Conformable and robust force sensors to enable precision joint replacement surgery

Balancing forces within weight-bearing joints such as the hip during joint replacement surgeries is essential for implant longevity. Minimising implant failure is vital to improve patient wellbeing and alleviate pressure on healthcare systems. With improvements in surgery, hip replacement patients are now often younger and more active than in previous generations, and their implants correspondingly need to survive higher stresses. However, force balancing currently depends entirely on surgical skill: no sensors can provide quantitative force feedback within the hip joint’s small, complex geometry. Here, we solve this unmet clinical need by presenting a thin and conformable microfluidic force sensor, which is compatible with the standard surgical process. We optimised the design using finite element modelling, then incorporated and calibrated our sensor in a model hip implant. Using a bespoke testing rig, we demonstrated high sensitivity at typical forces experienced during hip replacements. We anticipate that these sensors will aid implant positioning, increasing the lifetime of hip replacements, and represent a powerful new surgical tool for a range of orthopaedic procedures where force balancing is crucial.


Sensor Design and Operation
is a schematic of the sensor, whose design is based on the group's previous work 25 . The sensor consists of (1) a microfluidic chip (Fig. 1b) with an embedded microfluidic channel (20 mm x 0.75 mm x 0.3 mm) and fluid reservoir (2 mm x 2 mm x 0.3 mm), and (2) an electrode layer (Fig. 1c) comprising interdigitated silver electrodes on a flexible polyimide (PI, Kapton) substrate for mechanical support. The chip can be made using a silicone elastomer such as polydimethylsiloxane (PDMS) poured into a stereolithography (SLA) 3D-printed mold, or directly SLA 3D-printed by photo-curing an elastomeric resin (Flexible Resin, Formlabs). The channel is aligned above the electrodes and is open to air at the end opposite to the reservoir. The reservoir is the active sensing area. It has a square cross section, is filled with fluid by a syringe, and can contain internal columns for mechanical support. The fluid is a 2:1 volume mixture of glycerol and deionised (DI) water, to balance the low volatility and high dielectric constant 25 . The interdigitated electrodes are made by aerosol jet printing (AJP) silver nanoparticle ink onto Kapton and are protected by a printed PI layer. The chip and electrode layer are bonded using laser-cut double-sided tape. The operating principle is as follows 25 : when a force is applied to the fluid reservoir, the reservoir deforms and displaces fluid along the channel. The displaced fluid overlaps with the interdigitated electrodes, increasing the capacitance. The capacitance is calculated using the equation = 0 / , where is the relative dielectric permittivity, is the electrode area and is the inter-electrode distance. The fluid determines the value of . On releasing the force, the fluid returns to the reservoir.
The sensors are durable, with measurements reproduced over more than 2,000 loading cycles in a mechanical testing rig 25 . For each sensor, the force-capacitance relationship is calibrated by applying a known force and measuring the corresponding impedance using an impedance analyser (ISX-3, Sciospec), which is converted to a capacitance. For the present application, an array of sensors is embedded into the UHMWPE component.
Importantly, these sensors can be easily and rapidly customised to suit a range of applications by changing the reservoir and channel size, electrode size and spacing, chip material and device size. While this sensor has been developed for hip applications, we have fabricated sensors with alternative designs, including a sensor incorporating three sensing elements intended for the knee joint both for the TKR and unicompartmental applications.

Figure 1 | Incorporating functionalised microfluidic sensors into the THR implant for force measurements. a
The sensor is made of a soft elastomeric microfluidic chip layer and a Kapton substrate with aerosol-jet printed interdigitated electrodes. b The microfluidic chip layer contains an embedded microchannel with a fluid reservoir and optional supporting columns. c Interdigitated electrodes are aerosol jet printed onto the Kapton substrate and consist of silver with an insulating polyimide coating. d Photograph of the sensor, highlighting the channel and reservoir. e Photograph of sensor, showing its flexibility. In this design, the reservoir has a round cross-section. f Standard geometry of a hip implant. The cup containing the sensors is incorporated into the polymer part of the implant's acetabular cup. g Our additions to the implant, during the trial stage, consist of an outer cup, the sensors which lie in the grooves of the outer cup, and the inner cup to act as an articulating surface with the femoral head. h Photograph of the SLA 3D-printed inner and outer cup. i Photograph of a sensor located in a groove between the inner and outer cups. j Cross-section of the outer cup, showing how variable groove depth allows the channel to be shielded from force while the active sensing area (reservoir) is exposed.

