Customised knee implants within hours – thanks to AI

Customised knee implants within hours – thanks to AI

Total knee replacement (TKR) is the most effective treatment to relief pain and restore normal function in a diseased knee joint.

VOLMO LTD(UK) in partnership with the University of Birmingham have developed  an innovative design tool that enables these implants to be customised for individual patients. These bespoke implants can then be 3D printed in medical grade titanium alloys. The platform was tested for a real patient case in which all three implant components were produced, femur, spacer and tibial implant. Powered by artificial intelligence, medical image processing and computer aided design algorithms this innovative design platform automatically created a 3D model of the knee joint within minutes from which the final customised implant component designs can be produced within an hour and only requires minimal user interaction. These designs were then 3D printed in the Centre for Custom Medical Devices (University of Birmingham) using a RenAM 500S (Renishaw, UK) selective laser melting system in medical grade Ti-6Al-4V alloy. 


  1. The need for patient specific knee implants 

Although with current techniques implant survivorship has been excellent, reported patient satisfaction has lagged substantially behind other orthopaedic procedures, such as total hip arthroplasty. More than 20% of TKR patients report dissatisfaction with implant outcomes [4]. There are a number of reasons why such a large subset of patients are dissatisfied, with the most common reasons cited as pain, stiffness, and limited function [4].      Although many implants currently on the market perform well, they still fall short in successfully achieving normal knee kinematics [4]. Residual pain may be attributed to lack of fit or misalignment, with data suggesting up to 27% to internal rotation errors [3].After TKR patients may experience substantial functional impairment [3]. Even amongst satisfied TKR patients around 20% state their knee does not feel normal. Issues such as instability,  malalignment  and poor fixation contribute to around 50% of early revision TKR [3].  

Every patient has a unique anatomical geometry and these geometries can vary because of gender, ethnicity, and body type between patients [4]. There is also substantial variation between individual subjects within these groups, suggesting that a customised implant may be advantageous [4]. The optimal TKR implant should reproduce knee function, maintain bone-implant interface integrity, and resist wear. The kinematics and loads on the TKR implant are very important to the success of TKR.  

2.0 Methodology:   Patient Specific Knee Implant Design 

  1. a)   Patient CT scan to 3D model generation . The method used in this study was to obtain computed tomography (CT) scan data of a patient’s knee that required TKR. VOLMO’s proprietary software, ImageSim, was used to segment this data using an automatic function enabling isolation of the femur and tibia sections requiring replacement (Figure 1 – green). The artificial intelligence (AI) powered segmentation module of ImageSim allows users to automatically create a mask of the knee joint with specific length of femur and tibia bone segments. This enables the rapid generation of an accurate knee joint model within minutes. In the current case we generated the mask of a full knee joint in 3 minutes on a  laptop with a i7 processor and 16GB RAM.  From this masks in a few clicks the users can then generate a 3D model (Figure 1 – yellow).

Figure [1] Knee joint mask (green) and 3D knee joint model (yellow) generated within minutes using VOLMO’s AI driven software ImageSim

b) 3D Model to customized implantImageSim has an integrated computer aided design (CAD) engine, ImpCAD, that imports the segmented model and allows the user to apply various operations.  For this knee implant it allows the user to perform bone resections at various angles and distances from a centre landmark as shown in figure 2. These interactions are all that is required from the user to create the final customised femur and tibia components. Once defined the implant designs took less than an hour to create using a laptop with a i7 processor and 16GB RAM (Figure 3).  Once the final implant is created the user has an ability to edit the design if required enabling engaging with clinical colleagues. This unique design ability of ImageSim to quickly generate the final implant within hour or so is a total transformation from the current design process and would offer a staggering benefit  in cost and  time saving.  

Figure[2]: ImpCAD- Performing bone resections

Figure[3]: ImpCAD – Final Customized Implant along with spacer generated

c) 3D Print designed Implants :

Final implant design files created in ImpCAD were printed at the University of Birmingham in medical grade Ti-6Al-4V alloy using a RenAM 500S selective laser melting machine. The total print time for the metallic components was 15 hours and the corresponding anatomical model components (Figure 4 – white) were 3D printed using a i3 (Prusa) in 6 hours. . Thus, the complete design to 3D print of customized implant was achieved under 24 hours. We believe this radical transformation from the current industry practice and could add great value to industry . 

Figure 4: Customised total knee replacement implant components assembled within the corresponding anatomical structure produced via 3D printing technologies


Authors: Dr Ash Harkara VOLMO LTD (UK), Dr Sophie Cox, University of Birmingham (UK) 




Categories: ARTICLES

Write a Comment