specific mold from imperfect anatomy. [Diagram 2] FIG. 1 is a schematic detailing how new anatomical structures are added to the statistical atlas to generate correspondences. [Diagram 3]3 is an overview of a multi-resolution 3D registration algorithm corresponding to the multi-resolution 3D registration in FIG. 2. [Figure 4] Multi-scale registration of feature points using multi-scale features. [Diagram 5] A low-level decomposition of the multi-resolution registration shown diagrammatically in FIG. [Figure 6] It is a graphical representation that captures the change in the population during the generation of the response. [Figure 7] 1 is a schematic diagram of a complete bone reconstruction process using incomplete, deformed or crushed anatomy. [Figure 8] FIG. 2 is a schematic diagram of a defect classification process for generation of a defect template. [Figure 9] 1 is a schematic example of the existing AAOS classification for acetabular defects. [Figure 10] 1 is a schematic example of the existing Paprosky acetabular defect classification. [Figure 11] 1 is a three-dimensional model representation of a patient with significant pelvic discontinuity on the left side. On the right is an example of a three-dimensional model of the pelvis of the patient shown on the left. [Figure 12] A comparison of the reconstructed left side model and the original patient model, as well as the left and right anatomy. [Figure 13] 13 is a distance map between the mirror image of the reconstructed pelvis model and the reconstructed model. [Figure 14] Patient with complete pelvic discontinuity and reconstruction results with an rms error of 1.8 mm. [Figure 15] Average distance map of reconstruction results for an incomplete skull and reconstruction error. [Figure 16] This is the result of reconstruction of a shattered femur. [Figure 17] FIG. 1 is a schematic diagram of a process for forming a patient-specific reconstructive implant. [Figure 18] FIG. 18 is a schematic diagram of the process for generating the implant depicted in FIG. 17. [Figure 19] FIG. 1 is a process flow diagram showing various steps for reconstruction of a patient's complete anatomy from an incomplete anatomy and creation of a patient-specific cup implant for a pelvic discontinuity. [Figure 20] 1 is a graphical representation of a patient-specific placement guide for a patient-specific acetabular implant. [Figure 21] Provides images examining the relationship between the three attachment sites of the implant and cup orientation for mass customization. [Figure 22] 1 is a schematic diagram of a method for manufacturing mass produced acetabular components using a modular design. [Diagram 23] 1 is a schematic diagram of a process for producing a patient-specific hip stem for reconstructive surgery. [Figure 24] FIG. 1 is a schematic diagram of a process for mass-customized implant production. [Diagram 25] FIG. 1 is a schematic diagram illustrating a process for using a statistical atlas to generate both mass-customized and patient-specific hip implants. [Figure 26] FIG. 1 is a schematic diagram illustrating a process for using a statistical atlas to generate both mass-customized and patient-specific hip implants. [Figure 27] FIG. 1 is a schematic diagram outlining a process for designing population-specific hip stem components. [Figure 28] 13 is a graphical representation showing where the proximal femoral landmarks are located. [Figure 29] 3D model of a femur showing the femoral canal waist and femoral waist at the center of the femur along the length of the femur. [Diagram 30] 13 is a graphical representation showing where the proximal femoral axis is located. [Diagram 31] 13 is a graphical representation showing where the femoral neck center calculation is located. [Diagram 32]1 is a graphic representation of the two points used to define the anatomical axis of the proximal femur. [Diagram 33] 1 is a graphical representation of 3D proximal femoral measurements. [Diagram 34] Shown are typical Dorr ratios, typically 2D (from XR). [Diagram 35] 1 is a graphic representation of the B / A ratio at the IM isthmus. [Diagram 36] 1 is a graphic representation of IM femoral canal measurements. [Figure 37] The contour and approximate circle. [Figure 38] 1 is a graphical representation of the measurements taken to obtain the IM femoral canal radius ratio. [Figure 39] Two femur models are depicted showing the effect of changing the radius ratio, in this case the left femur model has a radius ratio of 0.69 and the right femur model has a radius ratio of 0.38. [Diagram 40] 1 is a graphical representation of the medial contour, femoral neck axis, and femoral head point of the proximal femur before alignment. [Diagram 41] 1 is a graphical representation of anatomical axis alignment with the Z direction. [Diagram 42] 13 is a graphical representation of the medial contour aligned using the femoral neck pivot point. [Diagram 43] 1 is a graphical representation of different models generated using interpolation between models to illustrate the smoothness of the interpolation. [Diagram 44] 1 is a graphical and pictorial representation of three-dimensional mapping of bone density. [Diagram 45] X-ray depictions showing three heights: proximal axis, femoral head offset, and IM width at the femoral head. [Figure 46] 13 is a plot showing the relationship between proximal angle and femoral head offset. [Figure 47] 1 is a plot showing the relationship between proximal angle and femoral head height. [Figure 48]1 is a plot showing the relationship between femoral head offset and femoral head height. [Figure 49] 13 is a proximal angle histogram. [Figure 50] 1 is a plot depicting a group of females and males with respect to femoral head offset and femoral calcaneus diameter. [Figure 51] 1 is a plot depicting a group of females and males in terms of femoral head offset and proximal angle. [Figure 52] 1 is a femoral head offset histogram. [Diagram 53] This is an IM size histogram. [Figure 54] 1 is a graphical representation of female measurements for the proximal femur. [Figure 55] 1 is a graphical representation of male measurements for the proximal femur. [Figure 56] 1 is a graphical representation of a woman's measurements for greater trochanter height. [Figure 57] 1 is a graphical representation of male measurements for greater trochanter height. [Figure 58] Differences in IM canal shape between the sexes. [Figure 59] Normal female, ie, T-score 1.1. [Figure 60] Women with osteopenia, i.e., T-score 1.3. [Figure 61] Women with osteoporosis, i.e., T-score 3. [Figure 62] 13 is an interpolated data set femoral head offset histogram. [Figure 63] 13 is a data set femoral canal size histogram. [Figure 64] AP femoral head height measurement. [Figure 65] Femoral head height relative to the pivot point versus AP femoral head height. [Figure 66] Femoral head height relative to the anatomical axis midpoint vs. AP femoral head height. [Figure 67]13A-13D are parameters used for the formation of hip stem implant groups that accommodate both ethnicity and gender differences.14A-13D are mass-custom implant shape parameters for femoral stem components extracted from clustering. [Figure 68] 1A and 1B are assembled and exploded views of the original femoral stem. [Figure 69] 1A and 1B are assembled and exploded views of a repair femoral stem. [Figure 70] 1 is an isolated acetabular cup geometry. [Figure 71] 1 is an anatomical template for an acetabular cup. [Figure 72] An anatomical acetabular cup and femoral stem ball shape with multiple cup radii. [Figure 73] We show that the curvature match between the acetabular cup and femoral head curvature influences kinematics and constraints. [Figure 74] 1 is a contour defining a cross-sectional analysis of an acetabular cup. [Figure 75] The transverse acetabular ligament is automatically detected as a method for cup orientation. [Figure 76] Extraction from micro-CT of porous shape and size that matches bone anatomy. This process is misrepresented. Source unknown. [Figure 77] Pet specific implant and cutting guide. The similar process outlined for human patients is incorrect. Source unknown. [Figure 78] Mass-customized orthopedic implants for pets using statistical atlases. The same process outlined for mass-customized implants in humans is incorrect. Source unknown. [Figure 79] 1 is a process for creating patient-specific cutting and placement guides for a hip joint system. [Figure 80]The process of non-rigid registration for the creation of patient-specific 3D pelvis and proximal femur models from x-rays. This is an incorrect detailed overview of the non-rigid registration components. Source unknown. [Figure 81] Multiple x-rays used for the reconstruction of the pelvis and proximal femur, incorrectly outlined. Source unknown. [Figure 82] FIG. 79 shows an automated segmentation of the pelvis and proximal femur from MRI and CT scans. [Figure 83] FIG. 79 is an automated segmentation of complex pulverized anatomy from MRI or CT shown diagrammatically. [Figure 84] A virtual templating process for both the acetabular cup and femoral stem components. [Figure 85] Automated stem placement using tip fixation. [Figure 86] The stem is automatically placed using a press fit and three contact points. [Figure 87] Automatic pelvic landmarking. [Figure 88] Automatic calculation of cup orientation and placement. [Figure 89] Evaluation of cup and stem placement. [Figure 90] Evaluation of cup and stem placement to ensure overall limb length restoration and orientation. [Figure 91] A pre-planning interface for evaluating and correcting implant placement and sizing. [Figure 92] The process of using a patient-specific guide for the resection and placement of a femoral stem. [Figure 93] A process that uses a patient-specific guide for reaming and placement of an acetabular cup. [Figure 94]A statistical atlas or template may be used to determine the patient-specific guide mating site, a mapping of the patient-specific labrum attachment site, in this example the acetabular labrum attachment site, used for the generation of the patient-specific guide and locking mechanism. [Figure 95] A process for forming trauma plates and fixation devices for a population. [Figure 96] Localization of plate shapes in the atlas average bone. [Figure 97] Here the propagation of plate positions throughout a population is shown in a single case. [Figure 98] Extraction of the plate midline curve. [Figure 99] Calculation of the 3D radius of curvature at the plate midline curve. [Figure 100] Calculation of plate length. [Figure 101] Calculation of intermediate plate width. [Figure 102] Calculation of cross-sectional radius. [Figure 103] Determining the optimal number of clusters. [Figure 104] Plate size clustering, shown in Figure 95 as "clustering". [Figure 105] Parameterization of plate size. Shown in Figure 95 as "Parameterized curve" and "Generative model". [Figure 106] The plates generated are seeded into populations for evaluation. [Figure 107] 3D surface distance map between the plate surface and bone for assessing plate fit. [Figure 108] Validation of design plates on cadavers to avoid muscle and ligament impingement. [Fig. 109] Identification of the clavicular midline curvature. The midline curvature is not symmetrically "S" shaped according to statistical analysis of anatomical populations. [Figure 110]Superior plate (left), plate midline curve (center), and midline plate curvature showing radius of curvature (right). [Figure 111] Anterior midshaft 7h plate (left), plate midline curve (center), and midline plate curvature showing a single radius of curvature (right). [Figure 112] Upper midshaft plate (left), plate midline curve (center), and midline plate curvature showing different radii of curvature (right). [Figure 113] Anterior lateral plate (left), plate midline curve (center), and midline plate curvature showing different radii of curvature (right). [Fig. 114] Anterior midshaft long plate (left), plate midline curve (center), and midline plate curvature showing different radii of curvature (right). [Figure 115] A process for generating a customized plate placement guide for trauma reconstruction. [Fig. 116] A process for producing customized cutting and placement guides for reconstruction surgery using bone grafts. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0017] Exemplary embodiments of the present disclosure are described and illustrated below to cover various aspects of orthopedics including bone and tissue reconstruction, patient-specific and mass-customized orthopedic implants, gender- and ethnicity-specific orthopedic implants, cutting guides, trauma plates, bone graft cutting guides and bone graft placement guides, and patient-specific instruments. Of course, it will be apparent to those skilled in the art that the embodiments discussed below are exemplary in nature and may be reconfigured without departing from the scope and spirit of the invention. However, for clarity and precision, the exemplary embodiments discussed below may include optional steps, methods, and features that those skilled in the art will recognize as not essential to fall within the scope of the invention.

[0018] Complete anatomy reconstruction With reference to Figures 1-8, reconstruction of deformed or incomplete anatomy is one of the complex problems facing healthcare providers. Loss of anatomy may be the result of birth conditions, tumors, disease, physical injuries, or failure of previous surgery. As part of providing treatment for various diseases, healthcare providers may find it advantageous to reconstruct or configure anatomy to facilitate treatment for various conditions, which may include, but are not limited to, fracture / bone comminution, bone degeneration, orthopedic implant repair, joint degeneration, and custom instrumentation design. For example, prior art hip joint reconstruction solutions require mirroring of a healthy patient's anatomy, which may not be an accurate reflection of healthy anatomy due to inevitable asymmetries, as shown in Figures 12-16.

[0019] The present disclosure provides systems and methods for bone and tissue reconstruction. To perform this reconstruction, the systems and associated methods utilize anatomical image representations of one or more humans. These images are processed to generate a virtual three-dimensional (3D) tissue model or a series of virtual 3D tissue models that mimic the appropriate anatomy of the subject. The systems and associated methods are then utilized to generate molds and / or other devices (e.g., fixation devices, implantation devices, patient-specific implants, patient-specific surgical guides) for use in conjunction with reconstructive surgery.

[0020] As depicted in FIG. 1, an overview of a typical system flow begins with receiving input data representing an anatomy. The anatomy may comprise an incomplete anatomy in the case of tissue degeneration or tissue absence due to genetic characteristics, or the anatomy may comprise a deformed anatomy due to genetic characteristics or environmental conditions, or the anatomy may comprise shattered tissue due to one or more anatomy breaks. The input anatomical data comprises two-dimensional (2D) images or three-dimensional (3D) surface representations of the target anatomy, which may be in the form of, for example, a surface model or a point cloud. In situations where 2D images are utilized, these 2D images are utilized to construct a 3D virtual surface representation of the target anatomy. Those skilled in the art are familiar with utilizing 2D images of anatomy to construct a 3D surface representation. Thus, a detailed description of this process has been omitted to promote brevity. By way of example, the input anatomical data may comprise one or more of X-ray images, computed tomography (CT) scan images, magnetic resonance images (MRI), or any other imaging data that can generate a 3D surface representation of the target tissue.

[0021] With reference to Figure 45 and Table I, it has been found that in the context of x-ray images used to construct virtual 3D bone models, bone rotation during imaging plays an important role in accurately constructing the model. In other words, when attempting to create x-ray images in situations where bone rotation has occurred between images, the x-ray images need to be normalized to take this bone rotation into account.

[0022] As an example, it has been found that in the context of the proximal femur, 6 degrees and 15 degrees of bone rotation result in significant changes in measurements extracted from the x-ray images. As an example, these measurements include, but are not limited to, proximal angle, femoral head offset, and intramedullary canal width. As reflected in Table I, for the same femur x-rayed at 0 degrees (i.e., the starting point defined by the initial x-ray), 6 degrees of rotation and 15 degrees of rotation show differences in the proximal angle, femoral head offset, and intramedullary canal width measured using pixels, where each pixel size was approximately 0.29 millimeters. Notably, the proximal angle increased with increasing rotation, as did the femoral head offset, but the same was not true for the intramedullary canal width. In this exemplary table, three cross sections were spaced along the longitudinal axis, where each plane corresponded to a location where the intramedullary canal width was measured. As reflected in Table I, the intramedullary canal width at the same location changes with the rotation angle. As a result, as discussed in more detail below, when constructing a 3D virtual model of a bone using x-rays, rotational bias must be taken into account to the extent that bone rotation occurs during imaging.

[0023] However, it should be understood that the above is an exemplary description of biological structures that may be used with the exemplary systems and methods, and as such is in no way intended to limit other biological structures from being used with the present systems in accordance with the disclosed methods. As used herein, tissue includes bone, muscle, ligament, tendon, and any other distinct type of structural material with a specific function in a multicellular organism. As a result, when the exemplary systems and methods are discussed in the context of bone, one of ordinary skill in the art should recognize the applicability of the systems and methods to other tissues.

[0024] Referring back to FIG. 1 , the bioanatomical data input into the system is directed to three modules, two of which involve processing the bioanatomical data (full bone reconstruction module, patient-specific module), while the third (anomaly database module) classifies the bioanatomical data as part of a database. The first processing module, i.e., the full bone reconstruction module, processes the input bioanatomical data together with data received from the statistical atlas module to generate a virtual 3D model of the subject's bone. This 3D model is a complete normal reconstruction of the subject's bone. The second processing module, i.e., the patient-specific module, processes the input bioanatomical data together with data received from the full bone reconstruction module to generate one or more final orthopedic implants, as well as one or more molds, fixation systems, implant forming tools, and renderings. A rendering is a visualization of the reconstructed anatomy for feedback regarding the expected surgical outcome. More specifically, the patient-specific module is adapted to generate a fully customized device that is designed to precisely fit the patient's specific anatomy despite significant deviations from normality of the patient's anatomy. The patient-specific module also utilizes the virtual 3D reconstructed bone model from the full bone reconstruction module to automatically identify anatomical regions and features (e.g., fitting regions and / or shapes) for device design parameters. In this aspect, the patient-specific data is used to define design parameters so that the output instrument and any implants precisely fit the patient's specific anatomy. Exemplary uses of the patient-specific module are discussed in more detail below. To further understand the system's functionality and processes, the following is a description of the system's modules, starting with the statistical atlas module.

[0025] As shown in Figures 1 and 2, the statistical atlas module records a virtual 3D model of one or more anatomical structures (e.g., bones) to capture inherent anatomical variability in a given population. In a typical form, the atlas records mathematical representations of anatomical features of one or more anatomical structures expressed as a mean representation and a change about the mean representation. By representing the anatomical features as mathematical representations, the statistical atlas allows for automated measurement of the anatomical structures as well as reconstruction of missing anatomical structures, as discussed in more detail below.

[0026] To extract anatomical variations across common anatomical structures, the input anatomical data is compared to a common reference frame across the population, commonly referred to as a template 3D model or anatomical 3D template model. This template 3D model is visually displayed on a graphic display as a 3D model that can be rotated or otherwise visually manipulated, but comprises a mathematical representation of the anatomical surface features / representations for all anatomical structures across the statistical atlas in the target tissue (i.e., for a given bone that shares all the properties of the bone across the statistical atlas population, generated from the template 3D model). The template 3D model may be a combination of multiple anatomical representations or a single representative example, and may represent the lowest entropy state of the statistical atlas. For each anatomical structure (i.e., input anatomical data) to be added to the statistical atlas, an anatomical 3D model is formed and both the anatomical 3D model and the template 3D model are subjected to a normalization process.

[0027] During the normalization process, the anatomical 3D model is normalized to the scale of the template 3D model. The normalization process may involve scaling one or both of the anatomical 3D model and the template 3D model to have a common unit scale. After normalization of the anatomical 3D model and the template 3D model, the normalized anatomical 3D model and the template 3D model are made scale invariant, which allows the use of shape features independent of scale (meaning size in this case). After normalization is completed, both 3D models are processed by a scale space mapping and feature extraction sequence.

