Background

The literature quotes up to 20% dissatisfaction rates for total knee replacements (TKR). Swedish registry and national joint registry of England and Wales confirm this with high volumes of patients included. This dissatisfaction rate is used as a basis for improving/changing/modernising knee implant designs by major companies across the world.

The knee joint displays a wide spectrum of laxity, from inherently tight to excessively lax even within the normal, uninjured population. The assessment of AP knee laxity in the clinical setting is performed by manual passive tests such as the Lachman test. Non-invasive assessment based on image free navigation has been clinically validated and used to quantify mechanical alignment and coronal knee laxity in early flexion. When used on cadavers the system demonstrated good AP laxity results with flexion up to 40°. This study aimed to validate the repeatability of the assessment of antero-posterior (AP) knee joint laxity using a non-invasive image free navigation system in normal, healthy subjects.

Twenty-five healthy volunteers were recruited and examined in a single centre. AP translation was measured using a non-invasive navigation system (PhysioPilot) consisting of an infrared camera, externally mounted optical trackers and computer software. Each of the volunteers had both legs examined by a single examiner twice (two registrations). The Lachman test was performed through flexion in increments of 15°. Coefficients of Repeatability (CR) and Interclass Correlation Coefficients (ICC) were used to validate AP translation. The acceptable limits of agreement for this project were set at 3mm for antero-posterior tibial translation.

The most reliable and repeatable AP translation assessments were at 30° and 45°, demonstrating good reliability (ICC 0.82, 0.82) and good repeatability (CR 2.5, 2.9). The AP translation assessment at 0°, 15°, 75° and 90° demonstrated moderate reliability (ICC ≤ 0.75), and poor repeatability (CR ≥3.0mm).

The non-invasive system was able to reliably and consistently measure AP knee translation between 30° and 45° flexion, the clinically relevant range for this assessment. This system could therefore be used to quantify abnormal knee laxity and improve the assessment of knee instability and ligamentous injuries in a clinic setting.

Utilisation of unicondylar knee arthroplasty (UKA) has been limited due in part to high revision rates. Only 8% of knee arthroplasty surgeries completed in England and Wales are UKAs. It is reported that the revision rate at 9 years for Total Knee Arthroplasty (TKA) was 3% compared to 12% for UKAs. In the last decade semi active robots have been developed to be used for UKA procedures. These systems allow the surgeon to plan the size and orientation of the tibial and femoral component to match the patient's specific anatomy and to optimise the balancing the soft tissue of the joint. The robotic assistive devices allow the surgeon to execute their plan accurately removing only ‘planned’ bone from the predefined area. This study investigates the accuracy of an imageless navigation system with robotic control for UKA, reporting the errors between the ‘planned’ limb and component alignment with the post-operative limb and component alignment using weight bearing long leg radiographs. We prospectively collected radiographic data on 92 patients who received medial UKA using an imageless robotic assisted device across 4 centres (4 surgeons). This system is CT free, so relies on accurate registration of intra-operative knee kinematic and anatomic landmarks to determine the mechanical and rotational axis systems of the lower limb. The surface of the condylar is based on a virtual model of the knee created intra-operatively by ‘painting’ the surface with the tip of a tracked, calibrated probe. The burring mechanism is robotically controlled to prepare the bone surface and remove the predefined volume of bone. The study shows the 89% of the patients' post-operative alignment recorded by the system was within 30 of the planned coronal mechanical axis alignment. The RMS error was 1.980. The RMS errors between the robotic system's implant plan and the post-operative radiographic implant position was; femoral coronal alignment (FCA) 2.6o, tibial coronal alignment (TCA) 2.9o and tibial slope (TS) 2.9o. In conclusion, the imageless robotic surgical system for UKA accurately prepared the bone surface of the tibia and femur which resulted in low errors when comparing planned and achieved component placement. This resulted in a high level of accuracy in the planned coronal mechanical axis alignment compared to that measured on post-operative radiographs.

Unicondylar knee arthroplasty (UKA) is growing in popularity with an increase in utilisation. As a less invasive, bone preserving procedure suitable for knee osteoarthritic patients with intact cruciate ligaments and disease confined to one compartment of the knee joint. The long term survival of a UKA is dependent on many factors, including the accuracy of prosthesis implantation and soft tissue balance. Robotic assisted procedures are generally technically demanding, can increase the operation time and are associated with a learning curve. The learning curve for new technology is likely to be influenced by previous experience with similar technologies, the frequency of use and general experience performing the particular procedure. The purpose of this study was to determine the time to achievement of a steady state with regards to surgical time amongst surgeons using a novel hand held robotic device.

This study examined consecutive UKA cases which used a robotic assistive device from five surgeons. The surgeons had each performed at least 15 surgeries each. Two of the surgeons had previous experience with another robotic assistive device for UKA. All of the surgeons had experience with conventional UKA. All of the surgeons have used navigation for other knee procedures within their hospital. The system uses image free navigation with infrared optical tracking with real time feedback. The handheld robotic assistive system for UKA is designed to enable precision of robotics in the hands of the surgeon. The number of surgeries required to reach ‘steady state’ surgical time was calculated as the point in which two consecutive cases were completed within the 95% confidence interval of the surgeon's ‘steady state’ time.

The average surgical time (tracker placement to implant trial acceptance phase) from all surgeons across their first 15 cases was 56.8 minutes (surgical time range: 27–102 minutes). The average improvement was 46 minutes from slowest to quickest surgical times. The ‘cutting’ phase was reported as decreasing on average by 31 minutes. This clearly indicates the presence of a learning curve. The surgeons recorded a significant decrease in their surgical time where the most improvement was in the process of bone cutting (as opposed to landmark registration, condyle mapping and other preliminary or planning steps). There was a trend towards decreasing surgical time as case numbers increase for the group of five surgeons. On average it took 8 procedures (range 5–11) to reach a steady state surgical time. The average steady state surgical time was 50 minutes (range 37–55 minutes).

In conclusion, the average operative time was comparable with clinical cases reported using other robotic assistive devices for UKA. All five surgeons using the novel handheld robotic-assisted orthopaedic system for UKA reported significant improvement in bone preparation and overall operative times within the first 15 cases performed, reaching a steady state in surgical times after a mean of 8 cases. Therefore, this novel handheld device has a similar learning curve to other devices on the market.

