Disclaimer: This dissertation has been written by a student and is not an example of our professional work, which you can see examples of here.

Any opinions, findings, conclusions, or recommendations expressed in this dissertation are those of the authors and do not necessarily reflect the views of UKDiss.com.

Effectiveness of Robotics and Computer Assisted Navigation in Surgery for Knee and Hip Replacements

Info: 9882 words (40 pages) Dissertation
Published: 10th Dec 2019

Reference this

Tags: MedicalMedical Technology

Robotics and computer assisted navigation in surgery has developed over the last 25 years to provide greater accuracy to orthopaedic surgeons in total knee and hip replacements. These methods have overtaken the traditional method of instrumentation. There are numerous systems on the market being applied in a variety of clinical settings. Whilst the intraoperative clinical benefits have been widely studied and reported, these advancements have yet to provide clinical benefits post operatively. Patient reported outcomes are yet to show any statistically significant increase, and their economic benefits are dependent upon a myriad of factors. In the future, as surgeons gain a greater understanding of how to best achieve natural kinematics for patients, these robotic systems will be well placed to accurately deliver these targets and potentially increase long term clinical outcomes, patient reported outcomes and economic benefits.


Osteoarthritis is the degenerative loss of cartilage tissue in the joint. It is the most common joint disease in Australia, affecting approximately 15% of the population [1]. The signs and symptoms include pain and loss of function. For the past 40 years, the treatment of end-stage osteoarthritis of the knee and hip has been total joint replacement. The earliest incarnations of the total knee replacement were designed and introduced by innovative surgical working in collaboration with the orthopaedic industry [2] . These implants had a high failure rate, knees in particular, and the ratio of loosening of the implants was exceptionally high due to overshearing and rotation during gait [3]. The earliest incarnations of total hip replacements (THR) also had problems. Wagner and Freeman, pioneers in hip surgery in the 1970’s, used cementless femoral stems coated with hydroxyapatite to increase the contact forces between metal and bone. Unfortunately, follow-up at 6 years showed failure rates of 35-58%, with complications mostly concerning acetabular loosening [3]. During the 1980’s, advancements in the design and fixation of prostheses led to improved results in arthroplasty, but still high rates failure of the implants persisted. The reason for failure of the implants however, shifted from implant loosening to delamination of bearing surfaces [3]. With the advancement of better materials came better results, and patient revision rates dropped. Radiographic evidence of total knee replacements (TKR), in the form of long leg X-Rays showed realignment of the coronal and sagittal planes. All of these operations were still done with  manual tools and instruments, and a reliance on a visual inspection by the surgeon [4], and as Plaskos et al. demonstrated, “cutting guides combined with an oscillating saw resulted in errors in cuts ranging from 0.6 degrees to 1.1 degrees in varus-valgus and 1.8 degrees in flexion-extension” [2]. Mechanical alignment jigs are used to assist in making these cuts on the femur and tibia, referencing the long axis of the bone, either externally or internally. Other complex instruments have been designed to reference other landmarks such as the transepiconylar axis, posterior condylar axis and the anterior-posterior axis, otherwise known as Whiteside’s line. It has been suggested that errors in surgical technique may be the most common reason for failure of implants in knees [2] and that the importance of mechanical axis alignment for implant longevity has been well documented [5]. Studies from the Mayo registry have shown that there was a significant loss of implant survivorship when implants fell outside of this range [5]. The postoperative alignment of the knee has a large effect on the load transferred through the implant, and to spread the load evenly manufacturers of the implants have traditionally recommended positioning the implants in mechanical alignment. This mechanical axis is defined as “a straight line passing from the centre of the femoral head to the centre of the talus” [2]. For total hip replacements, one of the main issues with