[0028] Scale space mapping and feature extraction is essentially a multi-resolution feature extraction process. In particular, this process extracts shape-specific features at multiple feature scales. First, multiple anatomical features are selected, each representing a feature present in a different scale space. Then, model-specific features are extracted for each scale-space representation of the selected anatomical features. These extracted features are used to derive robust (with respect to noise) registration parameters between the template 3D model and the anatomical 3D model. After this multi-resolution feature extraction process, the extracted data is processed by a multi-resolution 3D registration process.

[0029] With reference to Figures 2-5, the multi-resolution 3D registration process uses scale-space extracted features to perform affine registration calculation between the anatomical 3D model and the template 3D model to align the two models. In particular, the anatomical 3D model and the template 3D model are processed by a rigid registration process. As depicted in Figure 5, this rigid registration process acts to align the anatomical 3D model and the template 3D model so that both models are in the same space and are free of pose singularities. To align the 3D models, the centroids associated with each model are aligned. Also, the principal axes in each of the 3D models are aligned so that the principal directions of both 3D models are the same. Finally, the pose difference between the 3D models is minimized by performing an iterative closest point calculation.

[0030] After rigid registration, the 3D models are aligned using a similarity registration process. This process involves iteratively matching the template 3D model and the anatomical 3D model in normal scale by computing a similarity transformation that best matches normal scale features (i.e., ridges) for both the template 3D model and the anatomical 3D model. The iterative similarity alignment algorithm is a variant of iterative closest point. Within each iteration rotation, translation and scale are computed between point pairs until convergence. Pairwise matching or correspondence between the two sets of points is evaluated using a distance query that is computed using a Kd-tree or some other space partitioning data structure. In particular, the ridges in both models are utilized to perform the computational matching point pair process. In this exemplary description, a ridge is a point on the 3D model where a single principal curvature has an extreme value along that line of curvature. As part of the computational matching point pair process, points are identified on the ridges of the 3D models that fit together. Next, the edges of both 3D models are subjected to a similarity transformation calculation process, in which the rotation, translation, and scale that best matches the edges of both models are calculated. This is followed by a transformation point process that is adapted to apply the calculated rotation, translation, and scale to the template 3D model edges. This is followed by a calculation of the change in the relative root mean square error or relative distance error from the previous process, after which the root mean square error or distance error between each matched set of points is calculated. If the change in the relative root mean square error or relative distance error is within a predefined threshold, a transformation process is performed to apply the final rotation, translation, and scale to the template 3D model.

[0031] The similarity registration process is followed by an articulated registration process, which receives input data from a scale-space feature process, in which features are extracted from the template 3D model and the anatomical 3D model in different scale spaces, each of which is defined by convolving the original anatomical 3D model with a Gaussian smoothing function.

[0032] The objective of the articulated registration process is to match the "n" scale-space features of the template 3D model with the "m" scale-space features calculated on the anatomical 3D model. The difference between the number of features detected on the template 3D model and the anatomical 3D model is due to anatomical variations. This difference in the number of features detected may result in many relationships between the template 3D model and the anatomical 3D model. Therefore, to accommodate such variations and achieve accurate matching between all mutual features, two-way mutual feature matching is performed. Specifically, a feature set is calculated in scale space with respect to the template 3D model. In this exemplary process, the feature set is a connected set of points that represent raised anatomical structures (e.g., the acetabular cup of the pelvis, the spinal process of the lumbar spine). Similarly, the feature set is calculated in scale space with respect to the anatomical 3D model. The matching feature pair process matches the feature set calculated on the template 3D model to the feature set in the anatomical 3D model using shape descriptors (e.g., curvature index, shape index, etc.). The result of this process is a "nm" mapping of feature sets between the template 3D model and the anatomical 3D model. If necessary, a reorganization process is performed to reorganize the matched feature sets into a single feature set (e.g., if the acetabular cup is detected as two pieces, this process reorganizes the two pieces into one single feature set). Then, a computation process is performed to calculate the correspondence between each point in the matched feature sets on the template 3D model and the anatomical 3D model. Then, an affine computation transformation process follows to calculate the rotation, translation, and shear that transform each matched feature set in the template 3D model to each corresponding feature set in the anatomical 3D model. Then, the template 3D model is transformed using the calculated affine transformation parameters (i.e., rotation, translation, and shear). Finally, a rigid body alignment process is performed to match each matched feature set in the template 3D model and the anatomical 3D model.

[0033] The non-rigid registration process, which is performed after the articulated registration process and the normal scale feature process, involves fitting all surface vertices in the template 3D model to vertices in the anatomical 3D model to calculate an initial correspondence. This correspondence is then used to calculate a deformation region that moves each vertex in the template 3D model to the fitted point in the anatomical 3D model. Matching is performed between vertices in the same class (i.e., scale space feature vertices, normal scale feature vertices, or non-feature vertices). In the context of the normal scale feature process, shape features are calculated in initial scale space (ridges), which represents the initial input model, for the template 3D model and the anatomical 3D model.

[0034] Specifically, as part of the non-rigid registration process, scale-space features are computed for the template 3D model (TMssf) and the anatomical 3D model (NMssf). Each set of features in the template 3D model and the anatomical 3D model is grown using "k" nearest neighbors. An alignment process is applied to match the template 3D model scale-space features to their corresponding features in the anatomical 3D model. Given two point clouds, a reference (X) and a translation (Y), the goal is to iteratively align the two point clouds to minimize the overall error metric, under the constraints of a minimum relative root mean square error and a maximum angle threshold. A realignment process is performed to align the feature set in the template 3D model with the matched set in the anatomical 3D model using the iterative closest points at a normal scale. After realignment, point correspondences are computed between each feature set point in the template 3D model and the matched feature set points in the anatomical 3D model. The matched points in the anatomical 3D model should have a surface normal direction close to the template 3D model points. The output is forwarded to compute the transformation domain step.

[0035] In parallel with the scale-space feature calculation process, the remaining set of non-feature points of the template 3D model (TMnfp) and the anatomical 3D model (NMnfp), or points on the template 3D model surface that do not belong to either the scale-space features or the normal scale features, are processed according to the correspondence calculation to calculate the point correspondences between the non-feature points in the template 3D model and the non-feature points in the anatomical 3D model. The matching points in the new model should have a surface normal direction close to the template model points. The output is forwarded to calculate the deformation region step.

[0036] Also, in parallel with the scale-space feature computation process, normal scale features (i.e., ridges) in the template 3D model (TMnsf) are aligned with normal scale features (i.e., ridges) in the anatomical 3D model (NMnsf) using AICP. AICP is a variant of iterative closest point computation where in each iteration translation, rotation, and scale are computed between a set of matching points. After the alignment process, a correspondence process is performed.

[0037] The outputs from the scale-space feature computation, correspondence and alignment processes are subjected to a deformation process, in which deformation regions are computed to move each point in the template 3D model to its matching point in the anatomical 3D model.

[0038] The output of the non-rigid registration process is subjected to a relaxation process to smooth the output model by moving the vertices of the template 3D model mesh closer to the surface of the anatomical 3D model after the multi-resolution registration step. In particular, the template 3D model in normal space (TMns) and the anatomical 3D model in normal space (NMns) are processed by a correspondence calculation to calculate the vertex in the template 3D model that is closest to the anatomical 3D model using a normal-constrained sphere search algorithm. This calculation uses the closest vertices in both models to generate a correspondence vector from each vertex in the template 3D model and its matching vertex in the anatomical 3D model, which may result in multiple matching points from the anatomical 3D model. Using the matching points at each vertex of the template 3D model, a weighted average of the matching points in the anatomical 3D model is calculated based on the Euclidean distance from that point and the matching point. At this point, the template 3D model is updated using the weighted average to move each point in the template 3D model using the calculated weighted average distance. After the weighting calculation process, a relaxation process is performed for each point in the template model to find the closest point on the anatomical 3D model surface and move it to the respective point. Finally, a smoothing operation is performed on the deformed template 3D model to remove noise. The resulting aligned 3D models (i.e., the template 3D model and the anatomical 3D model) are then subjected to a free-form deformation process.

[0039] The free-form deformation process morphs the surface of the template 3D model with the surface of the anatomical 3D model. More specifically, the surface of the template 3D model is iteratively moved on a weighted point-to-point basis using mutually matched points on both the template 3D model surface and the anatomical 3D model surface.

[0040] With reference to Figures 2 and 6, after the free-form deformation process, the anatomical 3D model is subjected to a correspondence calculation process to determine the deviation between the anatomical 3D model and the morphed template 3D model. This correspondence calculation process refines the template 3D model from the free-form deformation step to perform a final match of the selected landmark positions in the deformed template 3D model and the deformed anatomical 3D model. In this manner, the correspondence calculation process calculates and records the size and shape changes between the 3D models, which are recorded as deviations with respect to the average model. The output of this correspondence calculation process is the addition of the normalized anatomical 3D model and the modified template 3D model, which has been updated to account for the changes in the anatomical 3D model. In other words, the output of the process shown diagrammatically in Figure 2 is a normalized anatomical 3D model, which has been modified to have properties (e.g., point correspondences) that match the modified template 3D model to facilitate a complete anatomical reconstruction (e.g., a complete bone reconstruction).

[0041] With reference to Figures 1 and 7, the input and anatomy data from the statistical atlas module are directed to a full anatomy reconstruction module. As an example, the target anatomy may be a bone or multiple bones. However, it should be noted that anatomy other than bones may be reconstructed using the exemplary hardware, processes, and techniques described herein. In an exemplary form, the full anatomy reconstruction module may receive input data relating to an incomplete, deformed, or fractured pelvis. The input anatomical data comprises two-dimensional (2D) images or three-dimensional (3D) surface representations of the target anatomy, which may be in the form of, for example, a surface model or a point cloud. In situations where 2D images are utilized, these 2D images are utilized to construct a 3D surface representation of the target anatomy. Those skilled in the art are familiar with utilizing 2D images of anatomy to construct a 3D surface representation. Accordingly, a detailed description of this process has been omitted to promote brevity. By way of example, the input anatomical data may comprise one or more of an X-ray image, a computed tomography (CT) scan image, a magnetic resonance image (MRI), or any other imaging data from which a 3D surface representation can be generated. As discussed in more detail below, this input anatomical data may be used, without limitation, for (1) a starting point for identifying the closest statistical atlas 3D bone model, (2) registration using a set of 3D surface vertices, and (3) a final relaxation step of the reconstruction output.

[0042] As depicted in FIG. 7, input anatomical data (e.g., a patient's bone model) is utilized to identify an anatomical model (e.g., a bone model) in the statistical atlas that most closely resembles the anatomy of the patient of interest. This step is depicted in FIG. 3 as finding the closest bone in the atlas. To initially identify the bone model in the statistical atlas that most closely resembles the patient's bone model, the patient's bone model is compared to the bone model in the statistical atlas using one or more similarity measures. The result of the initial similarity measures is the selection of a bone model from the statistical atlas to be used as an "initial guess" for the subsequent registration step. The registration step aligns the selected atlas bone model (i.e., the initial guess bone model) with the patient bone model such that the output is the patient bone model that is aligned with the atlas bone model. After the registration step, the shape parameters for the aligned "initial guess" are optimized so that the shape matches the patient bone shape.

[0043] In this case, shape parameters from the statistical atlas are optimized to minimize the error between the reconstruction and the patient bone model, either by using undeformed or existing bone regions. Changing the shape parameter values ​​allows the display of different anatomical shapes. This process is repeated at different scale spaces until convergence of the reconstructed shape (possibly measured as the relative surface change between iterations or as the maximum number of allowed iterations) is achieved.

[0044] A relaxation step is performed to morph the optimized tissue to best match the initial patient 3D tissue model. According to a typical case, the missing anatomy from the reconstructed pelvis model output from the convergence step is applied to the patient-specific 3D pelvis model, thereby forming a patient-specific 3D model of the patient's reconstructed pelvis. More specifically, the surface points in the reconstructed pelvis model are directly relaxed (i.e., morphed) into the patient-specific 3D pelvis model to best match the reconstructed shape to the patient-specific shape. The output of this step is a fully reconstructed patient-specific 3D tissue model that represents what should be the patient's normal / complete anatomy.

[0045] 1, an anomaly database is utilized as training and data input for the defect classification module. In particular, the anomaly database contains data specific to abnormal anatomical features, including anatomical surface indications and associated clinical and demographic data.

[0046] 1 and 8, a fully reconstructed patient-specific 3D tissue model representing normal / complete tissue and input anatomical data representing abnormal / incomplete tissue from an anomaly database (i.e., a 3D surface representation or data capable of generating a 3D surface representation) are input to a defect classification module. This anatomical data from the anomaly database may be an incomplete anatomy in the case of tissue degeneration or tissue absence due to genetic characteristics, or the anatomy may be a deformed anatomy due to genetic characteristics or environmental conditions (e.g., surgical repair, disease, etc.), or the anatomy may be a shattered tissue due to one or more anatomic breaks. By way of example, the input anatomical data may comprise one or more of an X-ray image, a computed tomography (CT) scan image, a magnetic resonance image (MRI), or any other imaging data capable of generating a 3D surface representation.

[0047] The defect classification module pulls multiple abnormal 3D surface representations from the anomaly database coupled with normal 3D representations of the target anatomy to form a quantitative defect classification system. This defect classification system is used to form a "template" for each defect class or cluster. More generally, the defect classification module classifies anatomical defects into classes of closely related defects (meaning defects with similar geometric, clinical, appearance, or other characteristics) to facilitate the generation of healthcare solutions to treat these defects. The defect classification module uses software and hardware to automatically classify defects as a means to eliminate or reduce discrepancies between preoperative data and intraoperative observer visualization. Traditionally, preoperative radiographs have been interpreted as a means to qualitatively analyze the extent of anatomical reconstruction required, which has resulted in preoperative planning that is haphazard at best. Currently, intraoperative observers make the final decision on the extent of the anatomy defect, often concluding that the preoperative plan that relies on radiographs was flawed or incomplete. As a result, the instant defect classification module improves on current classification systems by reducing inter- and intra-observer variability associated with defect classification and providing a quantitative metric for classifying new defect cases.

[0048] As part of the defect classification module, the module may take as input one or more classification types to be used as initial conditions. For example, in the context of the pelvis, the defect classification module may use as input defect features corresponding to the American Academy of Orthopaedic Surgeons (AAOS) D'Antonio et al. bone defect classification structure. This structure includes four different classes: (1) Type I corresponding to partial bone loss, (2) Type II corresponding to cavitary bone loss, (3) Type III corresponding to a combination of partial and cavitary bone loss, and (4) Type IV corresponding to pelvic discontinuity. Alternatively, the defect classification module may be programmed using the Paprosky bone defect classification structure. This structure includes three different classes: (1) Type I corresponding to a supportive edge without osteolysis, (2) Type II corresponding to a distorted hemisphere with intact supportive struts and less than 2 centimeters of superior medial or lateral translation, and (3) Type III corresponding to more than 2 centimeters of superior translation and severe ischiolysis with destroyed or intact Kohler's lines. The defect classification module may also be programmed with a modified Paprosky bone defect classification structure that includes six different classes: (1) Type 1, corresponding to a supportive rim with no component movement; (2) Type 2A, corresponding to a distorted hemisphere but less than 3 centimeters of superior displacement; (3) Type 2B, corresponding to a larger hemisphere distortion with less than 1 / 3 of the circumferential rim and a supportive dome residue; (4) Type 2C, corresponding to an intact rim, midline-Köhler line movement, and a supportive dome residue; (5) Type 3A, corresponding to more than 3 centimeters of superior displacement and severe sciatic lysis with intact Köhler lines; and (6) Type 3B, corresponding to more than 3 centimeters of superior displacement and severe sciatic lysis with disrupted Köhler lines and a rim defect of more than half the circumference. The defect classification module uses the output classification type and parameters to compare the anatomical data with that of the reconstructed data to discern which classification type the anatomical data most closely resembles and thus corresponds to the resulting assigned classification.

[0049] As an initial step, the statistical atlas addition step involves generating correspondences between the normal atlas 3D bone model and the abnormal 3D bone model. More specifically, the 3D bone models are compared to identify which bones in the normal 3D model are absent in the abnormal 3D model. In a typical form, the missing / abnormal bones are identified by comparing the points on the surface of each 3D bone model to generate a list of distinct points on the surface of the normal 3D bone model that are absent in the abnormal 3D bone model. The system can also record and list (i.e., identify) those surface points that are common between the two models, or can immediately notice that all other points are present in both bone modules (i.e., in both the normal and abnormal bone models) if they are not recorded as points that are absent in the abnormal 3D bone model. Thus, the output of this step is the abnormal 3D bone model with the statistical atlas correspondences and a list of features (points) from the normal atlas 3D bone model, which indicates whether the feature (point) is present or absent in the abnormal 3D bone model.

[0050] After generating the correspondence between the normal atlas 3D bone model (generated from the complete bone reconstruction module) and the abnormal 3D bone model (generated from the input anatomical data), the missing / abnormal region from the abnormal 3D bone model is localized in the normal atlas 3D bone model. In other words, the normal atlas 3D bone model is compared to the abnormal 3D bone model to identify and record the bone defects from the abnormal 3D bone model present in the normal atlas 3D bone model. The localization may be performed in a number of ways, including but not limited to curvature comparison, surface area comparison, and point cloud area comparison. Finally, in a typical form, the missing / abnormal bone is localized as a set of boundary points that identify the geometric boundary of the missing / abnormal region.

[0051] The defect classification module uses the boundary points to extract features from the missing / abnormal regions according to the input clinical data. In a typical form, the extracted features may include shape information, volume information, or any other information used to describe the overall characteristics of the defective (i.e., missing or abnormal) region. These features may be refined based on existing clinical data, such as continuous defect classification data or patient clinical information not necessarily associated with anatomical features (demographic data, medical history, etc.). The output of this step is a mathematical descriptor representing the defective region that is used in subsequent steps to group similar tissue (e.g., bone) deformations.

[0052] The mathematical descriptors are clustered or grouped based on statistical analysis. In particular, the descriptors are statistically analyzed and compared to other descriptors from other patients / cadavers to identify unique defect classes within a given population. Of course, this classification assumes multiple descriptors from multiple patients / cadavers refining the classification and identification of distinct groups as the number of patients / cadavers increases. The output from this statistical analysis is a set of defect classes that are used to classify new input anatomical data, determining the number of templates.