Knee osteoarthritis results in pain and functional limitations. In cases where the arthritis is limited to one compartment of the knee joint then a unicondylar knee arthroplasty (UKA) is successful, bone preserving option. UKA have been shown to result in superior clinical and functional outcomes compared to TKA patients. However, utilisation of this procedure has been limited due primarily to the high revision rates reported in joint registers. Robotic assisted devices have recently been introduced to the market for use in UKA. They have limited follow up periods but have reported good implant accuracy when compared to the pre-operative planned implant placement.

UKA was completed on 25 cadaver specimens (hip to toe) using an image-free approach with infrared optical navigation system with a hand held robotically assisted cutting tool. Therefore, no CT scan or MRI was required. The surface of the condylar was mapped intra operatively using a probe to record the 3 dimensional surface of the area of the knee joint to be resurfaced. Based on this data the size and orientation of the implant was planned. The user was able to rotate and translate the implant in all three planes. The system also displays the predicted gap balance graph through flexion as well as the predicted contact points on the femoral and tibial component through flexion. The required bone was removed using a bur. The depth of the cut was controlled by the robotically controlled freehand sculpting tool.

Four users (3 consultant orthopaedic surgeon and a post-doctoral research associate) who had been trained on the system prior to the cadaveric study carried out the procedures. The aim of this study was to quantify the differences between the ‘planned’ and ‘achieved’ cuts. A 3D image of the ‘actual’ implant position was overlaid on the ‘planned’ implant image. The errors between the ‘actual’ and the ‘planned’ implant placement were calculated in three planes and the three rotations. The maximum femoral RMS angular error was 2.34°. The maximum femoral RMS translational error across all directions was up to 1.61mm. The maximum tibial RMS angular error was 2.60°. The maximum tibial RMS translational error across all directions was up to 1.67mm.

In conclusion, the results of this cadaver study reported low RMS errors in implant position placement compared to the plan. The results were comparable with those published from clinical studies investigating other robotic orthopaedic devices. Therefore, the freehand sculpting tool was shown to be a reliable tool for cutting bone in UKA and the system allows the surgeon to plan the placement of the implant intra operatively and then execute the plan successfully.

For patients suffering from osteoarthritis confined to one compartment of the knee joint, a successful unicondylar knee arthroplasty (UKA) has demonstrated an ability to provide pain relief and restore function while preserving bone and cruciate ligaments that a total knee arthroplasty (TKA) would sacrifice. Long-term survival of UKA has traditionally been inconsistent, leading to decreased utilisation in favour of alternative surgical treatment. Robot-assisted UKA has demonstrated an ability to provide more consistent implantation of UKA prosthesis, with the potential to increase long-term survivorship.

This study reports on 65 patients undergoing UKA using an image-free, handheld robotic assistive navigation system. The condylar surface was mapped by the surgeon intra-operatively using a probe to capture a 3-dimensional representation of the area of the knee joint to be replaced. The intra operative planning phase allows the surgeon to determine the size and orientation of the femoral and tibial implant to suit the patients’ anatomy. The plan sets the boundaries of the bone to be removed by the robotic hand piece. The system dynamically adjusts the depth of bone being cut by the bur to achieve the desired result. The planned mechanical axis alignment was compared with the system's post-surgical alignment and to post-operative mechanical axis alignment using long leg, double stance, weight bearing radiographs.

All 65 knees had knee osteoarthritis confined to the medial compartment and UKA procedures were completed using the handheld robotic assistive navigation system. The average age and BMI of the patient group was 63 years (range 45–82 years) and 29 kg/m² (range 21–37 kg/m²) respectively. The average pre-operative deformity was 4.5° (SD 2.9°, Range 0–12° varus). The average post-operative mechanical axis deformity was corrected to 2.1° (range 0–7° varus). The post-operative mechanical axis alignment in the coronal plane measured by the system was within 1° of intra-operative plan in 91% of the cases. 3 out of 6 of the cases where the post-operative alignment was greater than 1° resulted due to an increase in the thickness of the tibia prosthesis implanted. The average difference between the ‘planned’ mechanical axis alignment and the post-operative long leg, weight bearing mechanical axis alignment was 1.8°. The average Oxford Knee Score (old version) pre and post operation was 38 and 24 respectively, showing a clinical and functional improvement in the patient group at 6 weeks post-surgery.

The surgical system allowed the surgeons to precisely plan a UKA and then accurately execute their intra operative plan using a hand held robotically assisted tool. It is accepted that navigation and robotic systems have a system error of about 1° and 1mm. Therefore, this novel device recorded accurate post-operative alignment compared to the ‘planned’ post-operative alignment. The patients in this group have shown clinical and functional improvement in the short term follow up. The importance of precision of component alignments while balancing existing soft-tissue structures in UKA has been documented. Utilisation of robotic-assisted devices may improve the accuracy and long-term survivorship UKA procedure.

NavioPFS™ unicondylar knee replacement (UKR) system combines CT-free planning and navigation with robotically assisted bone preparation. In the planning procedure, all relevant anatomic information is collected under navigation, either directly with the point probe or by kinematic manipulation. In addition to key anatomic landmarks and the maps of the articulating surfaces of the femur and tibia, kinematic assessment of the joint laxity is performed. Relative positions of femur and tibia are collected through the flexion/extension range, with the pressure applied to fully stretch the collateral ligament on the operative side.

The planning procedure involves three stages: (1) the implant sizing and initial placement,(2) balancing of the gap on the operative side and (3) evaluating the contact points for the recorded flexion data and the planned placement of implants. In the gap balancing stage, the implants are repositioned until they allow for a positive gap, preferably uniform, throughout the entire range of flexion. UKR was planned and prepared on six cadaver knees with the help of NavioPFS system. All knees were normal without any signs of osteoarthritis. Two surgeons have performed medial UKR (4+2), and the bones were prepared using the NavioPFS handheld robotic tool.

Postoperatively, we have re-used the data collected during the planning procedure to compare the kinematic (gap balancing) performance of the used implant with three different commercial implant designs. All implants were placed in the orientation recommended by the respective manufacturer, sized to best fit the original bone geometry, and repositioned optimally balance the gap curve through the entire flexion range, without any negative gaps (overlaps). Since these were nonarthritic cadaver knees, the intent was to restore the original preoperative varus/valgus in neutral (zero) flexion.