failure has been identified as improper acetabular cup placement, and has been associated with several complications, including dislocation, component impingement, leg length discrepancy, altered hip biomechanics, accelerated bearing surface wear and the need for revision surgery  [6]. Good alignment of acetabular components have several safe zones for inclination and anteversion, and these have been reported on by Lewinnek et al. and Callanan et al. [6]. These safe zones though are achieved with manual instrumentation. Navigation was introduced to surgery to obtain better accuracy in component alignment. By the start of the 1990’s, Computer-based motion tracking systems contributed to the easy access of surgical navigation systems for surgeons [4]. Companies such as Stryker, Zimmer and BrainLab developed systems that were utilised to aid surgeons in arthroplasty surgery to better accurately align prostheses. The purpose of the introduction of these new navigation systems was to achieve what conventional instrumentation could not, that being a greater degree of accuracy in component placement of the orthopaedic implants. The results in studies that followed were positive, with numerous papers demonstrating a higher amount of patients with their components closer to the mechanical axis than those receiving total joint replacements with standard instrumentation. These systems were able to assist surgeons in performing their osteotomies in both the coronal and sagittal planes, thereby ensuring a greater degree of accuracy in tibial and femoral placement. Not only were these navigational systems useful for standard joint replacement procedures, but also for extreme and sever deformity cases where conventional instrumentation is unable to be used [4]. It has been shown that “accurate reconstruction of leg alignment offers the best opportunity for achieving good long term results” [4]. Despite these better results, outcomes still need to improve in terms of patient satisfaction, especially in the case of younger patients. Patient satisfaction remains at only 82% to 89% after TKA” [2]. Robotics were introduced in the early 1990’s. Conditt et al. stated that “the primary value of robotics in joint replacement is the reduction of human error by improving accuracy and precision” [7]. The intended goal is not to replace the surgeon, but to improve the surgeon’s performance [7]. The first case involving a real patient was completed in 1992 by the Robodoc system. Justin Cobb first introduced robotic assistance in 2000 using the Acrobot robot to improve the accuracy of implant positioning during unicompartmental knee arthroplasty [8]. This was the first step to robotics being fully incorporated into the operating room (OR) at the turn of the millennium, with companies such as Mako and Praxiteles introducing their systems in the early to late 2000’s. These robotic systems were seen as the next step in the integration of technology into the field of orthopaedic joint replacement surgery.  They offered not only a more accurate navigational system, but also had the ability to load patient images in the form of computed tomography (CT) scans, so that the collated intra-operative data points can be reconciled with pre-loaded images. They also have haptic control capabilities, allowing the surgeon to restrict themselves in the cuts they make, thereby reducing the amount of surgical error in the procedure, and also to conserve a greater amount of bone by being unable to resect any more than is necessary. It is this greater amount of precision and reproducibility that separates robotic systems from the navigational systems. This has correlated to an increase in patient satisfaction in the first year of surgery, and a reduction in readmission rates in the first 90 days after the operation date. Both of these facts are significant when considering the role robotics has played in total joint replacements as these are two of the key metrics that surgeons and hospitals are respectively marked on.  The long term benefits are yet to be reported as the technology is still in its infancy, and the correlation between long term patient satisfaction, long leg alignment and revision rates are yet to be proven. This review investigates the historical progression of computer navigated and robotic surgery from primitive beginnings to highly complex automated systems capable of real time haptic feedback. Furthermore, analysis of the economic impact of these systems is presented to explain the current pervasion within the orthopaedic community.