[0053] The output of the defect classification module is directed to a template module. In an exemplary form, the template module includes data that is unique for each of the defect classifications identified by the defect classification module. As an example, each template for a given defect classification includes a surface representation of the defective bone, the location of the defect, and measurements associated with the defective bone. This template data may be in the form of surface shape data, a point cloud representation, one or more curvature profiles, dimensional data, and physical quantity data. The output from the template module and the statistical atlas is utilized by a mass customization module to design, inspect, and enable the manufacture of mass-customized implants, fixation devices, instruments, or molds. Exemplary utilization of the mass customization module is discussed in more detail below.

[0054] Patient-specific reconstruction implants With reference to Figures 1 and 17, an exemplary process and system for generating a patient-specific orthopedic implant guide and associated patient-specific orthopedic implant for a patient suffering from incomplete, deformed, and / or crushed anatomy will be described. For the purposes of the exemplary discussion, a total hip arthroplasty for a patient with incomplete anatomy will be described. However, it will be appreciated that the exemplary process and system can be applied to any orthopedic implant that is amenable to patient-specific customization in cases where incomplete or deformed anatomy is present. For example, the exemplary process and system can be applied to total shoulder replacement and knee replacement where bone degeneration (incomplete anatomy), bone deformity, or crushed bone is present. As a result, although a hip implant is discussed below, one skilled in the art will appreciate the applicability of the system and process to other orthopedic implants, guides, tools, etc. for use with an original orthopedic procedure or an orthopedic repair procedure.

[0055] Pelvic discontinuity is a specific form of bone loss most often associated with total hip arthroplasty (THA), where osteolysis or acetabular destruction can cause the upper surface of the pelvis to become separated from the lower surface. The amount and severity of bone loss, as well as the potential for biological ingrowth of the implant, are some of the factors that can influence the choice of treatment for a particular patient. In cases of severe bone loss and loss of pelvic integrity, a custom-made tri-flange cup may be used. This implant, first introduced in 1992, has several advantages over existing cages. It can provide stability to the pelvic discontinuity, eliminate the need for structural grafts and intraoperative contouring of the cage, and promote osseointegration of the construct to the surrounding bone.

[0056] Regardless of the context, where the patient's incomplete, deformed, and / or crushed anatomy may be unresolved, the exemplary system and process for generating patient-specific implants and / or guides utilizes the aforementioned exemplary process and system of 3D bone model reconstruction to generate a three-dimensional model of the patient's reconstructed anatomy (see FIGS. 1-7 and the aforementioned exemplary discussion of these figures). More specifically, in the context of a hip arthroplasty involving a pelvic discontinuity, the exemplary patient-specific system uses the patient pelvis data to generate a side-specific (right or left) 3D model of the patient's complete pelvis. As a result, a discussion of the system and process for generating a 3D reconstructed model of the patient's anatomy utilizing the patient anatomy data for the incomplete anatomy is omitted to promote brevity. Thus, a description of the process and system for generating patient-specific orthopedic implant guides and associated patient-specific orthopedic implants for patients afflicted with incomplete, deformed, and / or crushed anatomy will be described following the formation of the three-dimensional reconstruction model.

[0057] Specifically referring to Figures 17-19 and 23, after the patient-specific reconstructed 3D bone models of the pelvis and femur are generated, both the incomplete patient-specific 3D bone models (for the pelvis and femur) and the reconstructed 3D bone models (for the pelvis and femur) are utilized to form a patient-specific orthopedic implant and a patient-specific placement guide for the implant and / or its fastener. In particular, the defect geometry extraction step includes generating correspondences between the patient-specific 3D model and the reconstructed 3D model (correspondences between the pelvis models and between the femur models, but not between one femur model and one pelvis model). More specifically, the 3D models are compared to identify which bones in the reconstructed 3D model are not present in the patient-specific 3D model. In a typical form, the missing / abnormal bones are identified by comparing points on the surface of each 3D model to generate a list of distinct points on the surface of the reconstructed 3D model that are not present in the patient-specific 3D model. The system can also record and list (i.e., identify) those surface points that are common between the two models, or, if not recorded as points that are not present in the patient-specific 3D model, can immediately notice that all other points are common to both bone modules (i.e., in both the reconstructed 3D model and the patient-specific 3D model).

[0058] Referring to FIG. 18, after generating the correspondence between the reconstructed 3D model (generated from the complete bone reconstruction module) and the patient-specific 3D model (generated from the input anatomical data), the missing / abnormal regions from the patient-specific 3D model are localized in the reconstructed 3D model. In other words, the reconstructed 3D model is compared to the patient-specific 3D model to identify and record the bones that are missing from the patient-specific 3D model that are present in the reconstructed 3D model. The localization may be performed in a number of ways, including but not limited to curvature comparison, surface area comparison, and point cloud area comparison. Finally, in a typical form, the missing / abnormal bones are localized and the output comprises two lists: (a) a first list identifying vertices corresponding to bones of the reconstructed 3D model that are not present in the patient-specific 3D model or that are deformed in the patient-specific 3D model, and (b) a second list identifying vertices corresponding to bones of the reconstructed 3D model that are also present in the patient-specific 3D model and are normal.

[0059] 18, 19, and 23, after the defect geometry extraction step, an implant positioning step is performed. The two vertex lists from the defect geometry extraction step and the 3D model of the normal bone (e.g., pelvis, femur, etc.) from the statistical atlas (see Figs. 1 and 2 and the above exemplary discussion of Figs. 1 and 2) are input to identify fixation positions for the femoral implant or the pelvic implant. More specifically, the fixation positions (i.e., implant positions) are automatically selected such that each is located where the patient has remaining bone. Conversely, the fixation positions are not selected in defective areas of the patient's remaining bone. In this way, the fixation positions are selected independent of the final implant morphology / shape. The selection of the fixation positions may be automated using the shape information and the statistical atlas positions.

[0060] As shown in FIG. 18, after the implant positioning step, the next step is to generate patient-specific implant parameters. To complete this step, an implant parameterization template is input that defines the implant with a set number of parameters sufficient to define the underlying shape of the implant. As an example, in the case of pelvic reconstruction to replace / augment a missing or degenerated acetabular cavity, the implant parameterization template includes an angle parameter for the acetabular cup replacement direction and a depth parameter for accommodating the dimensions of the femoral head. Other parameters for the acetabular implant may include, without limitation, the acetabular cup dimensions, anterior orientation, flange position and shape, fixation screw position and orientation. In the case of a porous implant, the pore position and the structural features of the porosity should be included. As an example, in the case of femoral reconstruction to replace / augment a missing or degenerated femur, the implant parameterization template includes an angle parameter for the femoral head replacement direction, femoral neck length, femoral head offset, proximal angle, and cross-sectional analysis of the external femur and intercondylar cavity. Those skilled in the art will appreciate that the parameters selected to define the underlying shape of the implant will vary depending on the anatomy to be replaced or supplemented. As a result, an exhaustive listing of parameters sufficient to define the underlying shape of the implant is unrealistic. Nevertheless, as depicted in FIG. 19 for example, the reconstructed 3D pelvis model may be utilized to obtain the radius of the acetabular cup, an indication of the pelvic bone that comprises the periphery of the acetabular cup, and an indication of the orientation of the acetabular cup relative to the remaining pelvis. The parameters may also be refined taking into account the implant position so that the implant best / better fits the patient-specific anatomy.

[0061] After compiling a set number of parameters sufficient to define the underlying shape of the implant, the design of the implant begins. More specifically, an initial iteration of the overall implant surface model is constructed. This initial iteration of the overall implant surface model is defined by a combination of an estimated contour and a patient-specific contour in the implantation region. The estimated contour is determined from the reconstructed 3D bone model, the missing anatomical bone, and features extracted from the reconstructed 3D bone model. These features and the location of the implantation site, which can be automatically determined, are used to determine the overall implant shape, as depicted in FIG. 19 for an acetabular cup implant, for example.

[0062] Referring back to FIG. 17, the initial iterations of the entire implant surface model are processed according to a custom (i.e., patient-specific) planning sequence. This custom planning sequence may involve input from the surgeon and engineer as part of the iterative review and design process. In particular, the surgeon and / or engineer may view the entire implant surface model and the reconstructed 3D bone model to determine if changes are required to the entire implant surface model. This review may result in iterations of the entire implant surface model until an agreement is reached between the engineer and the surgeon. The output from this step is a surface model for the final implant, which may be in the form of a CAD file, CNC machine encoding, or rapid manufacturing instructions to form the final implant or a tangible model.

[0063] 17, 19, and 20, the design of the patient-specific orthopedic implant is performed simultaneously or subsequently to the design of the patient-specific placement guide. In connection with the acetabular cup implant, one or more surgical instruments can be designed and manufactured to aid in the placement of the patient-specific acetabular cup, as discussed in the previous exemplary embodiment. In designing the patient-specific implant to have a size and shape to match the size and shape of the remaining bone, the contour and shape of the patient-specific implant may be utilized and incorporated as part of the placement guide.

[0064] In a typical form, the acetabular placement guide comprises three flanges configured to contact the surfaces of the ilium, ischium, and pubis, where the three flanges are interconnected via a ring. The flanges of the placement guide may also be of the same shape, size, and contour as the acetabular cup implant, such that the placement guide assumes the same position planned for the acetabular cup implant. In other words, the acetabular placement guide is formed as a negative imprint of the patient's anatomy (the imperfect surfaces of the ilium, ischium, and pubis) just as the acetabular cup implant does, so that the placement guide fits precisely to the patient's anatomy. However, the implant guide is significantly different from the implant in that it includes one or more fixation holes configured to guide fastener holes and / or drilling for placement. In a typical configuration, the placement guide includes holes that are sized and oriented based on image analysis (e.g., micro-CT) to ensure proper orientation of any drill bits or other guides (e.g., dowels) utilized in securing the acetabular cup implant to the remaining pelvis. The number and orientation of the holes varies depending on the remaining bone, which also affects the shape of the acetabular cup implant. FIG. 20 depicts an example of a patient-specific placement guide for use in a total hip arthroplasty. In other cases, the guide can be formed such that it fits into the implant and only guides the direction of the fixation screws. In this configuration, the guide is formed as a negative of the implant, allowing the guide to be placed directly over the implant. Nonetheless, the incorporation of at least a portion of the size, shape, and contour of the patient-specific reconstruction implant is a theme that carries over regardless of the bone to which the patient-specific implant is intended to be coupled.

[0065] Utilizing the exemplary systems and methods described herein can provide significant information that can result in greater orthopedic placement accuracy, better anatomical integration, and the ability to preoperatively measure true angle and plane orientation via a reconstructed three-dimensional model.

[0066] Creating customized implants using mass-customizable components With reference to Figure 22, an exemplary process and system for producing customized orthopedic implants using mass customizable components will be described. For purposes of exemplary discussion, a total hip arthroplasty for a patient with a significant acetabular defect will be described. However, it will be appreciated that the exemplary process and system may be applied to any orthopedic implant amenable to mass customization in instances where imperfect anatomy is present.

[0067] Significant acetabular defects require specialized procedures and implant components to repair. One approach is a custom tri-flange, which is a fully custom implant consisting of three flanges that attach to the ilium, ischium, and pubis, and an acetabular cup. In contrast to typical processes and systems, prior art tri-flange implants comprise a single complex component that is cumbersome to manufacture and requires the entire implant to be redesigned for each case (i.e., fully patient-specific). Typical processes and systems produce custom tri-flange implants that use fully custom components as well as mass customizable components in a modular manner to allow for custom fit and porosity.

[0068] A pre-planning step according to an exemplary process is performed to determine the orientation of the three flanges relative to the cup, the flange contact locations, and the orientation and size of the acetabular cup. This pre-planning step is performed according to the "Patient-Specific Implant" discussion immediately preceding this section. As an example, the specific locations of the implant fixation are determined according to the implant positioning step using the preamble data inputs discussed in the immediately preceding section. Recall that as part of this implant positioning step, the two vertex lists from the defect geometry extraction step and the 3D model of the normal pelvis from the statistical atlas (see FIGS. 1 and 2 and the above exemplary discussion of FIGS. 1 and 2) are input to identify fixation locations for the custom tri-flanges. More specifically, the fixation locations (i.e., implant locations) are selected such that each is located where the patient has residual bone. In other words, the fixation locations are not selected in the defective areas of the patient's residual pelvis. In this manner, the fixation locations are selected independent of the final implant morphology / shape.

[0069] After determining the fixation location, the tri-flange component (i.e., flange) is formed using the "Patient-Specific Implant" discussion immediately preceding this section. The flange is formed to orient the cup orientation relative to the replacement acetabular cup to provide acceptable articulation function. The flange contact surface is also contoured to match the patient's pelvic anatomy in that the tri-flange contact surface is formed as a "negative" of the pelvic bone surface. The exemplary process of FIG. 20 utilizes the final step of the process depicted in FIG. 14 to rapidly fabricate a flange prototype (or use conventional computer numerical control (CNC) equipment). After the flange is fabricated, further machining or steps may be performed to provide a cavity within which porous material can be added to the tri-flange.

[0070] One part of the Tri-Flange system that does not need to be a custom component is the acetabular cup component. In a typical process, a group of acetabular cups is first manufactured, which provides a base upon which the Tri-Flange system is to be formed. These "blank" cups are kept in stock for use as needed. If a particular porosity is desired for the cup, mechanical features are added to the cup that allow for the pressing of a porous material into the cup. Alternatively, if a particular porosity is desired for the cup, the cup may be coated with one or more porous coatings.

[0071] After the blank cup is formed and any porosity issues are addressed as described above, the cup is made patient-specific by machining the cup to accept the flanges. In particular, using a virtual model of the flange, the system software constructs virtual locking features for the flange, which are converted to machine coding so that the locking features are machined into the cup. These locking features allow the cup to be fastened to the flange such that when the flange is attached to the patient's remaining bone, the cup is properly oriented relative to the residual pelvis. This machining may use conventional CNC equipment to form the locking features into the blank cup.

[0072] After fabrication of the locking features as part of the blank cup, the flanges are attached to the cup using the interface between the locking features. The tri-flange assembly (i.e., the final implant) is subjected to an annealing process to promote a strong bond between the components. After annealing of the tri-flange implant, a sterilization process is performed, followed by appropriate packaging to ensure a sterile environment for the tri-flange implant.

[0073] Mass-customized implant creation With reference to FIG. 24, an exemplary process and system for generating a mass-customized orthopedic implant guide and associated mass-customized orthopedic implant for patients suffering from incomplete, deformed, and / or shattered anatomy will be described. For the purposes of the exemplary discussion, a total hip arthroplasty for a patient requiring an initial joint replacement will be described. However, it will be appreciated that the exemplary process and system can be applied to any orthopedic implant and guide that is amenable to mass customization in cases where an incomplete anatomy exists. For example, the exemplary process and system can be applied to total shoulder replacement and knee replacement where bone degeneration (incomplete anatomy), bone deformity, or shattered bone exists. As a result, although a hip implant is discussed below, one skilled in the art will appreciate the applicability of the system and process to other orthopedic implants, guides, tools, etc. for use with an initial orthopedic procedure or an orthopedic repair procedure.

[0074] A typical process utilizes input data from macroscopic and microscopic perspectives. In particular, the macroscopic perspective involves determining the overall geometry of the orthopedic implant and the corresponding anatomy. Conversely, the microscopic perspective involves considering the shape and structure of the cancellous bone and its porosity.

[0075] The macroscopic view includes a database in communication with a statistical atlas module, which records virtual 3D models of one or more anatomical structures (e.g., bones) to capture inherent anatomical variability in a given population. In a typical form, the atlas records mathematical representations of anatomical features of one or more anatomical structures expressed as mean representations and changes about the mean representations in a given anatomical population. See FIG. 2 and the previous discussion of statistical atlases and how anatomical structures are added to the statistical atlas of a given population. Output from the statistical atlas is directed to an automatic landmarking module and a surface / shape analysis module.

[0076] The automatic landmarking module utilizes input from the statistical atlas (e.g., regions likely to contain a particular landmark) and local geometric analysis to calculate anatomical landmarks for each instance of anatomy in the statistical atlas. This calculation is specific to each landmark. For example, the approximate shape of the region is known, and the location of the landmark to be searched is known relative to the local shape features. For example, locating the medial epicondyle bone point of the distal femur is achieved by refining a search based on the approximate location of the medial epicondyle bone point in the statistical atlas. Thus, it is known that the medial epicondyle bone point is the most medial point in this search window, and therefore a search for the most medial point is performed on each bone model in the medial epicondyle bone region defined in the statistical atlas, where the output of the search is identified as the medial epicondyle bone point landmark. After the anatomical landmarks are automatically calculated for each virtual 3D model in the statistical atlas population, the virtual 3D models of the statistical atlas, along with the shape / surface analysis output, are directed to a feature extraction module.

[0077] The shape / surface output comes from a shape / surface module that also receives input from a statistical atlas. In the context of the shape / surface module, the virtual 3D models in the statistical atlas population are analyzed with respect to shape / surface features that are not covered by the automatic landmarking. In other words, features that correspond to the overall 3D shape of the anatomical structure but do not belong to the features defined in the previous automatic landmarking step are calculated as well. For example, curvature data is calculated for the virtual 3D models.

[0078] The outputs from the surface / shape analysis module and the automatic landmarking module are directed to a feature extraction module. Using a combination of landmarks and shape features, mathematical descriptors relevant to the implant design (i.e., curvature, dimensions) are calculated for each case in the atlas. These descriptors are used as input to the clustering process.

[0079] The mathematical descriptors are clustered or grouped based on statistical analysis. In particular, the descriptors are statistically analyzed and compared to other descriptors from the remaining anatomical population to identify groups (of anatomies) within the population that have similar characteristics. Of course, this clustering assumes multiple descriptors from multiple anatomies across the population. As new examples are presented to the clustering that were not present in the initial clustering, the output clusters are refined to better represent the new population. The output from this statistical analysis is a finite number of implants (including implant groups and sizes) that cover all or most of the anatomical population.