The three implant designs demonstrated variable degree of capability to uniformly balance the knee gap over the entire range of flexion. The first implant (A) required a gap larger than 2 mm in one case out of six, the second (B) was capable of producing the positive gap curve under 2mm of gap in all six cases, and the third (C) required a gap larger than 2 mm in 3 (50%) of cases. All three designs exhibit the reduced gap space in mid (30°–90°) flexion.

Despite the best attempts, the artificial implants do not fully replicate the healthy knee kinematics. This is manifested by increased tightness in the mid flexion. In order to balance the gap in mid flexion, additional laxity has to be allowed in full flexion, extension, or both. NavioPFS allows for patient specific planning that takes into account this information, only available intraoperatively. This kind of evaluation on a patient specific basis is a very important planning tool and it allows the insight on the implant performance in mid flexion, typically not available using conventional planning techniques. It can also help in improving kinematic performance of future implant designs.

The Columbus® knee system was designed as a standard knee implant that allows high flexion without the need for additional bone resection. The aim of this retrospective study was to investigate the correlation between the maximum flexion achieved at five years and the slope of the tibial component. The hypothesis was that increased slope would give increased flexion.

The study design was a retrospective cohort study at a single centre. The inclusion criterion was having had a navigated cemented Columbus primary TKA implanted between March 2005 and December 2006 using the image free OrthoPilot® navigation system (Aesculap, Tuttlingen, Germany) in our institution. Follow-up had been carried out at review clinics by an independent arthroplasty team. Patient-related data had been recorded either in case notes, the departmental proprietary database or as radiographic images. In addition to demographics, five-year follow-up range of motion (ROM) was collected. All available radiographs on the national Picture Archiving and Communication System (Eastman Kodak Company, 10.1_SP1, 2006), whether taken at our institution or at the patient's local hospital, were analysed by a trainee orthopaedic surgeon (NCS) who was independent of the patients' care. Component position according to the Knee Society TKA scoring system was determined from the five-year review lateral x-ray. The tibial slope was calculated as 90° minus the angle of the tibial component so giving a posterior slope as a positive number and an anterior slope as a negative number. The correlation between maximum flexion angle and tibial slope was calculated. Further to this a subgroup of only CR prostheses and patients with BMI <35 were analysed for a relationship. The tibial slope of the group of patients having 90° or less of flexion (poor flexion) was compared to those having 110° or more (good flexion) using a t-test, as was the flexion of the those with BMI <30 to those with BMI > 35.

A total of 219 knees in 205 patients were identified. 123 had five-year radiograph and maximum flexion measurement available. Cohort demographics were mean age 68(8.6), mean BMI 32.0(5.9) and mean maximum flexion at five years of 101°(11°). The tibial slope angle showed variation around the mean of 2°(2.8°). There was no correlation between tibial slope and maximum flexion for either that whole cohort (r=-0.051, p=0.572, Figure 1b) or the subgroup of CR and BMI <35 patients (n=78, r = −0.089, p=0.438). The mean tibial slope of those patients having poor flexion was 2° (SD2.6°) and this was not significantly different to the mean for those with good flexion, 3° (SD3.1°) p=0.614. The mean flexion of those with BMI <30 was 100° (SD8.7°) and this was not significantly different to those with BMI >35, mean 101° (SD11.4°).

This study did not find any correlation between the tibial slope and maximum flexion angle in 123 TKAs at five year follow up. Further studies with a more accurate measurement of tibial slope should be carried out to confirm whether a relationship exists in the clinical setting.

Non-invasive assessment of lower limb mechanical alignment and assessment of knee laxity using navigation technology is now possible during knee flexion owing to recent software developments. We report a comparison of this new technology with a validated commercially available invasive navigation system.

We tested cadaveric lower limbs (n=12) with a commercial invasive navigation system against the non-invasive system. Mechanical femorotibial angle (MFTA) was measured with no stress, then with 15Nm of varus and valgus moment. MFTA was recorded at 10° intervals from full knee extension to 90° flexion. The investigator was blinded to all MFTA measurements. Repeatability coefficient was calculated to reflect each system's level of precision, and agreement between the systems; 3° was chosen as the upper limit of precision and agreement when measuring MFTA in the clinical setting based on current literature.

Precision of the invasive system was superior and acceptable in all conditions of stress throughout flexion (repeatability coefficient <2°). Precision of the non-invasive system was acceptable from extension until 60° flexion (repeatability coefficient <3°), beyond which precision was unacceptable. Agreement between invasive and non-invasive systems was within 1.7° from extension to 50° flexion when measuring MFTA with no varus / valgus applied. When applying varus / valgus stress agreement between the systems was acceptable from full extension to 20° & 30° knee flexion respectively (repeatability coefficient <3°). Beyond this the systems did not demonstrate sufficient agreement.

These results indicate that the non-invasive system can provide reliable quantitative data on MFTA and laxity in the range relevant to knee examination.

Conventional computer navigation systems using bone fixation have been validated in measuring anteroposterior (AP) translation of the tibia. Recent developments in non-invasive skin-mounted systems may allow quantification of AP laxity in the out-patient setting.

We tested cadaveric lower limbs (n=12) with a commercial image free navigation system using passive trackers secured by bone screws. We then tested a non-invasive fabric-strap system. The lower limb was secured at 10° intervals from 0° to 60° knee flexion and 100N of force applied perpendicular to the tibial tuberosity using a secured dynamometer. Repeatability coefficient was calculated both to reflect precision within each system, and demonstrate agreement between the two systems at each flexion interval. An acceptable repeatability coefficient of ≤3mm was set based on diagnostic criteria for ACL insufficiency when using other mechanical devices to measure AP tibial translation.

Precision within the individual invasive and non-invasive systems measuring AP translation of the tibia was acceptable throughout the range of flexion tested (repeatability coefficient ≤1.6mm). Agreement between the two systems was acceptable when measuring AP laxity between full extension and 40° knee flexion (repeatability coefficient ≤2.1mm). Beyond 40° of flexion, agreement between the systems was unacceptable (repeatability coefficient >3mm).

These results indicate that from full knee extension to 40° flexion, non-invasive navigation-based quantification of AP tibial translation is as accurate as the standard invasive system, particularly in the clinically and functionally important range of 20° to 30° knee flexion. This could be useful in diagnosis and post-operative follow-up of ACL pathology.

Unicondylar knee arthroplasty (UKA) is a treatment for osteoarthritis when the disease only affects one compartment of the knee joint. The popularity in UKA grew in the 1980s but due to high revision rates the usage decreased. A high incidence of implant malalignment has been reported when using manual instrumentation. Recent developments include surgical robotics systems with navigation which have the potential to improve the accuracy and precision of UKA.