Early Innovation

The introduction of robots in health care aims to overcome human limits and improve accuracy and reliability [9]. Robotic technology in orthopaedics was first introduced in the late 1980’s, with the first case being completed in 1992. The first case was a total hip replacement, but robotic systems developed to be used for total knee replacements, and cruciate ligament surgery. There are several robotic systems currently available in the market place. They all have a generic platform off which they operate, while possessing their own unique capabilities which differentiate themselves from one another in the market. The tools and instruments used in robotics are calibrated to ensure a high degree of accuracy. Calibration can be defined as “a procedure that matches geometric computer models of implants or tools to the coordinates of the position of the actual implants or tools” [10]. The implants and tools are then registered, which is defined as “a computational procedure that matches preoperative images or planning information to the position of the patient on the OR table” [10]. There are three registration methods. The first is paired point matching registration where the surgeon must identify three or more points on the preoperative images to the corresponding points on the patient during surgery [10]. The issue with this method is that if anatomical landmarks are used for matching points then it may not be reproducible. To overcome this, fiducial markers are placed on target bones before volumetric images, such as CT’s and MRI’s, are taken. The fiducials create a 3D reference for the patient’s anatomy, and the intraoperative location of these markers are referenced back to their location in the preoperative plan. This type of registration does require an extra operation, but is accurate, and is used by the ROBODOC system. The second registration method is shape-based, or surface, registration. This method does not require fiducial markers, but rather the point on the computer model, nearest the measured surface point, is designated as the corresponding point, and the calculation is repeated to reduce the average distance between each point and the corresponding surface point [10]. The mathematical calculations used are iterative closest-point algorithms and the least-squares method. To avoid local minima, the baseline registration is performed using a paired point method. This is done to obtain the starting position for registration [10]. Surface points are then used for final matching, for which the calculations are repeated until it becomes saturated [10]. Using this technique, intraoperative surface data points are reconciled with preoperative imaging data points. ROBODOC adopted this method in 1999 and the results showed that this method of registration has been reported as accurate as the fiducial-based registration method [10].  The third method of registration is 2D-3D or 3D-3D registration using fluoroscopic images. Although 2D-3D has been shown to be accurate in a laboratory setting for robotic-assisted total hip arthroplasty, the robotic applications of this registration method are yet to be used clinically [10].  A patient-specific digital image is formed and serves as a map. The systems use 3 types of digital maps. The first is a preoperatively imaged systems, which require anatomic information collected before the operation in the form of a CT scan or MRI, adapted to the anatomic site and the use of point-to-point registration. The second is a perioperatively imaged system, where anatomic imaging occurs in the operating suite at the time of surgery [11]. During fluoroscopic-based navigation, images are uploaded into the system directly from a navigated C-arm, an imaging device that can be used with flexibility in the OR. For this method, manoeuvring is required for a simultaneous representation of the position of the tracked surgical instruments in relation to the patient’s anatomy, and in different image planes during the operation [11]. The final category is image free, where an anatomic model is embedded in the software and upgraded by the registration process [11].


Robotic systems can be classified into three groups:

  1. Passive
  2. Active
  3. Semi-active

Passive robotic systems are systems that perform part of the surgical procedure under continuous and direct control of the surgeon during the operation [12]. The passive system performs no direct action on the patient, but uses 3-dimensional (3D) position sensors to track the target bone and surgical tools to provide a visual representation and facilitate surgical planning [13]. The robotic system will hold jigs and guides in the correct position. The cutting or drilling is then performed by the surgeon. Examples of this type of system include Stryker navigation and Zimmer Orthosoft. Active, or autonomous, robotic systems are computer-controlled systems that develop a preoperative surgical plan with surgeon supervision [13]. They then perform a surgical task without the direct intervention of the surgeon, such as allowing the robotic arm to cut the bone without direct manipulation of the cutter by the surgeon [2]. The systems not only hold the cutting tool, but they also make the appropriate cuts. Examples of this type of system include the ROBODOC system, the CASPAR system and the BRIGIT system. The surgeon is involved in the planning, and the robotic system executes the plan. This is done under close examination of the surgeon, who usually has an emergency button to disable the system if needed. There are 4 generic steps within this type of system:

  1. The placement of fiducial marker pins
  2. CT scanning
  3. Pre-operative planning with virtual implantation
  4. Actual surgical implantation