[0080] For each cluster, the parameterization module extracts mathematical descriptors within the cluster. The mathematical descriptors form the parameters (e.g., CAD design parameters) for the final implant model. The extracted mathematical descriptors are fed to the implant surface generation module. This module is responsible for converting the mathematical descriptors into surface descriptors in order to generate a 3D virtual model of the anatomical structure for each cluster. The 3D virtual model complements the microscopic view prior to stress testing and implant manufacturing.

[0081] At a microscopic level, data is obtained that indicates the structural integrity of each biological structure in a given population. In a typical form, this data on bone may comprise micro-CT data that provides structural information on the cancellous bone. More specifically, the micro-CT data may comprise images of the bones of interest (multiple micro-CT images for multiple bones across the population). These images are then segmented via a trabecular structure extraction module to extract the three-dimensional geometry of the cancellous bone and form a virtual 3D model for each bone in the population. The resulting 3D virtual model is the input to the pore size and shape module. As depicted diagrammatically in FIG. 76, the 3D virtual model includes pore size and shape information that is evaluated by the pore size and shape module to determine the pore size and size in the cancellous bone. This evaluation serves to analyze the pore size and shape of the bone in the intramedullary canal so that the stem of the femoral implant can be treated with a coating or otherwise processed to form a porous outer surface to promote integration between the remaining bone of the femur and the femoral implant. The output from this module, combined with the 3D virtual model output from the implant surface generation module, is directed to a virtual stress testing module.

[0082] The stress testing module combines the implant porosity data from the pore size and shape module with the implant shape data from the implant surface generation module to define the final implant shape model and properties. For example, the shape and properties include providing the final implant model with a porous coating that approximately matches the cancellous bone porosity in the target bone. Once the shape and properties are incorporated, the final implant model undergoes virtual stress testing (finite element analysis and mechanical analysis) to verify the functional quality of the model. In the range where the functional quality is unacceptable, the parameters defining the implant shape and porosity are altered until acceptable performance is obtained. Assuming the final implant model meets the stress testing criteria, the final implant model is utilized to generate the machine instructions necessary to convert the virtual model into a tangible implant (which may be further refined by manufacturing processes known to those skilled in the art). In a typical form, the machine instructions may include rapid manufacturing machine instructions to manufacture the final implant by a rapid prototyping process (to obtain the appropriate porous structure) or by a combination of traditional manufacturing and rapid prototyping.

[0083] Gender / ethnicity specific hip implant creation With reference to Figures 25-76, an exemplary process and system for generating gender and / or ethnicity specific implants will be described. For purposes of the exemplary discussion, a total hip arthroplasty for a patient requiring an initial joint replacement will be described. However, it will be appreciated that the exemplary process and system may be applied to any orthopedic implant that is amenable to customization. For example, the exemplary process and system may be applied to total shoulder and knee replacements, as well as other initial joint replacements. As a result, although a hip implant is discussed below, one skilled in the art will appreciate the applicability of the system and process to other orthopedic implants, guides, tools, etc., for use with an initial orthopedic procedure or orthopedic repair procedure.

[0084] The hip joint is composed of the head of the femur and the acetabular fossa of the pelvis. The anatomy of the hip joint makes it one of the most stable joints in the body. Stability is provided by a rigid ball-and-socket configuration. The femoral head is approximately spherical at its articulation, forming two-thirds of a sphere. Data has shown that the diameter of the femoral head is smaller in women than in men. In a normal hip joint, it is believed that the center of the femoral head coincides exactly with the center of the acetabular fossa, and this assumption is used as the basis for the design of most hip joint systems. However, the natural acetabular fossa is not deep enough to cover all of the natural femoral head. The approximately circular portion of the femoral head is slightly flattened at the top, making it spheroidal rather than spherical. This spheroid distributes the load in a ring-like pattern around the upper pole.

[0085] The geometric center of the femoral head is crossed by the three axes of the joint: the horizontal, vertical, and anterior / posterior. The femoral head is supported by the neck of the femur, which joins the shaft. The axis of the femoral neck is set obliquely and extends superiorly, medially, and anteriorly. The angle of inclination of the femoral neck relative to the shaft in the frontal plane is the neck-shaft angle. In most adults, this angle varies between 90 and 135 degrees, and is important because it determines the effectiveness of the hip abductors, the length of the limb, and the forces imposed on the hip joint.

[0086] An inclination angle greater than 125 degrees is called coxa valga, while an inclination angle less than 125 degrees is called coxa vara. An inclination angle greater than 125 degrees corresponds to an extension of the limb, a decrease in the effectiveness of the hip abductors, an increase in the load on the femoral head, and an increase in the stress on the femoral neck. In coxa vara, an inclination angle less than 125 degrees corresponds to a contraction of the limb, an increase in the effectiveness of the hip abductors, a decrease in the load on the femoral head, and a decrease in the stress on the femoral neck. The femoral neck forms an acute angle with the transverse axis of the femoral condyle. This angle faces medially and anteriorly and is called the anteversion angle. In adults, this angle averages about 7.5 degrees.

[0087] The acetabulum lies on the lateral surface of the hip joint where the ilium, ischium, and pubis meet. These three separate bones combine to form the acetabulum, with the ilium and ischium each contributing approximately two-fifths of the fossa, and the pubis contributing one-fifth of the fossa. The acetabulum is a socket that is not deep enough to cover all of the femoral head and has both articular and non-articular portions. However, the acetabular labrum deepens the socket and increases stability. Together with the acetabular labrum, the acetabulum covers slightly more than 50% of the femoral head. Only the lateral side of the acetabulum is lined on its medial side with articular cartilage, which is interrupted inferiorly by a deep acetabular notch. The center of the acetabular cavity is deeper than the articular cartilage and is non-articular. This center is called the acetabular fossa and is separated from the joints of the pelvic bones by a thin plate. The acetabular fossa is a unique area for every patient and is used in creating a patient-specific guide for reaming and placement of the acetabular cup component, and variations in anatomical features further increase the need for population-specific implant designs.

[0088] Some of the problems associated with the prior art use of cementless components can be attributed to the wide variation in femoral canal size, shape, and orientation. One of the challenges to femoral stem orthopaedic implant design is the large variation in medial-lateral and anterior-posterior dimensions. There is also considerable variation in the ratio of proximal to distal canal size. The different combinations of various arcs, tapers, curves, and offsets in the normal population are staggering. However, that is not the only problem.

[0089] Ancestral differences in femoral morphology and the lack of clear standards for modern populations make the design of an appropriate hip implant system problematic. For example, significant differences in anterior curvature, twist, and cross-sectional shape exist between American Indians, American Blacks, and American Whites. Differences between Asian and Western populations in the femur are found in the anterior arch of the femur, where Chinese are more anteriorly bowed and externally rotated with a smaller intramedullary canal and smaller distal condyles than Whites. Similarly, White femora are larger than Japanese femora in terms of distal condyle length dimension. Ethnic differences also exist between American Blacks and Whites in proximal femoral bone mineral density (BMD) and hip shaft length. The combined effect of higher BMD, shorter shaft length, and shorter intertrochanteric width may account for a lower prevalence of osteoporotic fractures in American Black women compared to American White women. Similarly, older Asian and American black men were found to have thicker cortices and higher BMD than white Spanish men, which may contribute to greater bone strength in these ethnic groups. In general, American blacks have thicker bone cortices, narrower endosteal diameters, and greater BMD than American whites.

[0090] The combination of femoral ancestral differences and pelvic ancestral differences present an even greater challenge to the original hip joint system. Revision introduces greater complexity. In addition to these normal anatomical and ethnic variations, the difficulties faced by the revision surgeon are compounded by (a) distortion of the femoral canal caused by bone loss around the originally placed prosthesis, and (b) iatrogenic defects resulting from removal of components and cement.

[0091] All of the aforementioned factors have led many hip surgeons to seek ways to improve the design of cementless femoral prostheses. In total hip arthroplasty (primary or revision), it is ideal to provide an optimal fit between the femoral ball and the acetabular cup. The femoral stem neck should have a cruciform cross section to reduce stiffness. The stem length should be such that the stem makes parallel contact with the femoral wall over 2-3 internal canal diameters. The proximal third of the stem is porous or hydroxyapatite (HA) coated. The stem is cylindrical (i.e., not tapered) to control bending loads and allow all rotational and axial loads to be transferred to the proximal end. The femoral head position should replicate the patient's own head center unless abnormal.

[0092] One way to attempt to meet these goals is to individually manufacture femoral prostheses for each patient -- in other words, to create a prosthesis that is unique to that particular patient, rather than attempting to reshape the patient's bone to fit an off-the-shelf prosthesis.

[0093] There are several common design rules for patient-specific (or mass-customized) primary and revision hip replacements. These design rules include: (1) the hip stem must be collarless (except in repairs) to allow for uniform distribution of loads to the femur, (2) the hip stem must have a modified diamond cross-section to maximize fit / fill but maintain rotational stability, (3) the hip stem must be bowed, if necessary, to fit the patient's bone, (4) the hip stem must be inserted along a curved path without gaps between the prosthesis and the bone, and (5) the hip stem neck must be curved to reduce stiffness. (6) the hip stem length should be such that the stem is in parallel contact with the femoral wall over two to three inner canal diameters; (7) the proximal third of the hip stem is porous coated or hydroxyapatite (HA) coated; (8) the hip stem is cylindrical (i.e., not tapered) to control bending loads and allow the transfer of all rotational and axial loads to the proximal end; (9) the femoral head position of the hip stem should replicate the patient's own head center unless this is abnormal.

[0094] The following is an exemplary process and system for generating a mass-customized orthopedic implant for patients requiring initial joint replacement, taking into account the gender and / or ethnicity of the patient population. For the purposes of the exemplary discussion, a total hip arthroplasty for patients with incomplete anatomy will be described. However, it will be appreciated that the exemplary process and system can be applied to any orthopedic implant amenable to mass customization in cases where incomplete anatomy is present. For example, the exemplary process and system can be applied to total shoulder and knee replacements where bone degeneration (incomplete anatomy), bone deformity, or comminuted bone is present. As a result, although the femoral component of a hip joint implant is discussed below, one skilled in the art will appreciate the applicability of the system and process to other orthopedic implants, guides, tools, etc., for use with an initial orthopedic procedure or an orthopedic repair procedure.

[0095] Referring to FIG. 25, an overall process flow for using a statistical atlas to generate both mass-customized and patient-specific hip implants is depicted. Initially, the process includes a statistical atlas including several embodiment series in which one or more bones are analyzed. In a typical situation of a hip joint implant, the statistical atlas includes several instances of bone models for the pelvic bone and the femur. An articular surface geometric analysis is performed for at least the acetabular component (i.e., the acetabular fossa) and the proximal femoral component (i.e., the femoral head). In particular, the articular surface geometric analysis involves the calculation of landmarks, measurements, and shape features for each bone from a given population of the statistical atlas. The articular surface geometric analysis also includes generating quantitative values, such as statistics, that represent the calculations. From these calculations, a distribution of the calculations is plotted and the distribution of the calculations is analyzed based on the distribution. For example, in a ball-shaped distribution, it can be observed that about ninety percent (90%) of the population is grouped, so that a non-patient-specific implant (e.g., a mass-customized implant) can be designed and the non-patient-specific implant can be appropriately matched to this grouping, thus reducing the cost for the patient compared to a patient-specific implant. For the remaining ten percent (10%) of the population, a patient-specific implant may be a better approach.

[0096] In the context of mass-customized implants, statistical atlases may be utilized to quantitatively evaluate how many different groups (i.e., different implants) can encompass the vast majority of a given population. These quantitative evaluations may result in a body of data that indicates general parameters for a basic implant design that is not patient-specific, but is more specific than off-the-shelf alternatives.

[0097] In the context of patient-specific implants, the statistical atlas may be utilized to quantitatively assess what normal bone embodies and to quantitatively assess the differences between the patient's bone and normal bone. More specifically, the statistical atlas may include curvature data associated with an average or template bone model. This template bone model may then be used to estimate what the patient's correct bone morphology is to successfully create the implants and surgical instruments used to perform the implantation procedure.

[0098] Figure 26 graphically summarizes the use of a statistical atlas in designing a mass-customized patient-specific hip implant. In relation to the Implant box, reference is made back to Figures 17 and 18 and the associated discussion in these figures. Similarly, in relation to the Planner box, reference is made back to Figure 17 and the associated discussion of the Custom Planning Interface. Finally, in relation to the Patient-Specific Guide box, reference is made back to Figure 19 and the associated discussion in this figure.

[0099] As depicted in FIG. 27, a flow chart is depicted for an exemplary process that may be utilized to design and manufacture gender and / or ethnicity specific hip implants. In particular, the process includes utilization of a statistical atlas that encompasses various samples of proximal femora (i.e., femur including femoral head) that are deemed by the ethnicity of the person to whom the bone belongs and the associated data as being from either male or female. The statistical atlas module also records a virtual 3D model of one or more anatomical structures (e.g., bones) to obtain the specific anatomical variations in a given gender and / or ethnicity population. In a typical form, the atlas records a mathematical representation of the anatomical features of one or more anatomical structures expressed as the average representation and the change about the average representation in a given anatomical population that may have a common gender and / or ethnicity (or may be grouped to have one of multiple ethnicities where there is an anatomical commonality). See FIG. 2 and the previous discussion of statistical atlases and how to add anatomical structures to the statistical atlas of a given population. Output from the statistical atlas is directed to an automatic landmarking module and a surface / shape analysis module.

[0100] 27-39, the automated landmarking module utilizes input from the statistical atlas (e.g., regions likely to contain a particular landmark) and local geometric analysis to calculate anatomical landmarks for each instance of anatomy in the statistical atlas. Examples include (1) femoral head center, which is the center point of the femoral head approximated by a sphere; (2) greater trochanter point, which is the point on the greater trochanter that has the smallest distance to a plane passing through the neck diaphysis point perpendicular to the anatomical neck centerline; (3) osteotomy point, which is the point 15 millimeters from the end of the lesser trochanter (approximately 30 millimeters from the lesser trochanter point); (4) neck diaphysis point, which is the point on the head sphere whose tangent plane accommodates the smallest femoral neck cross-sectional area; and (5) the smallest diameter along the femoral diaphysis. Various proximal femoral landmarks are calculated for each 3D virtual model of the femur, including, but not limited to, (1) the femoral waist, which is the cross section having the smallest diameter along the intramedullary canal; (2) the femoral neck pivot point, which is a point on the femoral anatomical axis that forms an angle equal to the femoral neck angle with the femoral head center and the distal end of the femoral anatomical axis; and (3) the lesser trochanter point, which is a point on the lesser trochanter region that protrudes most outward. As a further example, various proximal femoral axes are calculated for each 3D virtual model of the femur using identified anatomical landmarks including, but not limited to, (a) a femoral neck anatomical axis coaxial with a line connecting the femoral head center and the femoral neck center, (b) a femoral neck axis coaxial with a line joining the femoral head center point and the femoral neck rotation point, and (c) a femoral anatomical axis coaxial with a line connecting two points located at distances of 23 percent and 40 percent of the total length of the femur starting from the proximal end of the femur.As yet another example, various proximal femoral measurements include (i) proximal angle, which is the 3D angle between the femoral anatomical axis and the femoral neck anatomical axis; (ii) femoral head offset, which is the horizontal distance between the femoral anatomical axis and the femoral head center; (iii) femoral head height, which is the vertical distance between the lesser trochanter point (already mentioned) and the femoral head center; (iv) greater trochanter-to-femoral head center distance, which is the distance between the femoral head center and the greater trochanter point (already mentioned); and (v) (vi) femoral head radius, which is the radius of a ball fitted to the femoral head; (vii) neck diameter, which is the diameter of a circle fitted to the neck cross section in a plane perpendicular to the femoral neck anatomical axis and passing through the neck center point (already mentioned); (viii) femoral neck anteversion transepicondylar angle, which is the angle between the transepicondylar axis and the femoral neck axis; and (ix) the angle between the posterior condylar axis and the femoral neck axis. (x) LPFA, which is the angle between the mechanical axis and the vector pointing towards the greater trochanter; (xi) femoral talar index area, which is defined by the equation (ZX) / Z, where Z is the femoral area 10 centimeters below the mid-lesser trochanter point and X is the intramedullary canal area 10 centimeters below the mid-lesser trochanter point; (xii) canal femoral talar area ratio, which is the ratio between the intramedullary canal area 3 centimeters below the mid-lesser trochanter height and the intramedullary canal area 10 centimeters below the mid-lesser trochanter height; (xiii) intramedullary canal area 10 centimeters below the mid-lesser trochanter height. (xiv) minor axis / major axis ratio, which is the ratio between the minor and major axes of an ellipse fitted to the intramedullary canal cross-sectional area at its narrowest point on the intramedullary canal; and (xv) femoral radius-intramedullary canal radius ratio, which is the ratio of the circle radii using a circle that best fits around the circumference of the femur and intramedullary canal in a plane perpendicular to the femoral anatomical axis (this ratio reflects the cortical bone thickness and therefore the cortical bone loss in the case of osteoporosis).

[0101] 27 and 40-42, using the output from the automatic landmarking module, parameters from the femoral stem are evaluated for a given population, specifically the medial contour, neck angle, and femoral head offset, whether the population is grouped based on ethnicity, gender, or a combination of the two.

[0102] For the medial contour, this contour for the intramedullary canal of each femur in the population is generated by intersecting the intramedullary canal with a plane that extends through the femoral pivot point and has a normal axis perpendicular to both the femoral anatomical axis and the neck axis (vector on the product). After the contour is generated for each femur in the population, the population is divided into groups using the intramedullary canal size. Once divided, the contours may be out of plane, so an alignment process is performed to align all the contours to a common plane (e.g., the XZ plane). The alignment process includes aligning an axis perpendicular to both the femoral neck axis and the anatomical axis with the Y axis, followed by aligning the anatomical axis with the Z axis. In this method, all the contours are translated relative to a specific point so that they have a common coordinate system.

[0103] After the contour has a common coordinate system, the femoral neck points are utilized to verify that the contour points are in a plane. In particular, the femoral neck points are consistent points that reflect the actual anatomy and ensure that the points on the contour are in a plane. By verifying that the contour points are in a plane, the alignment variability between femurs in a population can be significantly reduced, thereby facilitating the utilization of the contour for the design of femoral head offset and implant angle.