UKA was carried out using an imageless navigation system – the Navio Precision Freehand Sculpting system (Blue Belt Technologies, Pittsburgh, USA) with a medical Uni Knee Tornier implant (Tornier, Montbonnot Saint Martin, France) on nine fresh frozen cadaveric lower limbs (8 males, 1 females, mean age 71.7 (SD 13.3)). Two users (consultant orthopaedic surgeon and post doctoral research associate) who had been trained on the system prior to the cadaveric study carried out 4 and 5 implants respectively. The aim of this study was to quantify the differences between the planned and achieved cuts.

A 3D image of the ‘actual’ implant position was overlaid on the planned implant image. The errors between the ‘actual’ and the planned implant placement were calculated in three planes and the three rotations. The maximum femoral implant rotational error was 3.7° with a maximum RMS angular error of 2°. The maximum femoral implant translational error was 2.6mm and the RMS translational error across all directions was up to 1.1mm. The maximum tibial implant rotational error was 4.1° with a maximum RMS angular error was 2.6°. The maximum translational error was 2.7mm and the RMS translational error across all directions was up to 2.0mm.

The results were comparable to those reported by other robotic assistive devices on the market for UKA. This technology still needs clinical assessment to confirm these promising results.

In arthritic knees with severe valgus deformity Total Knee Arthroplasty (TKA) can be performed through medial or lateral parapatellar approaches. Many orthopaedic surgeons are apprehensive of using the lateral parapatellar approach due to lack of familiarity and concerns about complications related to soft tissue coverage and vascularity of the patella and the overlying skin. However surgeons who use this approach report good outcomes and no added complications. The purpose of our study was to compare outcomes following TKA performed through a medial parapatellar approach with those performed through a lateral parapatellar approach in arthritic knees with severe valgus deformity.

We conducted a retrospective review of patients from two consultants using computer navigation for all their TKAs. All patients with severe valgus deformities (Ranawat 2 & 3 grades) operated on between January 2005 and December 2011 were included. 66 patients with 67 TKAs fulfilled the inclusion criteria. Patients were group by approach; Medial = 34TKAs (34 patients) or Lateral = 33 TKAs (32 patients). Details were collected from patients' records, AP hip-knee-ankle (HKA) radiographs and computer navigation files. Outcome measures included lateral release rates, post-operative range of knee movements, long leg mechanical alignment measurements, post-operative Oxford scores at six weeks and one year, patient satisfaction and any complications. Comparisons were made between groups using t-tests.

The total cohort had a mean age of 69 years [42–82] and mean BMI of 29 [19–46]. The two groups had comparable pre-operative Oxford scores (Medial 41[27–56], Lateral 44 [31–60]) and pre-operative valgus deformity measured on HKA radiographs (Medial 13° [10°–27.6°], Lateral 12° [6°–22°]). Three patients in the Medial group underwent intra-operative lateral patellar release to improve patellar tracking. Seven patients in the Lateral group had a lateral condyle osteotomy for soft tissue balancing (one bilateral). There was no statistically significant difference between groups at one year follow up for maximum flexion (Medial 100° [78°–122°], Lateral 100° [85°–125°], p=0.42), fixed flexion deformity (Medial 1.2° [0°–10°], Lateral 0.9° [0°–10°], p=0.31) or Oxford score (Medial 23 [12–37], Lateral 23 [16–41], p=0.49). Similarly there was no difference in the patient satisfaction rates between the two groups at one year follow up. However there was a statistically significant difference in the mean radiographic post-operative alignment angle measurement (Medial 1.8° valgus [4° varus to 10° valgus], Lateral 0.3° valgus [5° varus to 7° valgus], p=0.02). One patient in the Medial group had a revision to hinged knee prosthesis for post-operative instability. There was no wound breakdown or patellar avascular necrosis noted in either of the groups.

The lateral parapatellar approach resulted in slightly better valgus correction on radiographs taken six weeks post-operatively. We found no major complications in the Lateral parapatellar approach group. Specifically we did not encounter any difficulties in closing the deep soft tissue envelope around the knee and there were no cases of patellar avascular necrosis or skin necrosis. Hence we conclude that lateral parapatellar approach is a safe and reliable alternative to the medial parapatellar approach for correction of severe valgus deformity in TKA.

The amount of time spent in theatre by trainees is decreasing and therefore it seems crucial to fully optimis e these to enable adequate training. Trainees at the beginning of their practice, despite their exposure to surgery, cannot always take advantages of the surgical procedure they are assisting with. An obvious example of this is total hip replacement during posterior approach. Although the posterior approach and less invasive or minimally invasive approaches are certainly beneficial for patients, they are very difficult for a young trainee to comprehend, as they spend most of the time hanging onto the retractor without or rarely seeing the important anatomic steps of the procedure. Our goal was to develop a tool that would help a trainee to fully see and understand the surgical steps of total hip replacement during a posterior approach.

To enable visualisation of the operation from the senior surgeon's perspective we developed a device to film the surgery and output the video feed to a screen. The prototype used an HD Replay XD1080 camera connected to a WDHI Xenta transmitting dongle (transmitting frequency −5.8 GHz), with an onboard 6600 mAh external Li-Mh battery providing 1A of current to the system. The Replay camera was fixed to the surgeon's ventilation helmet, and took its power from the battery supplying both the fan system and the transmitting unit. The surgeon can then clip both of these items to his belt and the connecting wires and cables run up his back. The device provided a Full HD video output of the surgery from the surgeon's perspective. The receiving unit used a Xenta WHDI wireless receiver with HDMI and DVI-I/D connections allowing the video to be displayed on any screen in the operating room with these connections.

The prototype has been trialled by the senior author and was successful in allowing the direct surgeon's view of the procedure to be displayed on a screen in the theatre so that other staff involved in the operation could see it.

Although the use of virtual training, presentations and video are essential to training, surgical training still relies greatly upon surgical assistance. The introduction of an intra-operative video feedback device would enable trainees to observe the operation from a first-person perspective which could lead to a considerable reduction in the amount of training time required, as well as a better understand of the specific surgical steps in a procedure. This would be particularly use for operations where a trainee assists the surgeon from the opposite side of the operating table, for example when undergoing total hip replacement during posterior approach. We can also envision this device also being used by surgeons to monitor their trainees when operating, and perhaps to keep a record of the operations undertaken in an establishment for archiving or assessment.