Incorporation into Implant
To test the sensor's performance in the hip, a biomimetic trial insert ( Fig. 1f) was designed using Creo Parametric (PTC) and produced by SLA 3D printing. The insert (Fig. 1g,h) consists of an 'outer cup', which contains grooves to fit sensors; and an 'inner cup', which has a smooth inner surface to act as an articulating surface with the femoral head, and pegs to secure it in the outer cup. It was decided to incorporate six sensors into the trial insert to provide a balance between providing many sensing regions and mechanical stability when loading the sensors. The sensors were incorporated into the grooves (Fig. 1i) such that the reservoirs were at 30° to the cup axis. In this study, the outer cup has an external diameter of 43 mm and an internal diameter of 41 mm, but this can be adapted to suit femoral heads of different sizes. Typically, the femoral head has a diameter between 28 and 40 mm 1 . The groove depth is deliberately reduced going from the outside of the cup towards the centre (Fig. 1j). This gives the effect of raising the reservoirs out of the grooves, such that an applied force is concentrated on the fluid reservoirs (the active sensing area) while shielding the channel.

Mechanical Testing of Sensors within Implant Geometry Validates Simulation Results
Finite element models were produced to: 1) improve the sensor design by modelling electrode behaviour  which the capacitance response starts to plateau, which would complicate calibration. The maximum capacitance change Δ also decreases with increase in above 700 N, indicating that fluid may be lost from the sensor. This is an appropriate force range for applications in hip surgery, but it can be increased by modifying the sensor design, for example by changing the channel and reservoir dimensions.  S15). The data in Fig. 2c and 2d also show a delay of a few seconds between applying the force and the sensors' response, which may be due to force being applied too quickly relative to the fluid's relaxation time. In this loading setup, the gradient of the capacitance-force calibration curve was lowerapproximately 0.03 pF N -1 compared to 0.06 pF N -1 when calibrated outside of the cup 25 . It is therefore likely that the cup is acting to shield the sensors from these forces as previously indicated by the modelling results. This agrees with the observation that the maximum operating applied force for the sensors was considerably higher in this trial insert than when the sensors were calibrated on their own.
We have also characterised how modifications to both the chip design and material may affect device performance; such data (experimental and computational) is in the Supporting Information. [Bottom] In order to reach higher forces, a different machine was usedtherefore, the measurement duration is longer. The sensor is operational up to at least 400 N, above which the response begins to plateau. c Loading six sensors in the trial insert up to = 20 N at a cup angle = 0°. The coloured lines and dots indicate different sensors. d Loading six sensors in the trial insert up to = 20 N. The force is applied normal to a sensor's reservoir, at a cup angle = 30°.

Discussion
This paper summarises the design, production and characterisation of a next-generation hip implant technology that contains microfluidic force sensors, designed to replace part of the trial insert during surgery while the surgeon dynamically assesses the THR positioning, soft tissue tensioning, size and prosthetic impingement. This will give the surgeon quantitative data on the interfacial forces between the implant's femoral head and insert during the THR, so that imbalance and improper positioning can be detected and corrected, decreasing the rate of implant failure. This is crucial, as surgeon experience is linked to fewer complications during surgery and better patient satisfaction post-surgery 28,29 .
Furthermore, it is well-documented that specialised hip surgeons are less likely to have complications as compared with general orthopaedic surgeons, and patients operated on by orthopaedic trainees report less satisfaction than surgeons with greater than 15 years of experience according to data from the Swedish Joint Registry [28][29][30] . The objective force feedback our technology provides has the potential to augment training and reduce learning curves in hip replacement surgery by allowing trainee surgeons to receive visual and haptic feedback of the accuracy of implant positioning during the trialling process.
Therefore, quantitative force data could be a useful tool to both improve the technique of experienced surgeons and as a training tool for non-specialists to perform the same surgery with the same outcomes.
Although accurate implant placement has now been addressed to an extent with navigation or robotic assisted techniques, difficulties with inappropriate soft tissue balance and prosthetic impingement remain unaddressed, leading to poor functional outcomes and accelerated wear 21 . Furthermore, they require a significant amount of investment and are not available for routine use in most hospitals worldwide 22 .
There are currently no commercial force sensors that are thin, conformable, and capable of bearing the large loads of a THR in a durable manner. The closest analogue is the VERASENSE (Orthosensor Inc., USA) for total knee replacement, which uses piezoelectric elements to quantitatively assess soft tissue balance and implant positioning to guide surgeons. However, this cannot be adapted for the hip as the hip joint contact zone is curved and has much less space than the knee.
Many different existing sensing mechanisms could potentially meet the clinical need for quantitative force sensing, including capacitive, resistive, piezoresistive, optical, triboelectric, magnetic, passive resonator and FET-based sensors 31 . Capacitive sensors have limited spatial resolution compared to resistive ones, but have a higher reliability [32][33][34][35][36][37] . Flexible microfluidic force sensors have been made previously for tactile and haptic sensing [38][39][40] , but could not reach the high forces required in this application. Some devices can reach higher resolutions, but each sensing element requires a large support system featuring electrical connections 41 or power supplies.
Here, we have developed a capacitive microfluidic force sensor that is conformable and functional at the forces used in hip surgery. The sensors were experimentally and computationally validated for application in THR. Mechanical characterisation has demonstrated that the sensors can measure up to at least 400 N, which meets the expected requirements for the force range applied by the surgeon during THR 31,42 . This is