The first active robotic system was the ROBODOC system, an example of the apparatus is shown in Figure 1. The concept was to bring computer-aided design and computer-aided manufacturing (CAD/CAM) into the operating room for the accurate placement of femoral implants [10]. It was designed to improve outcomes in cementless total hip arthroplasty by reducing technical errors. This robotic-assisted total hip replacement system consists of locator pin implantation, CT scans, preoperative planning using the workstation, robot diagnostics and preparation, exposure and registration of pins, and robotic milling of the femur. The ROBODOC uses a helical computed tomography (CT)-based preoperative planning system performed using ORTHODOC to create a 3D image of the femur and tibia [13]. The femur is registered with the implantation of two locator pins. One is placed in the greater trochanter, and the other is placed in the femoral lateral condyle. Once these pins are in place, a CT scan is performed. The protocol requires 1mm thick slices, a 1-6mm scan interval and a 200mm field of view. In total, 200 slices are acquired, and these are transferred to a 3D computer modelling station called ORTHODOC [14]. With the CT image in the system, the surgeon can then evaluate the 3D computer model of the femur and apply various implant configurations/ geo-metry to determine the optimal implant type, position of the implant, and center of rotation of the THA, whilst also accounting for leg length and lateral offset [14] . The surgeon can also see whether damage to surrounding structures and soft tissue may occur, as the system has the ability to display the cutting paths three-dimensionally. Construction of the images and planning takes approximately thirty minutes. Once the surgeon is satisfied with the plan, they can then load the plan into the ROBODOC system and begin the operation. The surgeon registers the robot with the patient by guiding its probe into contact with the pins. The pin location is then recorded and verified against the data loaded into the system. The femur is then fixed to the robot base by a special clamp and a bone motion sensor/monitor is activated, then the bone is registered by moving the robotic manipulator to each fiducial marker and anatomical landmark [14]. The surgeon then attaches a cutting bur and guides the robotic arm into the surgical field and begins milling the femoral cavity. Once completed, the femoral stem can be inserted. It is easy to ascertain the osteotomy level of the femoral neck as the milling will show the neck cut by making a notch in the medial cortex of the proximal femur. The height of the stem is measured and differences from the plan are known to the surgeon.  In much the same way as computer navigation, the CASPAR system utilises fiducial markers that are fixated to the femur and tibia to obtain “spatial orientation and geometrical calculations” [4], see Figure 2. Once the pins are in place, a spiral CT is performed, and the femoral head, knee joint, ankle, and pins are scanned whilst the patient is anaesthetised.  Once the scanned is completed the virtual surgery can occur on the systems screen through the section of specific implant sizes. The implant can then be adjusted in the coronal, sagittal and axial planes so that alignment and rotation positions are satisfactory. Once this has been achieved, the milling area can be specified and adapted in order to avoid milling into areas such as soft tissue regions [4]. This is done to reduce soft tissue trauma. All data is saved and transferred to the robotic unit. A standard medial parapatellar incisional approach is made and the leg is flexed and rigidly attached to the robot through two Steinman pins running through the distal femur and proximal tibia. The pins themselves are connected to a frame which is in turn linked to the robot. Reflective markers are attached to the frame to detect and control undesired movements. They are continuously monitored by an infrared camera, which would stop the robot once undesired motion was detected. After verification of the fiducial markers, the surgeon commences the robotic action. A milling cutter is used by the robot to resect the bone. Constant water irrigation is used to cool the cutter, as well as wash away debris. The surgeon maintains control over the whole process through a manually held button, which when pressed, shuts down the robotic system. They can at this point switch to a conventional manually instrumented technique.  Semi-active systems, sometimes referred to as haptic robots, are a tactile feedback system that augments the surgeon’s ability to control the tool. These robotic systems differentiate themselves from computer-assisted surgery programs by their ability to restrain the  movement of the surgeon with the operative tool [13], typically by restricting the cut volume through the definition of constraining boundaries of the cut motion [2]. Haptics can be described as “the science of applying touch (tactile) sensation and control to interaction with computer applications” [15]. In this instance, the robot guides the cutting tool within a predetermined anatomical plane. The surgeon can then operate under the constraints the robot has provided. They combine intraoperative versatility and adaptability that navigation systems offer along with the accuracy of bone preparation and cutting. During bone resection, the surgeon is directed by the system, which steers and controls their actions based upon the pre-operative plan. This is done through a haptic feedback mechanism, in which the surgeon is prevented from resecting bone from outside the pre-operative plan. Examples of this system include the MAKO system, the ACROBOT system and the Praxiteles system. The MAKO RIO system from Stryker is a closed implant platform for medial and lateral unicompartmental arthroplasties, as well patellofemoral arthroplasty, and more recently, total knee and hip replacements, see Figure 3. This system uses a preoperative CT scan of the patient’s knee or hip to create a 3D virtual model of the patient’s anatomy [13],  allowing pre-operative planning, by aiding in the evaluation of the hip, knee and bone quality, as well as extra articular deformities. They also have the ability to adjust the plan intra-operatively based on the patient’s pathology and deformity. With this CT-based approach, the surgeon can set the implant position prior to surgery, so that they can minimize potential failures in the implants.  A pre-operative CT scan is taken of the patient, with 1 mm increment slices through the knee joint, and 5 mm increment slices taken through the hip and ankle. The scan is then saved in digital imaging and communications in medicine (DICOM) format and sent to the RIO system. Sagittal slices of the distal and proximal femur are segmented, defined and reconstructed into 3D models, so that implant templates can be placed on top of the model for a patient-specific plan. Cartilage cannot be visualized on a CT scan, and thus limits the pre-operative approach. Only bone alignment can be planned for, and intraoperative changes can be made to achieve long leg alignment and gap balancing. CT scans do assist in identifying cysts, subchondral defects and osteophytes, as well as areas of avascular necrosis. The pre-operative plan is based upon 4 areas:

  1. 3D visualization of implant position
  2. Component alignment
  3. Lower limb alignment
  4. Gap kinematics

To ensure the accurate positioning of the prosthesis, the implant computer assisted design (CAD) models are positioned over the 3D Models of the distal femur and proximal tibia.  The alignment parameters are then displayed on the screen. This allows the surgeon to visualize the predicted positioning of the implants and make changes accordingly to ensure congruency and to avoid edge loading.  Continuous feedback is supplied by the system on soft tissue and bone anatomy. The bone resection values are automatically defined and the boundaries for the cutting instrument are defined to prevent inadvertent harm. The initial plans are based upon alignment parameters, with changes made during surgery based on gap measurements through the range of motion and alignment values. These parameters are set and recommended by MAKO, and used in conjunction with parameters cited in literature.  An example of this is placing the tibial component in 2-4 degrees of varus and no greater than 7 degrees slope in patients with medial compartment osteoarthritis to reduce the risk of rupturing the anterior cruciate ligament, as found by Collier and colleagues [4]. The surgeon positions the robot to ensure maximum accessibility to the operative knee. Unlike other active and semi-active robot systems, the RIO system does not require rigid fixation of the robot to the patient. Osseous reference markers track the position of the tibia and the femur, and as the bones move during surgery, the haptic 3D resection volume move in sync. The surgeon will then calibrate the robotic arm before making the first cut [13]. The robotic arm is taken through by all axes of motion in a predefined 3D path to calibrate the movements of the robotic arm during surgery. This creates a center point for the cutting instrument, then intracortical pins are inserted into the tibia and femur, and the optical arrays are mounted. The arrays are tracked by the camera, which is integrated in the robot-assisted system [13]. The surgeon then goes through a process of registering a set of anatomical landmarks to map the patient’s preoperative CT scan to their anatomy on the operating table. The initial incision is made, and the leg is taken through a full range of motion. Once the skin incision has been made, the acquired data is matched to the CT model. Once the surgeon has registered the anatomy and set the implant position, soft tissue balancing is initiated. This is done through virtual kinematic modelling, and then intraoperative tracking allows real-time adjustments to be made so that the data collated is as close as possible to correct knee kinematics. Osteophytes are removed if they interfere with ligaments and other soft tissue structures. Removing the impediments allows correct kinematics and tissue tension during the passive force applied to the leg during a range of motion.  The articulating surfaces of the implants are adjusted to fill the flexion and extension gaps. Once an optimal position for the implants has been achieved, the plan will incorporate implant congruency, alignment values and gap kinematics. The final step is the application of a valgus force to the knee. The navigation system monitors this force and corrects the varus deformity appropriately by optimizing the position of the virtual components, and then the lower limb alignment is predicted. The target is typically 2 degrees of varus.