Distal femur resection for correction of flexion contractures in total knee arthroplasty (TKA) can lead to joint line elevation, abnormal knee kinematics and patellofemoral problems. The aim of this retrospective study was to establish the contribution of soft tissue releases and bony cuts in the change in maximum knee extension in TKA.

Data were available for 209 navigated TKAs performed by a single surgeon using a medial approach. All patients had the same cemented implant, either CR or PS, which both required a minimum thickness of 10 mm for the tibial and 9mm for the femoral component. Intra-operatively pre- and post-implant extension angles and the size of bone resection were collected using a commercial navigation system. The thickness of polyethylene insert and the extent of soft tissue release performed (no release, moderate and extensive release) were collected from the patient record. A univariate linear regression model was used to predict change in maximum extension from pre- to post-implant.

The mean bone resection was 19mm (15 to 28 mm) (Figure 1).79% of polyethylene inserts were 10mm thick (10 to 16 mm). 71% of knees had no soft tissue release. The mean increase in extension was 5° (11° decrease to 23° increase) (Figure 1). The analysis showed that bone cuts (p<0.001), soft tissue release (p=0.001) and insert thickness (p=0.010) were all significant terms in the model (r²_adj=0.170). This model predicted that carrying out a TKA with 19mm bone cuts, 10mm insert and no soft tissue release would give 4.2° increase in extension. It predicted that a moderate release would give a 2.8° increase in extension compared to no release, with an extensive release giving 3.9° increase over no release. For each mm increase in bone cuts the model predicted a 0.8° increase in extension and for each mm increase in insert size a decrease extension by 1.1°.

Preoperative FFC contracture is a frequent condition in TKA that the surgeon has to address either by resecting more bone or by extending soft tissue release to increase the extension gap to fit the knee implant. This analysis of 209 navigated knee arthroplasty showed that both options are suitable to increase the extension gap. The modelling results show that in general to increase maximum extension by the same as an extensive soft tissue release that bone cuts would have to be increased by 4–5mm. However this model only accounted for 17% of the variation in change in extension pre- to post-implant so is poor at predicting outcomes for specific patients. The large variation in actual FFC correction indicates that this relies on factors other than bone cuts and soft tissue releases as quantified in this study.

Introduction

The success of total hip replacement (THR) is closely linked to the positioning of the acetabular component. Malalignment increases rates of dislocation, impingement, acetabular migration, pelvic osteolysis, leg length discrepancy and polyethylene wear. Many surgeons orientate the cup to inherent anatomy of the acetabulum. Detailed understanding of the anatomy and orientation of the acetabulum in arthritic hips is therefore very important. The aim of this study was to describe the anteversion and inclination of the inherent acetabulum in arthritic hips and to identify the number that fall out with the ‘safe zone’ of acetabular position described by Lewinnek et al. (anteversion 15°±10°; inclination 40°±10°).

The aim of the study was to investigate rotational behaviour of the arthritic knee before (preimplant) and after (postimplant) total knee replacement (TKR) using (image-free navigation system as a measurement tool which recorded the axial plane alignment between femur and tibia, in addition to the coronal and sagittal alignment as the knee is flexed through the range of motion. The data on the rotation of the arthritic knee was collected after the knee exposure and registration of the lower limb (preimplant data). The position of rotation between the femur and tibia was recorded in 30° flexion, 45°, 60°, 90° and maximum degrees of flexion of the knee. The data was divided into subsets of varus and valgus knees and these were analysed pre and postimplant for their rotational position using SPSS for statistics.

The system was used in 117 knees of which 91 had full data set available (43 male 48 female). These included 71 varus knees, 16 valgus knees and 4 neutral knees to start in extension. Preimplant data analysis revealed there is tendency for the arthritic knees to first go in internal rotation in the initial part of flexion to 30 degrees and then the rotation is reversed back. This happens irrespective of the initial starting rotational relationship between femur and tibia in full extension. This happens in both varus as well as valgus arthritic knees. This trend of internal rotation in this initial part of flexion is followed in TKR as well implanted with fixed bearing CR knees irrespective of the preoperative deformity. Also noteworthy was the difference in rotation at 30°, 60° and 90 degrees of flexion between preimplant and postimplant knees (irrespective of varus and valgus groups).

When calculated at different points of flexion, there was statistically significant difference in the change of rotation at each point of flexion except 45 degree of flexion. The pre-operative values of change in rotation (internal being positive) at each step from the extended position being 5.4° (SD 4.5°) at 30 ° flexion, 4.7°(5.2°) at 45°, 3.6°(6.1°) at 60°, 3.5°(7.2°) at 90° and 4.2°(8.3°) at maximum flexion. Corresponding post-operative rotations were 2.2°(4.8°), 4.1°(6.4°), 6.6°(7.3°), 9.9°(8.8°) and 7.7°(8.9°). There was also an increase in the total range of rotation that the knee goes through after it has been implanted with prosthesis although it may not happen in every knee. This is statistically significant (p value <0.001) and seems more so in valgus group. The rotational movements and interrelationship of the femur and tibia is a complex issue, especially in the arthritic knees. Preimplant arthritic knee behaved generally similarly to normal knees according to the literature. Normal gait pattern demonstrates that the tibia moved through a 4° to 8° arc of internal rotation relative to the femur. The overall range (10.2° =/−4.2°) of knee rotation in this study greater than 8° might be explained by preimplant data acquired after the knee was approached and therefore releasing knee soft tissue envelop. This study confirmed that during the first 30° both varus and valgus knees moved internally. In our study there is increased range of total rotation postimplant (14° =/−6.8°) which may be explained by the fact that the anterior cruciate ligament is lost in all the TKRs and the posterior cruciate ligament may be dysfunctional as well. Thus the constraints on the knee rotation are decreased postimplant leading to increased rotation. We found some difference between varus and valgus post implant knees in that internal rotation seen in initial 30 degrees of flexion is much more pronounced in valgus knees as compared to varus knees (p value <0.001). This study confirmed knee internal rotation in initial stages of flexion, preimplant in arthritic knees during a passive knee flexion assessment. Varus and valgus knee seemed to behave similarly. This mimics the normal knee rotation. Postimplant knees in TKR behave differently.