[0104] With reference to FIG. 43, the statistical atlas may also be useful for interpolating between normal and osteoporotic bones. When designing and dimensioning a femoral stem, one of the important considerations is the intramedullary canal dimension. In the case of normal bone, the intramedullary canal is significantly narrower with respect to the femur compared to femora with osteoporosis. This narrow intramedullary canal dimension is at least partially the result of a reduction in bone thickness (measured perpendicular to the femur's major axis), which correspondingly results in recession of the inner surface of the femur that defines the intramedullary canal. In this method, an artificial population is formed by generating a virtual 3D model having a thickness that is interpolated between that of healthy bone and that of severely osteoporotic bone. This data set thus includes bones corresponding to different stages of osteoporosis. This data set may now be used as an input to the implant stem design.

[0105] In a typical form, the statistical atlas includes populations of normal, non-osteoporotic and osteoporotic bones, where the bone is a femur. Each of these normal femurs in the atlas is quantified and represented as a 3D virtual model according to the process described herein for adding bones to a statistical atlas. Similarly, each of the osteoporotic bones in the atlas is quantified and represented as a 3D virtual model according to the process described herein for adding bones to a statistical atlas. As part of the 3D models for normal and osteoporotic bones, intramedullary canal dimensions are recorded along the longitudinal length of the femur. Using atlas point correspondences, the intramedullary canal is identified for the atlas bone as extending proximally a predetermined percentage (e.g., 5%) of the total bone length beyond the lesser trochanter and proximally a second predetermined percentage (e.g., 2%) beyond the distal cortical point. Additionally, points on the outer bone surface that fall within these proximal and distal boundaries are used to determine the bone thickness, which is defined as the distance from an external point to the closest point on the IM canal.

[0106] In the context of the proximal femur, Figures 46-57 confirm that gender differences exist across any ethnic population. As depicted in Figures 54 and 55, the statistical atlas template 3D model for the female proximal femur shows statistically significant measurements when compared to the template 3D model for the male proximal femur. In particular, the femoral head offset is approximately 9.3% smaller for females than for males. With current implants, the femoral head offset increases with stem size, which is acceptable for normal females. However, problems arise when considering femoral head offset in osteopenic and osteoporotic cases where bone loss leads to increased intramedullary canal size, which means larger stem size and larger offset. Similarly, the neck diameter and femoral head radius are approximately 11.2% smaller for females than for males. Also, the neck length is approximately 9.5% smaller for females than for males. Additionally, the proximal angle is approximately 0.2% smaller for females than for males. Finally, femoral head height is approximately 13.3% smaller in women than in men. As a result, the gender-specific bone data support the notion that simply scaling generic femoral implants (i.e., gender-neutral implants) does not account for differences in bone geometry, and therefore that gender-based femoral implants are needed.

[0107] 58-63, not only do the dimensions of the proximal femur vary widely across gender lines, but the cross-sectional shape of the femur also varies widely along the length of the intramedullary canal. In particular, across a given population in a statistical atlas of male and female femurs, males have intramedullary canal cross sections that are more circular than females. More specifically, females have intramedullary canal cross sections that are 8.98% more eccentric than males. As discussed in more detail below, this gender-specific data comprises a portion of the feature extraction data that is plotted to arrive at clusters from which numerical and general shape parameters are extracted to arrive at gender-specific femoral implants.

[0108] As depicted in Figures 64-66, the statistical atlas includes calculations corresponding to measurements across a given population of femurs (separated by gender) for femoral head center offset in the anterior-posterior (AP) direction. In a typical configuration, the AP direction was determined by a vector pointing anteriorly perpendicular to both the mechanical axis and the posterior condylar axis. The offset was measured between the femoral head center and two reference points, where the first reference point is the midpoint of the anatomical axis and the second reference point is the femoral neck pivot point. In summary, the AP femoral head height relative to the neck pivot point and the anatomical axis midpoint did not show significant differences between male and female femurs. Again, this gender-specific data comprises a portion of the feature extraction data that is plotted to arrive at clusters from which numerical and general shape parameters are extracted to arrive at gender-specific femoral implants.

[0109] 24 and 27, the femoral head center offset, intramedullary canal cross-sectional shape data, and medial contour data for the femurs in the statistical atlas population comprise part of the extracted feature data that is plotted to discern a predetermined number of clusters (a gender-specific cluster and an ethnicity-specific second cluster, assuming the statistical atlas includes data regarding ethnicity associated with each bone) present across the predetermined population to design a gender- and / or ethnicity-specific mass-customized implant consistent with the flowchart and related discussion regarding FIG 24. The gender- and / or ethnicity-specific identified clusters are utilized to extract the necessary parameters to design a mass-customized femoral implant.

[0110] Referring to FIG. 68, an exemplary mass-customized femoral component according to the present disclosure is depicted. In particular, the mass-customized femoral component comprises four major elements including a ball, a neck, a proximal stem, and a distal stem. Each of the major elements includes an interchangeable interface to allow the ball, neck, and stem to be interchangeable with other interchangeable elements. In this embodiment, if a larger femoral ball is required, only the femoral ball is replaced. Similarly, if a larger neck offset is desired, the neck element is replaced with a different neck element that provides the required offset, while leaving the other three elements intact, if appropriate. In this manner, the femoral component can be customized to fit the patient without any particular limitations and without necessarily sacrificing the fit or kinematics that would otherwise be compromised by using a one-size-fits-all implant. Thus, all of the femoral elements can be replaced with other mass-customized elements to better fit the patient anatomy.

[0111] In this exemplary embodiment, the neck is configured to rotate about the axis of the proximal stem, whereby the rotational orientation of the neck relative to the proximal stem can be adjusted intraoperatively. In particular, preoperative measurements may provide a planned rotational position of the neck relative to the proximal stem. Nevertheless, intraoperative considerations such as endokinetic testing may lead the surgeon to change the preoperative rotational orientation to improve the kinematics or avoid certain impingements. As an example, the neck includes a cylindrical stud having an inset circumferential groove with a textured surface. The cylindrical stud is received within an axial cylindrical channel of the proximal stem. In addition to the cylindrical channel, a second channel intersects the cylindrical channel, the second channel being formed to receive a plate having a semicircular groove that is also textured and configured to engage the textured surface of the inset circumferential groove. A pair of screws fastened to the proximal stem press the plate into engagement with the cylindrical stud, which eventually no longer allows rotational movement of the cylindrical stud relative to the proximal stem. Thus, when this fixed engagement is reached, the screw may be loosened to allow rotational movement between the cylindrical stud and the proximal stem, for example if rotational adjustment is required intraoperatively.

[0112] While the engagement between the neck and ball may be conventional, the engagement between the proximal and distal stems is not. In particular, the proximal stem includes a distal shank that is threadedly engaged to be threadedly received within a threaded opening extending into the distal stem. The proximal stem is then attached to the distal stem by rotating the proximal stem relative to the distal stem such that the threads of the shank mate with the threads of the opening in the distal stem. Rotation of the proximal stem relative to the distal stem is completed when the proximal stem abuts the distal stem. However, if rotational adjustment is required between the proximal and distal stems, a washer may be utilized to provide a spacer that corresponds to the precise rotational adjustment. As a further example, if a greater rotational adjustment is required, the washer is thicker, but a thinner washer provides a correspondingly smaller rotational adjustment.

[0113] Each of the primary elements may be manufactured with a predetermined alternative that accounts for size and contour variations within a given gender and / or ethnicity range. In this configuration, alternatives of the primary elements may be mixed and matched to approximate a patient-specific implant that conforms closer to the patient's anatomy than traditional mass-customized femoral components, but at a fraction of the cost and process to produce a patient-specific femoral implant.

[0114] 69 illustrates a further alternative exemplary mass-customized femoral component according to the present disclosure. In particular, the mass-customized femoral component comprises five major elements including a ball, a neck, a proximal stem, a mid-stem, and a distal stem. Each of the major elements includes an interchangeable interface to allow the ball, neck, and stem to be interchangeable with other interchangeable elements. One skilled in the art can appreciate that by increasing the number of elements of the mass-customized femoral component that are akin to stacked slices of the patient's natural femur to regenerate this bone, one can increasingly approximate a patient-specific implant fit by using mass-customized elements.

[0115] Similar to the anatomical differences between genders and ethnicities in the proximal femur, Figures 70-75 confirm that gender and ethnicity differences exist across the general pelvic population in the statistical atlas. Referring back to Figure 24, a series of mass-customized acetabular cup implants are designed and manufactured by using statistical atlas data (i.e., pelvic populations) that are grouped based on at least one of gender and ethnicity. The grouped atlas data is subjected to an automated landmarking process and a surface / shape analysis process to isolate the geometry of the acetabular cup within the population as depicted diagrammatically in Figure 70. Also depicted diagrammatically in Figures 74 and 75, the landmarking process (for locating the acetabular ligament) and the contour analysis process (for assessing the contour of the acetabular cup) result in feature extraction from which an anatomical cup implant surface is finally generated as depicted in Figure 71. This analysis shows that the acetabular cup and femoral head are not composed of a single radius of curvature but of several radii of curvature as depicted in Figures 72 and 73.

[0116] Creation of animal-specific implants Referring to FIG. 77, an exemplary system and method for designing and manufacturing animal-specific (i.e., patient-specific for an animal) implants and associated devices is similar to the process depicted and previously described with respect to FIG. 17, which is incorporated herein. As a prelude, an image of the animal anatomy is acquired and automatically segmented to yield a virtual 3D bone model. Although depicted diagrammatically as a CT scan image, it should be understood that other imaging modalities other than CT may be utilized, such as, but not limited to, MRI, ultrasound, and X-ray. In accordance with the previous exemplary disclosure, the virtual 3D bone model of the diseased anatomy is imported into a statistical atlas. Input from the statistical atlas is then utilized to reconstruct the bone to form the reconstructed virtual 3D bone model. Bone landmarks are calculated with respect to the surface of the reconstructed virtual 3D bone model to enable accurate implant size determination. The geometry of the diseased bone is then mapped and converted to a parametric form, which is then used to form an animal-specific implant that closely resembles the remaining anatomical geometry. In addition to the animal-specific implants, animal-specific instruments are manufactured and utilized for preparation of the animal's remaining bone and placement of the animal-specific implants.

[0117] Referring to FIG. 78, an exemplary system and method for designing and manufacturing mass-customized animal implants is similar to the process depicted and previously described with respect to FIG. 24, which is incorporated herein. As a prelude, 3D animal bone models from a statistical atlas associated with the bones of interest are subjected to automatic landmarking and surface / shape analysis. The automatic landmarking module utilizes information stored in the atlas (e.g., regions likely to contain specific landmarks) and local geometric analysis to automatically calculate anatomical landmarks for each 3D animal bone model. For each animal bone of interest in the statistical atlas, the shape / surface analysis directly characterizes the surface geometry of the 3D virtual animal bone model. Each of the 3D animal bone models is then subjected to a feature extraction process that uses a combination of landmarks and shape features to calculate features relevant to the implant design. These features are used as input to a clustering process, in which the animal bone population is divided into groups with similar features using a predefined clustering methodology. Each resulting cluster represents an instance that is used to define the shape and size of a single animal implant. A parameterization process is then performed for each cluster center (implant size) to extract parameters (e.g., computer-aided design (CAD) parameters) for the entire implant model. Using the extracted parameters, an entire implant surface and size is then generated for each cluster. Depending on the cluster into which the animal patient is classified, a mass-customized implant is selected from the required group and implanted.

[0118] Creation of patient-specific cutting guides 79-94, an exemplary process and system for integration of multi-dimensional medical imaging, computer-aided design (CAD), and computer graphic features to design a patient-specific cutting guide will be described. For purposes of exemplary illustration only, the patient-specific cutting guide will be described in the context of a total hip arthroplasty procedure. Nonetheless, one skilled in the art will recognize that the exemplary process and system are applicable to any surgical procedure in which a cutting guide may be utilized.

[0119] As depicted in FIG. 79, an overview of a typical system flow begins with receiving input data representing an anatomy. The input anatomical data comprises two-dimensional (2D) images or three-dimensional (3D) surface representations of the target anatomy, which may be in the form of, for example, a surface model or a point cloud. In situations where 2D images are utilized, these 2D images are utilized to construct a 3D surface representation of the target anatomy. Those skilled in the art are familiar with utilizing 2D images of anatomy to construct a 3D surface representation. Thus, a detailed description of this process has been omitted to promote brevity. By way of example, the input anatomical data may comprise one or more of X-ray images (taken from at least two views), computed tomography (CT) scan images, magnetic resonance images (MRI), or any other imaging data from which a 3D surface representation can be generated. In a typical form, the anatomy comprises a pelvis and a femur.

[0120] However, it should be understood that the above is an exemplary description of biological structures that may be used with the exemplary system, and as such is in no way intended to limit other biological structures from being used with the present system. As used herein, tissue includes bone, muscle, ligament, tendon, and any other distinct type of structural material with a specific function in a multicellular organism. As a result, when the exemplary systems and methods are discussed in the context of bones involved in the hip joint, one skilled in the art will recognize the applicability of the systems and methods to other tissues.

[0121] The input anatomical data of the femur and pelvis of the system is directed to one of two modules depending on the type of input data: in the case of X-ray data, 2D X-ray images are input to a non-rigid module to extract the 3D bone contours; if the input data is in the form of CT scans or MRI images, these scans / images are directed to an automatic segmentation module where the scans / images are automatically segmented to extract the 3D bone contours (and 3D cartilage contours).

[0122] With reference to FIG. 80, the non-rigid module uses multiple X-ray images taken from at least two fields of view that are subjected to one or more pre-processing steps. These steps may include one or more of the following: noise reduction and image processing. The resulting pre-processed X-ray images are subjected to a calibration step to align the X-ray images. Preferably, the X-ray images are taken in the presence of a fixed-position calibration device, whereby the X-ray images are aligned relative to this fixed-position calibration device. However, if a fixed-position calibration device is not present in the X-ray images, the images may nevertheless be calibrated using common features detected across multiple images. The output from this calibration process is the position of the anatomy relative to the imaging device, which is identified in FIG. 80 by the reference "Pose".

[0123] The resulting pre-processed X-ray image is subjected to a feature extraction step, which comprises one or more calculations of image features utilizing the pre-processed X-ray image. By way of example, these calculations may include gradient features, contours, texture components, or any other features derived from the image. In this exemplary process, the feature extraction step outputs the contours of the anatomical structures (e.g. bone shapes), represented by the reference character "Contour" in FIG. 80, and image features, represented by the reference character "Texture", derived from the X-ray image. Both the contoured anatomical structures and image feature data are directed to a non-rigid registration step.

[0124] The non-rigid registration step aligns the output from the feature extraction step and the calibration step to a 3D template model of the subject's anatomy from a statistical atlas. As an example, the 3D template model is generated according to non-linear principal components from an anatomical database that comprises part of the statistical atlas. During the non-rigid registration step, the 3D template model is optimized so that its shape parameters (non-linear principal components) match those of the X-ray image, which is driven by the pose, contour, and texture data. The output from the non-rigid registration step is a 3D patient-specific bone model that is directed to a virtual templating module, similar to the 3D patient-specific bone model output from an automatic segmentation module for CT scan or MRI images.

[0125] With reference to FIG. 83, the automatic segmentation process is initialized, for example, by taking a CT scan or MRI image and running an automatic segmentation sequence. With particular reference to FIG. 82, the automatic segmentation sequence includes aligning the scan / image to a base or starting 3D model of the target anatomy. After aligning the scan / image to the base 3D model, the scan / image is processed by a primary deformation process to calculate normal vectors to determine the location of the profile points, linearly interpolate intensity values, filter the resulting profile using a Savitsky-Golay filter, generate the profile gradient, weight the profile using a Gaussian weighted profile equation, determine maximum profiles, and use these maximum profiles to deform the base 3D model. The resulting deformed 3D model is projected onto a template 3D model from a statistical atlas for the target anatomy. Using the parameters of the template 3D model, the deformed 3D model is further deformed in a secondary deformation process to resemble the features inherent in the template 3D model. After this latter deformation process, the deformed 3D model is compared to the scan / image to tell if significant differences exist.

[0126] In situations where significant differences exist between the deformed 3D model and the scan / image, the deformed 3D model and the scan / image are again subjected to a primary deformation process and then a secondary deformation process, and this looping process continues until the deformed 3D model is within a predetermined tolerance for differences between the deformed 3D model and the scan / image.

[0127] After the deformed 3D model is determined to exhibit less than a significant difference from the previous iteration or a maximum number of iterations has been reached, the surface edges of the deformed 3D model are smoothed, followed by a high-resolution remeshing step to further smooth the surface to form a smoothed 3D model. This smoothed 3D model is subjected to a primary deformation sequence (identical to the primary deformation process described above before surface smoothing) to generate a 3D segmented bone model.

[0128] Referring back to Figure 83, the 3D segmented bone model is processed to generate contours. In particular, the intersection of the 3D segmented bone model and the scans / images is calculated, resulting in a binary contour for each image / scan.

[0129] The 3D segmented bone models are also processed to generate a patient-specific statistical 3D model of bone appearance. In particular, the appearance of the bone and any anatomical abnormalities are modeled based on the image information present within and outside the contour.

[0130] The bone contours are then reviewed by a user of the segmentation system. This user may be a segmentation expert or an infrequent user of the segmentation system who notes one or more areas of the 3D model that do not correlate with the segmented regions. This lack of correlation may be in the context of missing or obviously inaccurate regions. Upon identification of one or more erroneous regions, the user may select a "seed point" on the model that indicates the center of the area where the erroneous region is located, or may manually contour the missing region. The system's software uses the seed point to add or subtract local contours to the seed point using an initial scan / image of the anatomy from a CT or MRI. For example, the user may select an area where a bone spur should be present, and the software will compare the scan / image to the area on the 3D model and add the bone spur to the segmentation sequence. Any changes made to the 3D model are finally reviewed and verified or undone by the user. This review and correction sequence may be repeated as many times as necessary to account for anatomical differences between the scan / image and the 3D model. Once the user is satisfied with the 3D model, the resulting model may be manually manipulated to remove touch-up areas and bridges in the model as needed before being output to the virtual templating module.