Clinical laxity tests are frequently used for assessing knee ligament injuries and for soft tissue balancing in total knee arthroplasty (TKA). Current routine methods are highly subjective with respect to examination technique, magnitude of clinician-applied load and assessment of joint displacement. Alignment measurements generated by computer-assisted technology have led to the development of quantitative TKA soft tissue balancing algorithms. However to make the algorithms applicable in practice requires the standardisation of several parameters: knee flexion angle should be maintained to minimise the potential positional variation in ligament restraining properties; hand positioning of the examining clinician should correspond to a measured lever arm, defined as the perpendicular distance of the applied force from the rotational knee centre; accurate measurement of force applied is required to calculate the moment applied to the knee joint; resultant displacement of the knee should be quantified.

The primary aim of this study was to determine whether different clinicians could reliably assess coronal knee laxity with a standardised protocol that controlled these variables. Furthermore, a secondary question was to examine if the experience of the clinician makes a difference. We hypothesised that standardisation would result in a narrow range of laxity measurements obtained by different clinicians.

Six consultant orthopaedic surgeons, six orthopaedic trainees and six physiotherapists were instructed to assess the coronal laxity of the right knee of a healthy volunteer. Points were marked over the femoral epicondyles and the malleoli to indicate hand positioning and give a constant moment arm. The non-invasive adaptation of a commercially available image-free navigation system enabled real-time measurement of coronal and sagittal mechanical femorotibial (MFT) angles. This has been previously validated to an accuracy of ±1°. Collateral knee laxity was defined as the amount of angular displacement during a stress manoeuvre. Participants were instructed to maintain the knee joint in 2° of flexion whilst performing a varus-valgus stress test using what they perceived as an acceptable load. They were blinded to the coronal MFT angle measurements. A hand-held force application device (FAD) was then employed to allow the clinicians to apply a moment of 18Nm. This level was based on previous work to determine a suitable subject tolerance limit. They were instructed to repeat the test using the device in the palm of their right hand and to apply the force until the visual display and an auditory alarm indicated that the target had been reached. The FAD was then removed and participants were asked to repeat the clinical varus-valgus stress test, but to try and apply the same amount of force as they had been doing with the device.

Maximum MFT angular deviation was automatically recorded for each stress test and the maximum moment applied was recorded for each of the tests using the FAD. Means and standard deviations (SD) were used to compare different clinicians under the same conditions. Paired t-tests were used to measure the change in practice of groups of clinicians before, during and after use of the FAD for both varus and valgus stress tests.

All three groups of clinicians initially produced measurements of valgus laxity with consistent mean values (1.5° for physiotherapists, 1.8° for consultants and 1.6° for trainees) and standard deviations (<1°). For varus, mean values were consistent (5.9° for physiotherapists, 5.0° for consultants and 5.4° for trainees) but standard deviations were larger (0.9° to 1.6°). When using the FAD, the standard deviations remained low for all groups for both varus and valgus laxity. Introducing the FAD overall produced a significantly greater angulation in valgus (2.4° compared to 1.6°, p<0.001) but not varus (p = 0.67) when compared to the initial examination. In attempting to reach the target moment of 18Nm, the mean ‘overshoot’ was 0.9Nm for both varus and valgus tests. Standard deviations for varus laxity were lower for all groups following use of the FAD. The consultants' performance remained consistent and valgus assessment remained consistent for all groups. The only statistically significant change in practice for a group before and after use of the FAD was for the trainees testing valgus, who may have been trained to push harder (p = 0.01). Standardising the applied moment indicated that usually a lower force is applied during valgus stress testing than varus. This was re-enforced by clinicians, one third of whom commented that they felt they had to push harder for valgus than varus, despite the FAD target being the same.

We have successfully standardised the manual technique of coronal knee laxity assessment by controlling the subjective variables. The results support the hypothesis of producing a narrow range of laxity measurements but with valgus laxity appearing more consistent than varus. The incorporation of a FAD into assessment of coronal knee laxity did not affect the clinicians' ability to produce reliable and repeatable measurements, despite removing the manual perception of laxity. The FAD also provided additional information about the actual moment applied. This information may have a role in improving the balancing techniques of TKA and the management of collateral ligament injuries with regard initial diagnosis and grading as well as rehabilitation.

Finally, the results suggest that following use of the FAD, more experienced clinicians returned to applying their usual manual force, while trainees appeared to use this augmented feedback to adapt their technique. Therefore this technique could be a way to harness the experience of senior clinicians and use it to enhance the perceptive skills of more junior trainees who do not have the benefit of this knowledge.

There is increasing interest in the use of image free computer assisted surgery (CAS) in total hip arthroplasty (THA). Many of these systems require the registration of the Anterior Pelvic Plane (APP) via the bony landmarks of the anterior superior iliac spines (ASIS) and pubic tubercles (PT) in order to accurately orient the acetabular cup in terms of anteversion and inclination. Given system accuracies are within 1mm and 1° and clinical validation studies have given accuracy by cup position. However, clinical outcomes contain not only system inaccuracies but also variations due to clinical practice. To understand the effects of variation in landmark acquisition on the identification of the acetabular cup orientation, independent bench testing is required. This requires a phantom model that can represent the range of pelvises, male and female, encountered during THA and introduce deliberate known errors to the acquisition to see the effect on anteversion and inclination angles. However, there is a paucity of information in the literature with regards to these specific pelvic dimensions (pelvic width and height). Therefore the aims of this work were to generate the normal expected range of sizes of the APP for both males and females and to use these to manufacture a phantom model that could be used to assess CT free navigation systems.

In the first part of the study 35 human cadavers and 100 pelvic computed tomography (CT) scans were examined.

All cadavers had no gross pelvic abnormalities or previous surgeries. Measurements were carried out with cadavers placed in a supine position. The first author made three sets of measurements using a millimeter ruler. Solid steel pins were used to identify the palpated ASISs and PTs. String was tied between the two ASIS pins and the pelvic width measured. The midpoint of the pubic tubercles was taken to be the midpoint of the pubic symphysis. Pelvic height was measured from the midpoint of the ASIS distance (marked on the string) to the midpoint of the PTs. One hundred pelvic CT scans with no bony abnormalities, previous surgery or metal prosthesis (due to artefacts) were obtained retrospectively from the hospital radiological online system (PACS, Kodak). Mimics software (Mimics12 Materialise, Leuven, Belgium) was used to automatically reconstruct three-dimensional (3D) models using the ‘Bone’ thresholding function. This eliminated any soft tissue from the 3D models. The most anterior ASIS and PT points were then identified on the 3D model surface and measurements of distances made. As the software did not allow identification of points not on the model surface it was not possible to directly obtain the midpoint of the ASIS distance. Therefore to obtain the pelvic height measurements the distance between each ASIS and the ipsilateral and contralateral PTs was also measured. The pelvic height was then calculated using trigonometric functions. The ratio of width to height was calculated (ratio > 1 indicating pelvis width greater than pelvis height). Student's t test was used analyse any differences between male and female pelvic measurements with a p<0.05 being statistically significant.

Using the results from above an aluminium pelvic phantom model was designed and manufactured. It was machined from a billet of marine grade aluminium alloy using a vertical computer numerical controlled (CNC) milling machine. The top surface represented the APP and sides (which represented the acetabuli) were angled to give anteversion and inclination angles of 20° and 45° respectively. Co-ordinates for ASIS and PT points were given based on the 99% prediction intervals from the pelvic data and additional points were milled to give up to a 20 mm error mediolaterally and also in height. Each co-ordinate point was drilled with a 2.0mm diameter ball-nose cutter to a depth of 1.0mm, these holes designed to accommodate the ball-nosed pointer tip to ensure it remained at the same position in space at all orientations of the pointer. Further to this, known errors in height were introduced using accurately manufactured blocks with similar points milled on the surface to fit a ball-nosed pointer. These blocks could be secured to the top surface of the model using screws. A Perspex base unit with tracker attachments was made to hold the phantom and provide the reference frame. A further support that enables the phantom to also be used in the “lateral” position was manufactured.

For the assessment of pelvic size there were 66 females and 69 males, mean age 62.3 years (range from 20 to 99 years). The mean width was 238 mm (SD 20 mm) and mean height was 93 mm (SD 11 mm) with a mean ratio of 2.6 (SD 0.3). There were no statistically significant differences in mean between males and females (p>0.4 in all cases). From this data set the range of APP sizes required to cover 99% of population (width 186 to 290 mm and height 66 to 120 mm) and therefore the measurements for the model were generated. The manufactured model can be used to give the range of pelvis sizes from 170mm to 290mm in width and 60mm to 120mm in height and also to add up to 20 mm of error in palpation of each of the ASISs and PT.

This study generated APP sizes to cover 99% of the general population over a wide age range. It illustrated that a single pelvic model would fit both sexes. The model allows the determination of the effects of changes of the pelvic dimensions may have on the acetabular orientation measured on an image free CAS system including the assessment of point acquisition and deliberate errors. The model has been successfully used in preliminary testing and can be used to assess any CT free system.

Knee alignment is a fundamental measurement in the assessment, monitoring and surgical management of patients with osteoarthritis [OA]. In spite of extensive research into the consequences of malalignment, our understanding of static tibiofemoral alignment remains poor with discrepancies in the reported weight-bearing characteristics of the knee joint and there is a lack of data regarding the potential variation between supine and standing (functional) conditions. In total knee arthroplasty [TKA] the lower limb alignment is usually measured in a supine condition and decisions on prosthesis placement made on this. An improved understanding of the relationship between supine and weight-bearing conditions may lead to a reassessment of current surgical goals.

The purpose of this study was to explore the relationship between supine and standing lower limb alignment in asymptomatic, osteoarthritic and prosthetic knees. Our hypothesis was that the change in alignment of these three groups would be different.

A non-invasive infrared position capture system (accuracy ±1° in both coronal and sagittal plane) was used to assess the knee alignment for 30 asymptomatic controls and 31 patients with OA, both before and after TKA. Coronal and sagittal mechanical femorotibial (MFT) angles in extension (negative values indicating varus in the coronal plane and hyperextension in the sagittal plane) were measured with each subject supine and in bi-pedal stance. For the supine test, the lower limb was supported at the heel and the subject told to relax. For the standing position subjects were asked to assume their normal stance. The change in alignment between these two conditions was analysed using a paired t-test for both coronal and sagittal planes. To quantify the change in 3D, vector plots of ankle centre displacement relative to the knee centre from the supine to standing condition were produced.

Alignment in both planes changed significantly from supine to standing for all three groups. For the coronal plane the supine and standing measurements (in degrees, mean(SD)) were 0.1(2.5) and −1.1(3.7) in the asymptomatic group, −2.5(5.7) and −3.6(6) in the OA group and −0.7(1.4) and −2.5(2) in the TKA group. For the sagittal plane the numbers were −1.7(3.3) and −5.5(4.9); 7.7(7.1) and 1.8(7.7); 6.8(5.1) and 1.4((7.6) respectively. This change was most frequently towards relative varus and extension. Vector plots showed that the trend of relative varus and extension in stance was similar in overall magnitude and direction between the three groups.

Knee alignment can change from supine to standing for asymptomatic and osteoarthritic knees, most frequently towards relative varus and hyperextension. The similarities between each group did not support our hypothesis. The consistent kinematic pattern for different knee types suggests that soft tissue restraints rather than underlying joint deformity may be more influential in dynamic control of alignment from lying to standing. In spite of some evidence suggesting a difference between supine and standing knee alignment a mechanical femorotibial (MFT) angle of 0° is a common intra-operative target as well as the desired post-operative weight-bearing alignment. These results indicated that arthroplasties positioned in varus intra-operatively could potentially become ‘outliers’ (>3° varus) when measured weight-bearing. Mild flexion contractures may correct when standing, reducing the need for intra-operative posterior release. These potential changes should be considered when positioning TKA components on supine limbs as post-operative functional alignment may be different.

Computer assisted surgery is becoming more frequently used in the medical world. Navigation of surgical instruments and implants plays an important role in this surgery. OrthoPilot™ Hip Suite (BBraun Aesculap) is one such system used for hip navigation in orthopaedic surgery. However the accuracy of this system remains to be determined independently of the manufacturer. The manufacturer supplies a technical specification for the accuracy of the system (± 2 mm and ± 2°) and previous research has been undertaken to compare its clinical accuracy against conventional hip replacements by x-ray. This clinical validation is important but contains many sources of error or deviation from an ideal outcome in terms of the surgeons' use of the system, inaccurate palpation of landmarks, variation in actual cup position from that given by the navigation system and measurement of the final cup position. It is therefore not possible to validate the claims of the manufacturer from this data. There is no literature evaluating the technical accuracy of the software i.e. the accuracy of the system given known inputs. This study had two main aims 1) validating the accuracy of the OrthoPilot data while navigating the surgical instruments and 2) validating the accuracy of navigation algorithm inside the OrthoPilot system which determines cup implant placement. The OrthoPilot validation was performed and compared against the gold standard of a VICON movement analysis system.