[0131] As shown in Figures 79 and 84, the virtual templating module receives a 3D patient-specific model from either or both of the automatic segmentation and non-rigid registration modules. In relation to the hip joint, the 3D patient-specific model includes the pelvis and femur, both of which are input into an automatic landmarking process. This automatic landmarking step uses regions from similar anatomy present in a statistical atlas and local geometric searches to calculate anatomical landmarks relevant for implant placement on the femur 3D model and on the pelvis 3D model.

[0132] In the context of automatic placement of a femoral stem using distal fixation as shown in FIG. 85, automatic landmarking involves the definition of axes at the femur and the implant. For the femur, the anatomical femoral axis (AFA) is calculated, followed by the proximal anatomical axis (PAA). Then, the proximal neck angle (PNA) is calculated, which is defined as the angle between the AFA and the PNA. For the femoral implant, the implant axis is along the length of the implant stem, and the implant neck axis is along the length of the implant neck. Similar to the femoral PNA, the implant angle is defined as the angle between the implant axis and the implant neck axis. The implant with the implant angle closest to the PNA is then selected. Then, the implant fit angle (IF A) is defined as the intersection of the proximal anatomical axis with a vector drawn from the femoral head center at the selected implant angle.

[0133] When using automatic placement of a femoral stem using distal fixation and calculated anatomical landmarks as shown in Figure 85, the implant sizing step determines / estimates the appropriate implant size for the femoral component. The implant size is chosen by comparing the width of the implant to the width of the intramedullary canal and selecting the implant with the width that most closely resembles the intramedullary canal. The system then proceeds to the implant placement step.

[0134] In the implant placement step for distally fixed femoral stems, an initial implant position is determined / selected with respect to all relevant implanted components based on the surgeon's preferred surgical technique and the previously calculated anatomical landmarks. A resection plane is created to simulate a proximal femoral osteotomy and the implant fit is evaluated by analyzing the cross-sections of the aligned implant and femoral intramedullary canal at various positions along the implant axis. The implant is aligned to the femur by translating the implant so that the neck of the implant is in the general position of the proximal femoral neck after aligning the implant axis to the anatomical femoral axis. The implant is then rotated around the anatomical femoral axis to obtain the desired anteversion.

[0135] As part of this implant placement step, an iterative scheme is utilized that includes using an initial "educated guess" regarding implant placement as part of a kinematic simulation to evaluate the placement of the "educated guess." In a typical form, the kinematic simulation takes the implant (based on the placement of the selected implant) through a predetermined range of motion using estimated or measured joint kinematics. The kinematic simulation may then be used to determine collision locations and estimate the resulting range of motion of the implant after implantation. If the kinematic simulation produces unsatisfactory data (e.g., unsatisfactory range of motion, unsatisfactory mimicry of natural kinematics, etc.), other locations for implant placement may be utilized and then kinematic analysis may be performed to further refine the implant placement until a satisfactory result is reached. After implant locations have been determined / selected for all relevant implanted components, the template data is transferred to a tool generation module.

[0136] In the context of automatic placement of a femoral stem using a press fit and three contacts as shown in FIG. 86, automatic landmarking involves the definition of axes at the femur and the implant. For the femur, the anatomical femoral axis (AFA) is calculated, followed by the proximal anatomical axis (PAA). Then, the proximal neck angle (PNA) is calculated, which is defined as the angle between the AFA and the PNA. For the femoral implant, the implant axis is along the length of the implant stem, and the implant neck axis is along the length of the implant neck. Similar to the femoral PNA, the implant angle is defined as the angle between the implant axis and the implant neck axis. The implant with the implant angle closest to the PNA is then selected. Then, the implant fit angle (IF A) is defined as the intersection of the proximal anatomical axis with a vector drawn from the femoral head center at the selected implant angle.

[0137] When using automatic placement of the femoral stem using a press fit, three contacts, and calculated anatomical landmarks as shown in FIG. 86, the implant sizing step determines / estimates the appropriate implant size for the pelvic and femoral components. The implant size is selected by aligning the implant to the femur by aligning the implant axis to the anatomical femoral axis. The implant is then rotated to align its neck axis with the femoral neck axis. The implant is then translated so that it is in the anatomically appropriate position within the proximal femur. The system then proceeds to the implant placement step.

[0138] In the implant placement step for the press-fit femoral stem, an initial implant position is determined / selected with respect to all relevant implanted components based on the surgeon's preferred surgical technique and the already calculated anatomical landmarks. A resection plane is created to simulate a proximal femoral osteotomy and the implant fit is evaluated. The fit evaluation is performed by analyzing the contour of the implant and the femoral intramedullary canal. The contour is created by intersecting the intramedullary canal with a plane perpendicular to both the anatomical axis and the femoral neck axis and passing through the intersection of the anatomical axis and the femoral neck axis to yield a contour. Once the contour of the implant and the intramedullary canal is generated, only implants with a width smaller than the intramedullary canal width at the same position are kept, thereby resulting in a large number of possible correct implant sizes. The group of possible sizes is reduced by two strategies to reduce the mean square distance error between the implant and the intramedullary canal. The first strategy minimizes the mean square error (MSE) or other mathematical error measure of the distance between the implant and both the medial and lateral sides of the intramedullary canal. The second strategy minimizes the MSE of the distance between the sides of the implant and the intramedullary canal.

[0139] As part of this implant placement step, an iterative scheme is utilized that includes using an initial "educated guess" regarding implant placement as part of a kinematic simulation to evaluate the placement of the "educated guess." In a typical form, the kinematic simulation takes the implant (based on the placement of the selected implant) through a predetermined range of motion using estimated or measured joint kinematics. The kinematic simulation may then be used to determine collision locations and estimate the resulting range of motion of the implant after implantation. If the kinematic simulation produces unsatisfactory data (e.g., unsatisfactory range of motion, unsatisfactory mimicry of natural kinematics, etc.), other locations for implant placement may be utilized and then kinematic analysis may be performed to further refine the implant placement until a satisfactory result is reached. After implant locations have been determined / selected for all relevant implanted components, the template data is transferred to a tool generation module.

[0140] Referring back to FIG. 79, the jig generation module generates a patient-specific guide model. More specifically, from the template data and associated planning parameters, the shape and placement of the patient-specific implant is known with respect to the patient's remaining bone. As a result, the virtual templating module uses the patient-specific 3D bone model to calculate the position of the implant relative to the patient's remaining bone, thus providing the jig generation module with information regarding how much of the patient's remaining bone is to be preserved. According to this bone retention data, the jig generation module utilizes the bone retention data to allocate one or more bone cuts to reduce the patient's current bone to the remaining bone required to accept the implant as planned. Using the intended bone cuts, the jig generation module generates a virtual 3D model of a cutting guide / jig having a shape configured to mate with the patient's bone in a single position and orientation. In other words, the 3D model of the cutting jig is formed as a "negative" of the anatomical surface of the patient's remaining bone such that the tangible cutting guide precisely matches the patient's anatomy. In this manner, any guesswork associated with positioning the cutting jig is eliminated. After the jig generation module generates a virtual 3D model of the cutting jig, the module outputs the necessary machine code for a rapid prototyping machine, CNC machine, or similar device to manufacture a tangible cutting guide. As an example, a typical cutting jig for femoral head and neck resection includes a hollow slot that forms an associated guide to constrain the cutting blade within a specific range of motion and maintains the cutting blade in a predetermined orientation that replicates the virtual cut from the surgical planning and templating module. The jig generation module is also utilized to create a placement jig for a femoral stem.

[0141] Referring to FIG. 92, after removing the femoral head and neck, intramedullary reaming is performed, followed by femoral stem insertion. To prepare the femur for the insertion of the femoral implant, the intramedullary canal needs to be reamed along a direction that matches the direction of the femoral implant. If the reaming is offset, the orientation of the femoral implant may be compromised. To address this concern, the jig generation module generates a virtual guide that is a "negative" of the anatomical surface of the patient's remaining or resected bone so that a rapid prototyping machine, CNC machine, or similar device can manufacture a cutting guide that precisely matches the patient anatomy. As an example, the reaming jig may include an axial guide along which the reamer may traverse in a longitudinal direction. Using this reaming jig, the surgeon performing the reaming operation is assured of the proper reaming direction.

[0142] The intramedullary canal may receive a femoral stem. Again, the jig generation module generates a femoral stem placement guide to ensure that the femoral stem is properly positioned within the intramedullary canal both from a rotational and angular standpoint. As an example, the femoral stem placement guide is simultaneously a "negative" of the anatomical surface of the patient's remaining or resected bone and the top end of the femoral stem. In this manner, the placement guide includes a unique shape for sliding over the femoral shaft (the portion of the femoral stem to which the femoral ball is connected) while simultaneously connecting with the patient's remaining or resected bone, such that only a single orientation of the femoral stem is possible relative to the patient's femur, thereby ensuring proper implantation of the femoral stem consistent with the preoperative plan. However, it should be noted that although the exemplary jig has been described in the context of an initial hip implant, one skilled in the art should understand that the exemplary process and system described above is not limited to initial hip implants or to hip joint or repair surgical procedures. Alternatively, the process and system may be applied to any hip implant, as well as surgical procedures involving other areas of the body, including, but not limited to, the knee, ankle, shoulder, spine, head, and elbow.

[0143] In relation to the acetabular cavity, the tool generation module may generate instructions for manufacturing a reaming and acetabular implant placement guide for the acetabular cup, as depicted in Fig. 93. In particular, from the template data and associated planning parameters, the shape and placement of the patient-specific acetabular implant is known with respect to the patient's remaining pelvis. As a result, the virtual templating module uses the patient-specific 3D acetabular cavity model to calculate the size and position of the acetabular cup implant relative to the patient's remaining bone, thus providing the jig generation module with information on how much of the patient's remaining pelvis is to be preserved and the desired implant orientation. According to this bone retention data, the jig generation module utilizes the bone retention data to allocate one or more bone cuts / reams to reduce the patient's current pelvis to the remaining bone required to accept the acetabular implant as planned. Using the intended bone cuts, the jig generation module generates a virtual 3D model of a cutting guide / jig having a shape configured to mate with two parts of the patient's pelvis in only one direction. In other words, the 3D model of the cutting jig is formed as a "negative" of the anatomical surface of the patient's pelvis such that the tangible reaming guide precisely matches the patient's anatomy. In this manner, any guesswork associated with the positioning of the reaming jig is eliminated. After the tool generator module generates a virtual 3D model of the reaming tool, the module outputs the necessary machine code for a rapid prototyping machine, CNC machine, or similar device to manufacture a tangible reaming tool. As an example, a typical acetabular component tool for reaming the acetabulum may have a four-piece structure, where a first piece is configured to be received in the natural acetabulum and temporarily attach to a second piece until the second piece is secured to the pelvis using the first piece as a placement guide. After the second piece is fastened to the pelvis, the first piece may be removed. Additionally, the third piece includes a cylindrical or partially cylindrical component that uniquely connects with the second piece to allow the reamer to transverse longitudinally to the third piece, but whose orientation is fixed using the combination of the first and third pieces. After reaming, the reamer is removed and the third piece is detached from the first piece. The acetabular cup implant is seated into the reamed acetabular cavity using the fourth piece.In particular, the fourth piece is uniquely formed to engage with the first piece in only a single orientation, but at the same time, be received within the interior of the acetabular cup implant. After the implant cup is positioned, both the first and fourth pieces are removed. It should also be noted that additional jigs may be formed to drill one or more holes in the pelvis for seating the acetabular implant, where each of the drilling jigs is sequentially mounted against the first piece to verify the orientation of the drill bit.

[0144] Trauma plate formation With reference to Figures 95-108, an exemplary process and system for forming bone plates (i.e., trauma plates) across a given population will be described. Those skilled in the art are aware that bone can undergo regeneration to repair itself after fracture. Depending on the severity and location of the fracture, prior art trauma plates have been utilized that often require bending or other modification in the operating room to conform to the irregular bone shape and obtain maximum contact between the bone fragments. However, excessive bending reduces the useful life of the trauma plate, which can lead to bone plate breakage and / or trauma plate-screw fixation loosening. This process and system provides a more precise trauma plate shape, reducing or eliminating the need to contour the plate intraoperatively, thereby increasing plate useful life and increasing the time before any bone plate-screw fixation loosening occurs.

[0145] The foregoing exemplary description for forming a trauma plate applies to any and all bones to which a trauma plate can be applied. For simplicity, the exemplary description describes a system and process for forming a trauma plate for use with a humerus. However, it should be understood that the process and system are equally applicable to the manufacture of other bones of the body and corresponding trauma plates and are in no way limited to humerus trauma plates.

[0146] As part of the exemplary process and system for forming a trauma plate, a statistical bone atlas is formed and / or utilized for the bone of the subject. By way of illustration, the bone of the subject comprises a humerus. Those skilled in the art are familiar with statistical atlases and methods of constructing a statistical atlas in relation to one or more bones. As a result, a detailed discussion of constructing a statistical bone atlas has been omitted to promote brevity. Nevertheless, what may be unique about the statistical bone atlas of the exemplary system and process is the classification of the humerus within the statistical bone atlas based on gender, age, ethnicity, deformity, and / or incomplete configuration. In this manner, one or more trauma plates may be mass-customized for one or more of the aforementioned categories, where the one or more categories define a particular bone population.

[0147] In a typical form, a statistical bone atlas includes anatomical data that may take a variety of forms. As an example, a statistical bone atlas may include two-dimensional or three-dimensional images and information about bone parameters from which measurements can be obtained. Typical atlas input data may take the form of x-ray images, CT scan images, MRI images, laser scan images, ultrasound images, segmented bones, physical measurement data, and any other information from which a bone model can be formed. This input data is utilized by software that accesses the statistical atlas data to construct a three-dimensional bone model (or access a three-dimensional bone model that has already been formed and stored as part of the statistical atlas), from which the software creates an average or template bone model in three dimensions.

[0148] Using the template bone model, the software can automatically specify points on the outer surface of the template bone model or allow manual specification of points on the outer surface of the template bone model. By way of illustration, in relation to the average humerus model, a user of the software defines a general boundary shape for the final trauma plate by roughly contouring the shape of the trauma plate on the outer surface of the humerus model. The general boundary shape of the trauma plate can also be achieved by the user specifying a series of points on the outer surface of the humerus model that correspond to the outer boundary. Once the outer boundary or boundary points are defined, the software may automatically specify points on the outer surface of the humerus model within the defined boundary or allow manual specification of said points. As an example, the software may perform a percentage fill operation during which the user may specify that percentage within the boundary of the trauma plate to be specified by a series of points, each corresponding to a distinct location on the outer surface of the humerus model. The software also provides a manual point specification function, in which the user may specify one or more points on the outer surface of the humerus model within the boundary. Note that when manual point specification is utilized, the user does not need to predefine the boundary to specify points on the outer surface of the humerus model, rather, when the manual specification of points is completed, the boundary is defined by the outermost points specified.

[0149] After the designation of points on the outer surface of the template bone model, the localization points are propagated throughout the bone population of interest. In particular, the localization points are automatically applied by the software to each three-dimensional bone model in a given population via point correspondence in a statistical atlas. As an example, a given bone population may be gender-specific and ethnicity-specific to comprise humeri from Caucasian females. Using the propagated points for each bone model in the population, the software fills gaps between points within the boundary using a three-dimensional filling process to form a three-dimensional rendering of the trauma plate for each bone. The software then calculates the longitudinal midline of the three-dimensional rendering of each trauma plate via a thinning process.

[0150] The midline of each three-dimensional trauma plate rendering comprises a three-dimensional midline with varying curvatures along its length. The software extracts the three-dimensional midline and uses a least squares fit to determine a preferred number of radii of curvature that together best approximate the dominant curvature of the three-dimensional midline. In the context of the humerus, it has been found that three radii of curvature accurately approximate the midline curvature. However, this number may vary depending on the bone population and the boundaries of the trauma plate. Additional features may also be included herein, such as cross-sectional curvature at one or more locations along the length of the plate, locations of muscles, nerves, and other soft tissues to be avoided, or any other features relevant to defining the size or shape of the plate. As an example, the three radii of curvature for the midline represent the curvature of the trauma plate at the proximal humerus, the transition between the humeral shaft and the humeral head, and the curvature of the humeral shaft. Each radius of curvature was recorded, and a four-dimensional feature vector was applied to the radius of curvature data to cluster the radii into groups that best fit the population. In a typical form, the cluster data may indicate that multiple trauma plates are required to adequately fit the population. Once the radius of curvature data has been clustered, the trauma plate dimensions may be finalized.

[0151] During feature extraction associated with plate design, the software determines the best number of clusters to fit the population. It must be noted that there are some cases where there are more than one cluster that gives a local minimum as shown diagrammatically in FIG. 100. To determine the acceptable tolerance and the optimal selection that gives a reasonable number of plates in each group, the software generates three-dimensional surface models for the plates in each cluster. An automatic evaluation is then performed by placing the plates on the population and calculating the discrepancy between the plates and the bone surface. With the results of this analysis, the software chooses the optimal number of plates to be used for this particular population. The final plate model is then parameterized and screw positions are placed on each plate to maximize fixation while avoiding muscle and soft tissue positions. The screw width is determined by cross-sectional analysis of the bone at each screw position throughout the population.

[0152] The process and method were validated for humeri using a cadaveric study. In particular, CT scans of cadaveric humeri from Caucasian females were taken. These CT scans were utilized by the software to generate separate 3D models for each humerus. It is noted that none of the CT scans and 3D models utilized during this validation study were part of the statistical atlas and associated population utilized to generate the humerus trauma plates. As a result, none of the CT scans and 3D models provided new data and models to be used to validate the designed humerus trauma plates. After the 3D validation models were generated, each of the models was classified into a particular cluster (a cluster resulting from designing a humerus trauma plate from the design population). Based on which cluster the validation model was classified into, the humerus trauma plate designed for that cluster was fitted to the appropriate validation 3D humerus model, and a measurement was calculated indicating any spacing between the outer surface of the validation 3D humerus model and the underside of the humerus trauma plate. FIG. 107 depicts a distance map of a trauma plate mounted against a validation three-dimensional humerus model to show the area of ​​maximum distance between the bone and the trauma plate. It can be seen that the majority of the trauma plate is minimally spaced from the bone while only the poorly fitted areas show a spacing ranging from 0.06 to 0.09 centimeters. As a result, it was determined at the end of this cadaveric study that a trauma plate designed using the aforementioned system and following the aforementioned exemplary process has a remarkable contour fit that obviates the surgeon's practice of having to bend or manually reshape the bone plate when applied intraoperatively.