The system used was OrthoPilot™ with a Spectra camera from Northern Digital Inc. (Ontario, Canada). Software investigated was the Hip Suite THA cup only navigation software Version 3.1. The validation was performed and compared against the VICON Nexus version 1.4.116 with Bodybuilder software version 3.55. An aluminium pelvis phantom was used for measurement allowing accurate and repeatable inputs. The OrthoPilot system has three types of instruments sets; passive, active and hybrid. This study was carried out with the passive instruments set. Data were captured simultaneously from both the OrthoPilot and VICON systems for the supine position of the phantom. Distances between the anatomical land marks on the phantom were compared to test the data capturing accuracy of the OrthoPilot system. Anatomical land marks of right anterior superior iliac supine (RASIS), left anterior superior iliac supine (LASIS) and Pubic Symphasis (PS) were palpated to define the Anterior Pelvic Plane (APP). Distances between the anatomical landmarks of RASIS to LASIS, RASIS to PS and LASIS to PS were considered for comparison. Width and height of the pelvis was varied to examine different APPs. The width and height used were 170 mm and 53 mm, 230 mm and 88 mm, and 290 mm and 123 mm respectively. One hundred APP data sets were captured at each instance.

The accuracy of the hip navigation algorithm was tested by applying similar algorithm to calculate the native anteversion and inclination angles of the acetabulum using the VICON system. Data were captured simultaneously from both OrthoPilot and VICON systems. Radiographic anteversion and inclination angles were obtained with phantom model, which had 14° of anteversion angle and 45° of inclination angle. APP of 230 mm in width and 88 mm in height was used to obtain anterior pelvic plane data. Position vectors for each anatomical land mark from the OrthoPilot system were extracted from relevant transformation matrices, while position vectors from the VICON system were extracted from static trial modelling.

The distance data from both systems were compared with calibrated distance data from the phantom model. Mean values of the distances between anatomical landmarks were found to be similar for both OrthoPilot and VICON systems. In addition, these distances were comparable with the pelvic phantom model data, within 1 mm for all measured distances for the VICON and 2 mm for the OrthoPilot. Furthermore, the standard deviations were less than 1% of the measured value. Comparison was also made for the anteversion and inclination angles of the acetabulum of the pelvic model with OrthoPilot and VICON data. Both systems produced similar results for the mean angle values, within 0.5° of the known angles for the VICON and 1° for the OrthoPilot and with standard deviations of the measured values of less than 1%.

All the data were captured simultaneously from both OrthoPilot and VICON systems under the same laboratory conditions. According to the above results it is clear that the distance readings obtained from the OrthoPilot are comparable to the results obtained from the gold standard VICON system and the calibrated distance readings of the phantom. In addition, acetabular angle results obtained from OrthoPilot are almost equivalent to results obtained from VICON and the calibrated phantom angles. Finally it is can be concluded that, both the data palpation with OrthoPilot system and acetabular angle calculation algorithm of the OrthoPilot system are accurate enough for the real world clinical tasks they are expected to perform.

Over the past fifteen years, computer-assisted surgery systems have been more commonly used, especially in joint arthroplasty. They allow a greater accuracy and precision in surgical procedures and thus should improve outcomes and long term results.

New instruments such as guided handheld tools have been recently developed to ultimately eliminate the need for drilling/cutting or milling guides.

To make sure that the handheld tool cuts and/or drills in the desired plane, it has to be servo-controlled. For this purpose, the tool joints are actuated by computer-controlled motors. A tracking system gives the tool position and orientation and a computer calculates the corrections for the motors to keep the tool in the desired plane.

For this servo-control, a very fast tracking system would be necessary. It should be fast enough to follow human motion. Current optical tracking systems used for computer-assisted surgery have a bandwidth of about 10–60 Hz [3]. For servo-control, a bandwidth of about 200–300 Hz would be required to be faster than human reaction; the latency of the system should also be small, about 2–3 ms. Optical tracking systems with a higher bandwidth exist but are too expensive for applications in surgery; besides the latency – due to the complex computer vision treatment involved – is too big.

We have developed a hybrid tracking system consisting of two cameras pointed at the operating field and a sensor unit which can be attached to a handheld tool.

The sensor unit is made up of an inertial measuring unit (IMU) and numerous optical markers. The data from the IMU (three gyroscopes and three accelerometers placed such that their measurement axes are perpendicular to each other) and the marker images from the cameras looking at the optical markers are fed to a data fusion algorithm. This algorithm calculates the position and the orientation of any handheld tool. It can do so at the higher of the two sensor sample rates which is the IMU sample rate in our case.

Our experimental setup consists of an ADIS 16355 IMU which runs at a sample rate of 250 Hz and a pair of stereo cameras which are sampled at 16.7 Hz. The data collected from these sensors are processed offline by the data fusion algorithm. To compare the results of our hybrid system to those of a purely optical tracking system, we use only the marker image data to recalculate the sensor unit's position by triangulation.

The experiment we conducted was a fast motion in a horizontal direction starting from a rest position. The sensor unit position was calculated by the hybrid system and by the optical tracking system using the experimental data. The fast motion started right after the optical sample at t1 and the hybrid system detects it at once. The optical tracking system, on the other hand, only sees the motion at the next optical sample time t2.

These results show that our hybrid system is able to follow a fast motion of the sensor unit whereas a purely optical tracking system is not.

The proposed hybrid tracking system calculates position and orientation of any handheld tool at a high frequency of 250 Hz and thus makes it possible to servo-control the tool to keep it in the desired plane.

Several similar systems fusing optical and inertial data have been described in the literature. They all use processed optical data, i.e. 3D marker positions. Our algorithm uses raw image data to considerably reduce computation time. This hybrid tracking system can be used with any handheld tool developed to substitute existing drilling, cutting or milling instruments used in orthopaedic surgery and particularly in arthroplasty.

The sensor unit can be easily implemented into an existing optical tracking system. For the surgeon, the only change is an additional small inertial sensor besides the optical markers already attached to the tool.

The authors would like to thank the AXA Research Fund for funding G.C. Claasen's work with a doctoral grant and Guillaume Picard for his contributions to the experimental setup.