[0153] In another typical example of this process, a trauma plate was created for the clavicle. Here, a statistical atlas was created from several clavicles that adequately captured the variation within the Caucasian population. The locations relative to muscle attachment sites were also defined within the statistical atlas. Cross-sectional contours were extracted at 5% increments along the entire bone, and at the muscle attachment sites and at the clavicle waist. The maximum and minimum dimensions of each cross-sectional contour were calculated. The entire three-dimensional surface was also examined for asymmetry by analyzing the differences in magnitude and direction between similar points across the entire bone surface in the dataset. The results confirm existing studies of clavicle asymmetry, namely, that the left clavicle is longer than the right clavicle, but the right clavicle is thicker than the left clavicle. However, the pattern of asymmetry differs between males and females. Also, the clavicle midline does not follow a symmetrical "S" shape as in existing plate designs. Males are significantly more asymmetric (p<0.05) in all dimensions and in muscle and ligament attachment site contours, while female asymmetry is more variable. We hypothesize that this must address absolute and relative differences in muscle strength in men compared to women. However, the area without muscle attachments on the posterior mid-shaft was significantly asymmetric in both genders. From the extracted features, clustering was performed to find a set of clavicle plates to best fit the population. Also, screw fixation locations and lengths can be determined to best avoid soft tissue (muscle attachments) and prevent further fracture or plate loosening as a result of screws that are too long or too short. Several plates were designed using the process, as seen in Figures 110, 111, 112, 113, and 114.

[0154] Preparation of trauma plate placement guide With reference to FIG. 115, an exemplary process and system for forming a patient-specific trauma plate placement guide will be described. Those skilled in the art recognize that bones can be fractured in one or more locations, resulting in bone fragments that are separated from one another. As part of a reconstructive procedure to repair the bone, these fragments are held in a fixed orientation using one or more trauma plates. Reconstructive surgeons have attempted to connect the bone backs to one another using innate knowledge rather than patient-specific anatomical facts. As a result, to the extent that the patient's bone anatomy is altered from normal, to the extent that the bone fragments are significantly distorted, or to the extent that the number of bone fragments is large, surgeons resort to using prior art trauma plates and adapting the bone fragments to the shape of the plate rather than adapting the shape of the plate to the bone fragments. This process and system improves upon prior art trauma plate application by forming a trauma plate placement guide and customized trauma plate that adapts the trauma plate to the bone to replicate the original bone shape and orientation.

[0155] A typical system flow begins with receiving input data representing a disrupted anatomy. For purposes of illustration only, the disrupted anatomy comprises a human skull. It should be noted that the above process and system are equally applicable to other anatomy / bones, including but not limited to arms, legs, and torso. In a typical form, the anatomical data input may be in the form of x-rays, CT scans, MRIs, or any other imaging data that can represent the size and shape of bones.

[0156] The input anatomical data is utilized to construct a three-dimensional virtual model of the fractured anatomy. As an example, the input anatomical data comprises a computed tomography scan of the fractured skull, which is processed by the software to segment the scan and generate a three-dimensional model. Those skilled in the art are familiar with how to utilize a computed tomography scan to construct a three-dimensional virtual model. As a result, a detailed description of this aspect of the process has been omitted to promote brevity.

[0157] After generating the three-dimensional virtual model of the fractured skull remains, the software compares the three-dimensional virtual model of the skull with data from the statistical atlas to determine the areas in the three-dimensional virtual model where the skull is fractured. In particular, the software uses features extracted from the surface model of the input anatomy (e.g., surface roughness, curvature, shape factor, curvature, adjacent connectivity) to extract the areas of the fracture sites. Then, the outline contours of the fracture sites are extracted and matched to each other to find the matching fracture sites. The fractured fragments are also matched to the atlas to indicate the best position for placing the matched fracture sites to reconstruct the normal anatomy.

[0158] After the software generates a reconstructed three-dimensional virtual model of the fractured skull, buttresses may be manually and / or automatically placed on the exterior surface of the reconstructed three-dimensional virtual skull model. The automatic placement of the buttresses is the result of programmed logic to maximize the stability of the bone fragments while minimizing the number of buttresses. As used herein, the term buttress and its plurals refer to any support used to stabilize the bone fragments relative to one another. In certain cases, practical experience by a surgeon or other learned user supplements or replaces the logic when using the manual buttress placement feature. In any case, a series of buttresses are programmed into the software, and the software allows the software or a user of the software to select different buttresses for different applications. At the same time, the length of the buttresses may be manually or automatically manipulated based on the dimensions of the fracture and bone fragments.

[0159] After buttress allocation and buttress placement in the reconstructed three-dimensional virtual skull model, the software dimensions and contours of each buttress are recorded by the software. This record includes information necessary for the manufacture of each buttress, or, at a minimum, information useful to enable a surgeon or other knowledgeable individual to take existing buttresses and fit each to the placement guide. In conjunction with shaping the existing buttresses, the software extracts the contours of the reconstructed three-dimensional virtual skull model to generate computer-aided design (CAD) instructions for the formation of one or more physical models representative of the reconstructed three-dimensional skull model. These CAD instructions are sent to a rapid prototyping machine that forms one or more physical models representative of the reconstructed three-dimensional skull model. Each buttress may be applied to the physical model at the target location by recreating the appropriate anatomical surfaces as the physical model, or each buttress may be manually fitted prior to implantation and fastening to the patient's skull.

[0160] Based on the location and length of any buttresses, the software also extracts contours of the reconstructed three-dimensional virtual skull model to generate contour data for one or more patient-specific buttress placement guides. In particular, a placement guide may be generated for each buttress. In this manner, the placement guide includes a surface contour that matches the contour of the patient's skull in a single direction. Given that the location of the buttresses, as well as the contours of the adjacent outer skull surfaces, are known in the reconstructed virtual skull model, the software combines the two to form a virtual patient-specific placement guide. This virtual guide is output in the form of CAD instructions to a rapid prototyping machine for manufacturing.

[0161] In this exemplary embodiment, the manufactured patient-specific placement guide includes an elongated handle configured to be grasped by a surgeon. Extending from the end of the elongated handle is a block C-shaped contour plate. The underside of the contour plate is concave to match the convex morphology of the skull where the buttress is to be placed. Although not required, the end (or other portion) of the contour plate may be fastened to the buttress, or the contour plate may simply include a working window into which the buttress is aligned and ultimately fastened to the skull. After attachment of the buttress to the skull, the contour plate may be removed.

[0162] Customized cutting and placement guides and plates Referring to Fig. 116, reconstruction of deformed, destroyed, or incomplete anatomy is one of the complex problems facing health care providers. Abnormal anatomy may be the result of birth conditions, tumors, disease, or physical injuries. As part of providing treatment for various diseases, health care providers may find it advantageous to reconstruct or configure anatomy to facilitate treatment for various conditions, which may include, but are not limited to, fracture / bone comminution, bone degeneration, orthopedic implant repair, orthopedic primary implantation, and disease.

[0163] The present disclosure provides a system and method for bone and tissue reconstruction using bone grafts. To perform this reconstruction, the system and associated method utilizes current anatomy images of the patient to construct two virtual 3D models: (a) a first 3D model representing the current abnormal anatomy, and (2) a second 3D model representing the reconstructed anatomy of the patient. For a detailed description of using patient images (e.g., x-rays, CT scans, MRI images, etc.) to arrive at virtual models of the patient's abnormal anatomy and reconstructed anatomy, reference is made to the "Complete Anatomy Reconstruction" section above. The system and method builds upon the system described in the "Complete Anatomy Reconstruction" section to utilize the two 3D virtual models in conjunction with constructing a 3D virtual model of one or more bones (i.e., donor bones) from which a bone graft can be obtained. As explained in more detail below, the 3D virtual model of the patient's reconstructed abnormal anatomy is analyzed to generate a 3D virtual model of the bone graft required for reconstruction. This 3D virtual graft model is compared to a 3D virtual model of the donor bone to access one or more sites on the donor bone from which the bone graft can be resected. After determining the resection locations, a cutting guide and a graft placement guide are designed and manufactured to collect the bone graft and attach it to the site of reconstruction.

[0164] As an exemplary illustration, the systems and methods are described in the context of facial reconstruction, where the donor bone comprises the fibula. One skilled in the art should appreciate that the systems and methods may be applied to any reconstructive surgical procedure utilizing one or more bone grafts. Additionally, while facial reconstruction and the fibula as the donor bone are discussed, one skilled in the art should appreciate that the exemplary systems and methods may be used with toner bones other than the fibula.

[0165] As a preliminary step to discussing exemplary systems and methods for use in conjunction with reconstructive surgical planning and procedures using bone grafts, it is assumed that a patient's abnormal anatomy has been imaged and a virtual 3D model of the patient's abnormal anatomy and reconstructed anatomy has been generated according to the process described above in the "Complete Anatomy Reconstruction" section. As a result, a detailed discussion of utilizing patient images to generate virtual 3D models of both the patient's abnormal anatomy and reconstructed anatomy has been omitted in the interest of brevity.

[0166] After the virtual 3D models of the patient's abnormal anatomy and reconstructed anatomy are created, the software compares the anatomy and highlights areas of difference. In particular, common areas between the virtual 3D models indicate bones to be preserved, while different areas indicate one or more sites for reconstruction. The software extracts non-common areas from the virtual 3D model of the patient's reconstructed anatomy and isolates these areas as separate 3D virtual models of the target graft. The surgeon or other preoperative planner may view the virtual 3D bone graft model and make their own judgment regarding the bone or bones from which the bone graft can best be resected.

[0167] Regardless of the logic utilized to initially select potential bones as graft candidates, the bones of interest are imaged using conventional diagnostic techniques (X-ray, CT, MRI, etc.). Using the process described in the "Full Anatomy Reconstruction" section above, each imaged bone is segmented and a virtual 3D model of the imaged bone is created. This 3D donor bone model is compared to the virtual 3D bone graft model to isolate areas of commonality. In particular, the software compares the surface contours of the 3D donor bone model and the virtual 3D bone graft model to identify areas of commonality or areas with similar curvature. If no common or similar regions exist, the process can be started again by analyzing other possible donor bones. On the other hand, if one or more regions with common or similar curvatures exist in the donor bone, these regions are highlighted in the 3D donor bone model. In particular, the highlighted regions mimic the shape of the virtual 3D bone graft model. If the common regions are determined to be suitable for resecting the bone graft, the software virtually resects the bone graft as a virtual 3D model and applies the bone graft (which has a unique / unique contour with respect to the donor bone) to the virtual 3D model of the patient's abnormal anatomy, thereby verifying any regions of the patient's abnormal anatomy that may need to be resected as part of the reconstruction and the possible fit. In situations where the application of the virtual 3D model of the resected bone to the virtual 3D model of the patient's abnormal anatomy results in a less than satisfactory reconstruction, the process can be started again at the bone selection point or to resect a different region of the bone. However, if application of the virtual 3D model of the resected bone to the virtual 3D model of the patient's abnormal anatomy results in an appropriate fit, the system proceeds with a tooling design to facilitate resection of the bone graft and attachment of the bone graft to the patient's remaining bone.

[0168] In this exemplary embodiment, the system generates and outputs the necessary machine code for a rapid prototyping machine, CNC machine, or similar device to manufacture the bone graft cutting guide and the bone graft placement guide. To generate the necessary output to manufacture the bone graft cutting guide and the bone graft placement guide, the system utilizes a virtual 3D model of the resected bone and a virtual 3D model of the patient's abnormal anatomy.

[0169] In particular, the virtual 3D model of the resected bone defines the boundaries of the virtual 3D cutting guide. Also, in this exemplary situation, a portion of the fibula is to be resected to provide a bone graft. To ensure that the appropriate portion of the fibula is resected, the virtual 3D cutting guide includes a window through which a cutting device (saw, cutting drill, etc.) traverses to form a properly contoured bone graft. The virtual 3D cutting guide not only needs to be formed to provide the appropriate bone graft contour, but also needs to be formed to allow the placement of the cutting guide on the patient's donor bone to be specified. More specifically, the placement of the cutting guide on the donor bone needs to allow the resected bone to include the correct outline shape as well as the correct contour. In this embodiment, the underside of the virtual 3D cutting guide is formed to be the "negative" of the surface of the donor bone to which the cutting guide is attached. Exemplary attachment techniques for fixing the cutting guide to the donor bone can include, but are not limited to, screws, dowels, and pins. To accommodate one or more of these mounting techniques or others, the virtual 3D cutting guide is formed to include one or more through orifices other than the window through which the surgical cutter traverses. After the design of the virtual 3D cutting guide is completed, the system manufactures the bone graft cutting guide by generating and outputting the necessary machine code for a rapid prototyping machine, CNC machine, or similar device to then manufacture the actual cutting guide.

[0170] In addition to the cutting guide, the software also designs one or more bone graft placement guides. The bone graft placement guides are patient-specific and match the patient's anatomy (both the donor bone and the remaining bone to which the donor bone is attached) to ensure accurate placement of the bone graft relative to the remaining bone. In a typical form, the bone graft placement guide is configured for a mandibular reconstruction procedure. To design the bone graft placement guide, the software utilizes a virtual 3D model of the resected bone that is applied to a virtual 3D model of the patient's abnormal anatomy to construct a hybrid model. Using this hybrid model, junctions are identified where the bone graft connects (hopefully joined by bone growth) with the adjacent remaining bone. At these junctions, depending on various factors such as surgeon preferences, the system identifies bone graft plate locations and one or more guides for each plate to facilitate accurate placement and fixation of the plate relative to the bone graft and the remaining bone.

[0171] Those skilled in the art are familiar with conventional mandibular bone plates, and therefore a detailed discussion of the general design of mandibular bone plates has been omitted to promote brevity. What the present system and method achieves, unlike conventional systems and methods, is the creation of a patient-specific bone plate and placement guide that takes into account the shape of both the remaining bone and the bone graft. In particular, for each bone plate position identified (automatically or manually), the system designed a virtual 3D bone plate and associated placement guide. Each virtual 3D bone plate and guide model, whether it comprises a bone graft or a residual bone, is superimposed against the hybrid 3D model (including the bone graft and the patient's residual bone in their reconstructed positions) so that the underside of each virtual 3D bone plate and guide model is the negative of the underlying bone. In this way, the virtual 3D bone plate and guide models cooperate to ensure proper placement and corresponding engagement of the bone plate between the bone plate, the bone graft, and the residual bone. Exemplary attachment techniques for fastening the bone plates to the bone graft and residual bone may include, but are not limited to, screws, dowels, and pins. To accommodate one or more of these attachment techniques or other techniques, each virtual 3D bone plate and placement guide includes one or more through orifices. After the design of each virtual 3D bone plate and guide is completed, the system manufactures each 3D bone plate and guide by generating and outputting the necessary machine code for a rapid prototyping machine, CNC machine, or similar device to subsequently manufacture the actual bone plate and guide.

[0172] As will be apparent to those skilled in the art from the foregoing description and summary of the invention, the methods and apparatus described herein constitute exemplary embodiments of the invention, but the invention contained herein is not limited to this exact embodiment, and modifications may be made without departing from the scope of the invention as defined by the claims. It is also to be understood that the invention is defined by the claims, and that any limitations or elements describing the exemplary embodiments described herein are not intended to be incorporated into the interpretation of any elements described in the claims, unless such limitations or elements are expressly stated. Similarly, it should be understood that it is not necessary to meet any or all of the specified advantages or objectives of the invention disclosed herein in order to fall within the scope of any claim. This is because the invention is defined by the claims, and because inherent and / or unexpected advantages of the invention may exist even if they have not been expressly discussed herein.

[0173] [Additional note 1] 1. A method of constructing a patient-specific orthopaedic implant, comprising: comparing a patient-specific abnormal bone model derived from an actual anatomy of the patient's abnormal bone with a reconstructed patient-specific bone model also derived from the patient's bone anatomy, wherein the reconstructed patient-specific bone model reflects a normalized anatomy of the patient's bone, the patient-specific abnormal bone model reflects an actual anatomy of the patient's bone including at least one of an incomplete bone, a deformed bone, and a crushed bone, the patient-specific abnormal bone model comprises at least one of a patient-specific abnormal point cloud and a patient-specific abnormal bone surface model, and the reconstructed patient-specific bone model comprises at least one of a reconstructed patient-specific point cloud and a reconstructed patient-specific bone surface model; optimizing one or more parameters for a patient-specific orthopaedic implant to be fitted to the patient's abnormal bone using data output from the comparison of the patient-specific abnormal bone model and the reconstructed patient-specific bone model; generating an electronic design file for the patient-specific orthopaedic implant taking into account the one or more parameters; A method for providing the above. [Additional note 2] 2. The method of claim 1, further comprising manufacturing the patient-specific implant using the electronic design file. [Additional note 3] comparing the patient-specific abnormal bone model with the reconstructed patient-specific bone model to identify missing or deformed bones from the patient-specific abnormal bone model; localizing missing or deformed bone onto the reconstructed patient-specific bone model; The method according to claim 1 or 2, further comprising: [Additional note 4] generating said patient-specific abnormal bone model from data representative of an abnormal bone of a patient; generating the reconstructed patient-specific bone model from data representative of an abnormal bone of the patient and data from a statistical atlas, the statistical atlas data comprising at least one of a point cloud and a surface model of a normal bone similar to the abnormal bone of the patient; The method according to claim 3, further comprising: [Additional note 5] 5. The method of claim 4, wherein the data indicative of the patient's abnormal bone comprises at least one of a magnetic resonance image, a computed tomography image, an X-ray image, and an ultrasound image. [Additional note 6] 5. The method of claim 4, wherein the statistical atlas data is derived from at least one of magnetic resonance images, computed tomography images, X-ray images, and ultrasound images of normal bones. [Additional note 7] the identified missing or deformed bone comprises a set of boundary points; said step of localizing missing or deformed bone onto said reconstructed patient-specific bone model comprises the step of associating said set of boundary points with said reconstructed patient-specific bone model; The method according to claim 3. [Additional note 8] said comparing said patient-specific abnormal bone model to said reconstructed patient-specific bone model to identify missing or distorted bones from said patient-specific abnormal bone model includes outputting at least two lists of data, said at least two lists of data including a first list identifying missing or distorted bones and a second list identifying bones in common between said patient-specific abnormal bone model and said reconstructed patient-specific bone model; The method according to claim 3. [Additional note 9] 9. The method of claim 8, wherein the first list comprises vertices belonging to missing or deformed bones from the patient-specific abnormal bone model, and the second list comprises vertices belonging to bones that are common between the patient-specific abnormal bone model and the reconstructed patient-specific bone model. [Additional Note 10] 10. The method of any one of claims 1 to 9, further comprising a step of determining one or more patient-specific orthopedic implant fixation positions using data from the patient-specific abnormal bone model and data from the reconstructed patient-specific bone model. [Additional Note 11] 11. The method of claim 10, wherein the step of determining one or more patient-specific orthopedic implant fixation locations includes eliminating any locations where missing or deformed bone is identified. [Additional Note 12] 12. The method of any one of claims 1 to 11, wherein the step of optimizing one or more parameters for a patient-specific orthopedic implant comprises a step of using an implant parameterization template to define generic parameters that are subsequently optimized using the reconstructed patient-specific bone model. [Additional Note 13] 13. The method of claim 12, wherein the parameters include at least one of an angle parameter, a depth parameter, a curvature parameter, and a fixation device position parameter. [Additional Note 14] 14. The method of any one of claims 1 to 13, further comprising constructing an initial iteration of a surface model of the patient-specific orthopedic implant. [Additional Note 15] 15. The method of claim 14, wherein the step of constructing an initial iteration of a surface model includes a step of combining contours from the patient-specific abnormal bone model and contours from the reconstructed patient-specific bone model. [Additional Note 16] 15. The method of claim 14, wherein the step of constructing an initial iteration of a surface model includes taking into account an intended implantation location for the patient-specific orthopedic implant. [Additional Note 17] 15. The method of claim 14, further comprising constructing a subsequent iteration of the surface model of the patient-specific orthopedic implant. [Additional Note 18] 18. The method of claim 17, wherein the step of constructing a subsequent iteration of the surface model of the patient-specific orthopedic implant includes manual review of the surface model and the subsequent iteration of the reconstructed patient-specific bone model to determine whether further iterations of the surface model are required. [Additional Note 19] 19. The method of any one of claims 1 to 18, wherein the electronic design file comprises at least one of a computer-aided design file, a computer numerical control file, and a rapid manufacturing instruction file. [Additional Note 20] 20. The method of any one of claims 1 to 19, further comprising generating an electronic design file for a patient-specific implant placement guide using one or more parameters optimized for the patient-specific orthopedic implant. [Additional Note 21] 21. The method of claim 20, further comprising manufacturing the patient-specific implant placement guide using the electronic design file for the patient-specific implant placement guide. . [Additional Note 22] 21. The method of claim 20, wherein the one or more parameters optimized for the patient-specific orthopedic implant include at least one of a size parameter, a shape parameter, and a contour parameter. [Additional note 23] 23. The method of claim 22, wherein at least one of the contour parameters is common between the patient-specific orthopedic implant and the patient-specific implant placement guide. [Additional note 24] 24. The method of any one of claims 1 to 23, further comprising the step of designing the patient-specific implant placement guide to include a surface shape that is the negative of a surface shape of the patient's bone in which the patient-specific implant placement guide is intended to reside. [Additional note 25] the patient-specific abnormal bone model comprises at least one of a patient-specific abnormal femur model and a patient-specific abnormal pelvic bone model derived from the actual anatomy of the patient's abnormal hip joint; the reconstructed patient-specific bone model comprises at least one of a reconstructed patient-specific femur model and a reconstructed patient-specific pelvis model derived from the anatomy of the patient's hip joint; the reconstructed patient-specific model reflects normalized anatomy from a patient's hip joint; The patient-specific abnormal bone model reflects actual anatomy from a patient's hip joint. 25. The method according to any one of claims 1 to 24. [Additional note 26] the patient-specific abnormal bone model comprises a patient-specific abnormal femur model; The reconstructed patient-specific bone model comprises a reconstructed patient-specific femur model, the reconstructed patient-specific model reflects normalized anatomy from a proximal femur of a patient; the patient-specific abnormal bone model reflects actual anatomy from a patient's proximal femur; the patient-specific orthopaedic implant comprises a femoral stem implant. The method according to claim 25. [Additional note 27] the patient-specific abnormal bone model comprises a patient-specific abnormal pelvic bone model; The reconstructed patient-specific bone model comprises a reconstructed patient-specific pelvic bone model; the reconstructed patient-specific model reflects normalized anatomy from a patient's pelvis; the patient-specific abnormal bone model reflects the actual anatomy of a patient from a pelvis; the patient-specific orthopaedic implant comprises an acetabular cup implant. The method according to claim 25. [Additional note 28] 28. The method of any one of claims 1 to 27, wherein the electronic design file for the patient-specific orthopedic implant includes at least one of a computer-aided design file, a computer numerical control file, and a rapid manufacturing instruction file. [Additional note 29] 1. A method for generating an electronic reconstructed bone model of an abnormal bone, comprising: utilizing at least one of a point cloud and a surface model of the abnormal bone to at least one of: identify a bone from the statistical atlas that is similar to the abnormal bone; align the bone from the statistical atlas to the abnormal bone; and morph surface points on the reconstructed model of the abnormal bone to at least one of a point cloud and a surface model of the abnormal bone, the abnormal bone including at least one of an incomplete bone, a deformed bone, and a fractured bone; generating said reconstructed model of an abnormal bone; A method for providing the above. [Additional note 30] 30. The method of claim 29, wherein the step of utilizing at least one of a point cloud and a surface model of the abnormal bone includes a step of identifying a statistical atlas bone that is most similar to the abnormal bone. [Additional Note 31] The method of claim 29 or 30, wherein the step of utilizing at least one of a point cloud and a surface model of the abnormal bone includes a step of registering the statistical atlas to the abnormal bone. [Additional note 32] A method according to any one of appendix 29 to 31, wherein the step of utilizing at least one of a point cloud and a surface model of the abnormal bone includes a step of morphing surface points on a reconstructed model of the abnormal bone into at least one of a point cloud and a surface model of the abnormal bone. [Additional note 33] The method of claim 29 or claim 30, wherein the step of identifying a statistical atlas bone that is most similar to the abnormal bone comprises a step of using one or more similarity measures to identify the statistical atlas bone. [Additional note 34] 34. The method of any one of clauses 29, 30 or 33, wherein the statistical atlas comprises a plurality of mathematical representations, each of the plurality of mathematical representations representing a bone. [Additional note 35] 34. The method of any one of clauses 29, 30 or 33, wherein the statistical atlas comprises a plurality of virtual models, each of the plurality of virtual models representing a bone. [Additional note 36] 36. The method of any one of claims 29 to 35, further comprising a step of registering at least one of the point cloud and surface model of the abnormal bone to an identified bone from the statistical atlas that is similar to the abnormal bone. [Additional note 37] 37. The method of any one of claims 29 to 36, further comprising enhancing shape parameters between (a) at least one of a point cloud and a surface model of an abnormal bone and (b) an identified bone from the statistical atlas that is similar to the abnormal bone. [Additional note 38] 38. The method of claim 37, wherein the enhancement of the shape parameters includes interpolating between (a) at least one of the point cloud and surface model of the abnormal bone and (b) an identified bone from the statistical atlas that is similar to the abnormal bone to identify missing or deformed bones in at least one of the point cloud and surface model of the abnormal bone. [Additional note 39] 39. The method of claim 38, wherein the augmentation of the shape parameters results in the generation of surface points corresponding to missing or deformed bone. [Additional note 40] 39. The method of any one of claims 29 to 38, further comprising a step of morphing surface points interpolated from a bone from the statistical atlas that resembles the abnormal bone using at least one of the point cloud and a surface model of the abnormal bone to generate the reconstructed model of the abnormal bone. [Additional note 41] the abnormal bone comprises at least one of a deformed pelvic portion, a shattered pelvic portion, and an incomplete pelvic portion lacking bone; the reconstructed model of the abnormal bone comprises at least a complete pelvis model portion in which at least one of a bone deformation in the deformed pelvic portion, a comminuted bone comprising a portion of the comminuted pelvic portion, and a bone defect from the incomplete pelvic portion has been repaired; 41. The method according to any one of claims 29 to 40. [Additional note 42] 42. The method of claim 41, wherein the complete pelvis model portion includes an acetabular cup anatomy. [Additional note 43] the abnormal bone comprises at least one of a deformed femoral portion, a crushed femoral portion, and an incomplete femoral portion lacking bone; the reconstructed model of the abnormal bone comprises at least a complete femur model portion in which at least one of the bone deformation in the deformed femur portion, a comminuted bone comprising a portion of the comminuted femur portion, and a bone defect from the incomplete femur portion has been repaired; 41. The method according to any one of claims 29 to 40. [Additional note 44] 44. The method of claim 43, wherein the complete femur model portion comprises a proximal femur having a neck and ball anatomy. [Additional note 45] the abnormal bone comprises at least one of a deformed humerus portion, a crushed humerus portion, and an incomplete humerus portion lacking bone, a deformed ulna, a crushed ulna, an incomplete ulna lacking bone, a deformed radius portion, a crushed radius portion, an incomplete radius portion lacking bone, a deformed cranial portion, a crushed cranial portion, an incomplete cranial portion lacking bone, a deformed vertebral portion, a crushed vertebral portion, and an incomplete vertebral portion lacking bone, The reconstructed model of the abnormal bone comprises at least one of a complete humerus model portion, a complete ulna model portion, a complete radius model portion, a complete skull model portion, and a complete vertebral model portion, in which at least one of the bone deformation in the deformed ulna portion, the comminuted bone comprising a portion of the comminuted ulna portion, the bone defect from the incomplete ulna portion, the bone deformation in the deformed radius portion, the comminuted bone comprising a portion of the comminuted radius portion, the bone defect from the incomplete radius portion, the bone deformation in the deformed cranial portion, the comminuted bone comprising a portion of the comminuted cranial portion, the bone defect from the incomplete cranial portion, the bone deformation in the deformed vertebral portion, the comminuted bone comprising a portion of the comminuted vertebral portion, and the bone defect from the incomplete vertebral portion has been repaired. 41. The method according to any one of claims 29 to 40. [Additional note 46] 1. A method of constructing a mass customized orthopedic implant, comprising: Identifying features across a statistical atlas population of bones, said features comprising at least one of landmarks and shape features; generating a descriptor relevant to an implant design using features identified across said statistical atlas population of bones; categorizing at least some of the descriptors into groups having similar descriptors; parameterizing the group to extract parameters from the group; generating electronic design files for the mass-customized orthopaedic implant; A method for providing the above. [Additional note 47] 47. The method of claim 46, further comprising manufacturing a mass-customized orthopedic implant using the electronic design file. [Additional note 48] The method of claim 46 or claim 47, wherein the step of identifying features is performed automatically by a software program configured to calculate landmarks across a statistical atlas population of bones using positional parameters incorporated into the computational logic. [Additional note 49] The step of identifying features includes determining the location of bones using position parameters that are incorporated into computational logic. 49. The method of any one of clauses 46 to 48, which is performed automatically by a software program configured to calculate shape features across a statistical atlas population of. [Additional Note 50] 50. The method of any one of clauses 46 to 49, wherein the descriptor comprises a mathematical descriptor calculated across the entire statistical atlas population of bones. [Additional Note 51] 51. The method of any one of claims 46 to 50, wherein the step of classifying at least some of the descriptors into groups having similar descriptors comprises a step of using statistical analysis to define the groups. [Additional note 52] 52. The method of any one of claims 46 to 51, wherein the descriptors extracted from the group comprise design parameters for the shape of a mass-customized orthopedic implant. [Additional note 53] the descriptor comprises a mathematical descriptor; said step of parameterizing said groups to extract descriptors from said groups comprising converting said mathematical descriptors into surface descriptors; 53. The method according to any one of claims 46 to 52. [Additional note 54] 54. The method of any one of clauses 46 to 53, wherein the electronic design file for the mass-customized orthopedic implant includes a virtual three-dimensional model of the mass-customized orthopedic implant. [Additional note 55] The method of any one of claims 46 to 54, wherein the step of parameterizing the group to extract descriptors from the group includes a step of generating a virtual three-dimensional model of a mass-customized orthopedic implant. [Additional note 56] 56. The method of any one of claims 46 to 55, further comprising the step of extracting three-dimensional cancellous bone features across the entire statistical atlas population of bones, and generating a three-dimensional bone model for each bone in the statistical atlas population of bones incorporating the extracted cancellous bone features specific to that bone. [Additional note 57] 57. The method of any one of claims 46 to 56, further comprising the step of performing a porosity assessment for each bone in the statistical atlas population of bones to determine the size and pore diameter of cancellous bone. [Additional note 58] The method of any one of clauses 46 to 57, further comprising a step of performing a stress testing process that combines cancellous bone size data, pore size data, and surface descriptor parameters to generate the electronic design file for a mass-customized orthopedic implant. [Additional note 59] 59. The method of any one of clauses 46 to 58, wherein the electronic design file comprises at least one of a computer-aided design file, a computer numerical control file, and a rapid manufacturing instruction file. [Additional note 60] The method of any one of clauses 46 to 59, further comprising generating an electronic design file for the mass-customized implant placement guide using at least one of the extracted parameters. [Additional note 61] The method of claim 60, further comprising a step of manufacturing the mass-customized implant placement guide using the electronic design file for the mass-customized implant placement guide. [Additional note 62] 62. The method of any one of clauses 46 to 61, wherein the statistical atlas population of bones is ethnicity specific. [Additional note 63] 63. The method of any one of clauses 46 to 62, wherein the statistical atlas population of bones is gender specific. [Additional note 64] 64. The method of any one of claims 46 to 63, wherein said statistical atlas population of bones comprises at least a segment of the femur. [Additional note 65] 65. The method of any one of claims 46 to 64, wherein said statistical atlas population of bones comprises at least a segment of the pelvic bone. [Additional note 66] 1. A method of constructing a patient-specific cutting guide for preparing bone for an orthopaedic implant, comprising: processing a patient-specific bone contour to determine a size of an orthopaedic implant to be placed in the patient's bone and a position of said implant when placed relative to the patient's bone; designing a patient-specific cutting guide using the size of the orthopaedic implant and the position of the implant when installed in a patient's bone; manufacturing a patient-specific cutting guide that includes a shape that is the negative of the shape of the patient's bone to which the cutting guide is designed to be attached; A method for providing the above. [Additional note 67] 67. The method of claim 66, further comprising the step of extracting patient-specific contours from at least one X-ray depicting the patient's bones. [Additional note 68] 67. The method of claim 66, further comprising the step of extracting patient-specific contours from at least one computed tomography image or magnetic resonance image depicting the patient's bones. [Additional note 69] 1. A method of constructing a mass customized trauma plate, comprising: defining a virtual boundary for a mass-customized trauma plate relative to the virtual three-dimensional bone model template; selecting a number of surface points within the virtual boundary that correspond to surface locations on the virtual three-dimensional bone model template; Propagating a plurality of surface points throughout a statistical atlas that includes a plurality of virtual three-dimensional bone models; constructing a virtual three-dimensional bone plate using the plurality of surface points propagated to each of the plurality of virtual three-dimensional bone models, the virtual three-dimensional bone plate being fitted to that particular bone model; extracting a plurality of curvatures representative of each of the formed virtual three-dimensional bone plates; statistically analyzing the extracted curvatures to estimate shape parameters for the mass-customized wound plate; generating an electronic design file for the mass-customized wound plate using the shape parameters; A method for providing the above. [Additional note 70] 70. The method of claim 69, further comprising manufacturing the mass-customized trauma plate using the electronic design file.

Claims

1. 1. A method for designing a patient-specific bone graft placement guide as part of a facial reconstruction procedure, the method comprising: identifying a bone graft from a donor bone other than the mandible, the identifying comprising a region of the donor bone having a surface contour similar to a desired bone graft for the mandible; designing the patient-specific bone graft placement guide, the patient-specific bone graft placement guide incorporating a first shape configured to precisely align the patient-specific bone graft placement guide to the donor bone in only a single orientation and position, and a second shape configured to precisely align the patient-specific bone graft placement guide to the patient bone in only a single orientation and position; 23. A method comprising:

2. 2. The method of claim 1, wherein the step of identifying the bone graft from the donor bone comprises comparing a three-dimensional virtual model of the desired bone graft with a three-dimensional virtual model of the donor bone to identify the region of the donor bone having a surface contour similar to the desired bone graft.

3. 3. The method of claim 2, wherein comparing the three-dimensional virtual model of the desired bone graft with the three-dimensional virtual model of the donor bone comprises comparing a surface contour of the three-dimensional virtual model of the desired bone graft with the surface contour of the three-dimensional virtual model of the donor bone to identify one or more regions of the donor bone that have the same or similar curvature as the desired bone graft.

4. 2. The method of claim 1, wherein the step of designing a patient-specific bone graft placement guide comprises designing an underside of the patient-specific bone graft placement guide as a negative of the underlying patient bone and the underlying bone graft derived from the donor bone, such that the underside fits in a single direction and position to the underlying patient bone and the underlying bone graft derived from the underlying donor bone.

5. the patient-specific bone graft placement guide comprises an elongated handle and a block C-shaped contour plate; 5. The method of claim 4, wherein the lower surface of the contour plate comprises a negative of the underlying patient bone and the underlying bone graft from the donor bone.

6. 2. The method of claim 1, wherein the step of designing the patient-specific bone graft placement guide comprises utilizing a hybrid virtual model comprising the three-dimensional virtual model of the bone graft from the donor bone applied to a three-dimensional virtual model of the patient bone to obtain the shape and ensure placement of the patient-specific bone graft placement guide on the patient bone in only the single orientation and position.

7. The method of claim 1, further comprising outputting machine code for manufacturing the patient-specific bone graft placement guide.

8. The method of claim 1 , further comprising manufacturing the patient-specific bone graft placement guide.

Images