The RIO system has three components:

  1. Optical camera
  2. Computer
  3. Robotic arm

The camera is an infrared system, and the computer runs the software that operates the surgical plan. The robotic arm, has six degrees of freedom. Its movements are restricted in the incision site by the 3D boundaries that have been preset during the planning phase.  The robotic arm assists the surgeon with burring the femoral and tibial articulating surfaces. A high speed burr is attached to the distal end of the robotic arm. This burr can be either powered by hand or foot-pedal, and operates at 70,000 RPM [13]. Different size burrs are used for different functions. A 6 mm burr is used for the removal of major bone, and a 2 mm burr is used for the fine finishing of edges and corners. It can also be used for keel canal preparation for hip replacements, or alternatively, a 1 mm router can be used. The robotic arm controls the depth, width and length of burring through graphical real-time feedback. The arm is guided by the surgeon, who has control over the tip by guiding it within its boundaries. The robot gives the surgeon active tactile, visual and auditory feedback during the burring process. The navigation screen displays an overlay of the planed resection and the burred result to let the surgeon know the progress of the resection. When inside the safe zone, the arm will operate without resistance. Once the burr approaches the boundary, the system will make a series of beeps to alert the surgeon of the proximity to the boundary, and when the boundary is reached, the arm resists movement outside the area altogether.  Additionally, excessive force of the cutting volume or rapid movement of the patient anatomy will immediately stop unintentional resection. All burring is seen by the surgeon on the screen, which shows the 3D model of the distal femur and proximal tibia. The models are colour-coded and updated in real time to update the surgeon on resection progress. The resected area is coloured differently from surrounding areas so as to differentiate the two regions. When the robotic arm goes 0.5 mm outside the planned resection zone, which is green in colour, red appears on the screen and the arm progressively stiffens. If the surgeon attempts to further push the robotic arm outside of the green zone, then the robot will resist, the surgeon will be warned and the burr will immediately stop spinning [8]. This resistance is the haptic feedback feature provided by the system and intended to prevent surgeons from unintentionally cutting areas outside the planned resection zones. Once all burring is completed, the implants are inserted and cemented, and real-time long leg alignment is assessed so confirm satisfactory implant position. A final range of motion is performed to compare pre and post implant kinematics. For total knee replacements, the two components directly affected by the planes resected by the robot cutting tools are the femur and tibia. The key considerations when planning the tibial components are:

  • maximum cortical rim fit to prevent subsidence
  • avoid overhang and possible soft tissue impingement
  • minimize bone resection
  • manipulate implant position and or rotation to maximize implant congruency

The key considerations when planning the femoral components are:

  • maximize lateralization of medial and lateral femoral position to improve congruency with tibia
  • appropriate flexion and extension of components
  • avoiding patella maltracking

Pre-operative planning allows the surgeon prepare for implant size, minimize bone resection and preserve ligament structures. The robotic arm is under direct surgeon control and gives real-time tactile feedback to the surgeon as they perform the procedure.  This surgical platform combines interactive robotics, computer assisted planning and guidance with an intelligent bone cutting tool. It employs static referencing requiring rigid intraoperative fixation of the femur and tibia to a stereotactic frame. The enabling technology for this robotic platform is haptic, in which the computer system incorporates tactile, visual and auditory responses to provide a surgeon with a greater amount of dynamic feedback. These systems use optic motion-capture technology to dynamically track marker arrays fixed to the femur, tibia and robotic arm. This allows the surgeon to move the position of the limb and adjust its orientation during tactile-guided bone cutting [15].  For robotic total hip replacements, the MAKO system is used for reaming the acetabulum during bone preparation and cup placement. A computed tomography (CT) scan of the pelvis and both femurs is performed following the specified protocol as documented by the manufacturers, and the images are segmented and a 3-D reconstruction is performed. Anatomical landmarks, such as the anterior superior iliac spines and medial edges of the lesser trochanter are identified. The segmentation is completed by biomedical engineers, and can be verified by the surgeon if they so wish [16]. Misidentification of bone landmarks can distort the image and subsequent measurements, including offset, leg length and acetabular anteversion. The surgeon must visually confirm that the reconstructed pre-operative AP pelvic radiograph appears as optimal as possible. If various landmarks, such as the obturator foramina, the interteardrop or the sacrum, are not consistent with healthy bone or asymmetric in appearance, then the surgeon can detect these deformities or errors in segmentation and change the landmarks where appropriate. This process can be completed in either the operating theatre, or pre-operatively days in advance. The reference planes are established as such:

  • Anterior pelvic plane – tilt of pelvis whilst patient is supine in CT scanner
  • Medial-lateral axis – line connecting anterior superior iliac spines
  • Acetabular anteversion – calculated as functional plane

The surgeon must decide how to best use the information presented to him by the CT scans, as pelvic tilt can change between sitting and standing, as well as pre-operatively to post-operatively [16]. It is still relevant as a functional reference. Use of anatomic planes alone in the positioning of the acetabular component can be detrimental, as excessive functional abduction and anteversion can occur in the patient with decreased pelvic tilt, just as functional retroversion can occur when the patient has increased pelvic tilt. The MAKO software transforms the 3D models of the pelvis and femora into a 2D image that is the same as an anterior-posterior image. In patients with no pelvic deformity, the midline of the sacrum should align with the pubic symphysis. The femur is aligned so that the long axis is orientated parallel to the frontal plane, and this is done to eliminate any frontal distortion from flexion contracture, and thus show true femoral offset [16]. The software then measures all the pre-operative differences in leg length and offset, which is defined as the distance between the midsagittal plane of the pelvis and the diaphyseal axis of the femur. The surgeon assumes ultimate responsibility for how these measurements are interpreted, and how they are applied in conjunction with the surgeons own clinical judgement to make intraoperative decisions [16].

The procedure starts with a cut down to the joint and the acetabulum being virtually templated. The template is placed in 40 degrees of abduction and 20 degrees of anteversion, along the most lateral edge of the acetabular teardrop [16]. The size is selected with two goals in mind. The first being that the surgeon wants to remove as much cartilage and sclerotic bone as possible, and the second being  to preserve as much healthy bone as possible for implant fixation so that the centre of rotation can be restored to the anatomic location. The surgeon can preview the planned acetabular reaming on the 3D model and the following 3D appearance of the implanted cup sitting within the bone, and these views are in the coronal, sagittal and axial plane [16]. The multiple plane views allow the surgeon to anticipate potential issues such as the cup being exposed. Because the acetabular version references the functional plane, the initial plan does not always match the native anatomy, and so alterations related to cup size, anteversion and inclination can be made. It is simplest to match the contralateral hip when no abnormalities are present. In cases where the leg length and offset contain discrepancies, then the surgeon needs to make clinical assessment based off the pre-operative position of the surgical hip. The software does measure native femoral anteversion, but in instances where there are abnormalities, the surgeon can compensate by altering the acetabular version. The femur is templated by choosing an implant that wedges within the cortices of the bone of the proximal femur.  The location of the implant is important as it allows reconstruction of leg length and offset with multiple head and neck options, and planning the femoral options in 3D allows an accurate assessment of these variables. When 3D planning of the femur is combined with 3D acetabular planning, it allows the surgeon to anticipate possible changes to compensate for changes in the acetabular components [16]. The software measures the changes in limb length and offset for any combination of femur, head, cup and insert sizes, both in relation to the pre-operative state of the operative leg, as well as the contralateral leg. The surgeon needs to determine which of these measurements is of greater priority. In the operating theatre, the patient is placed in the lateral position. The navigation tower and robotic arm are positioned anterior to the patient and opposite the surgeon. When prepping the patient, the iliac crest and proximal half of the femur must be exposed. Whilst the incisional approach is being performed by the surgeon, the scrub nurse is registering the navigation system, taking approximately three minutes. The registration of the navigation instruments can be delegated as the final step before reaming begins. The verification process must be finalised by the surgeon [16]. There are two workflows that the surgeon can choose to follow, the express workflow, which  requires minimal registration, with just leg length and offset measured, and the enhanced workflow, which allows the surgeon to also navigate femoral version and the femoral neck osteotomy. The express workflow is sufficient for most cases, and the additional information received from the enhanced workflow is useful in cases where anatomical landmarks are not easily identifiable. The manufacturer suggests navigating the femur in instances where the plan has been changed intraoperatively. No femoral registration is required for the express workflow, as there is no femoral array. A single check point is placed on the distal femur before prepping the leg, and the greater trochanter is also selected as a checkpoint. A pelvic array is secured on the iliac crest with pins. The offset and leg length is then registered, and the joint then dislocated. The reason for the pelvic array being secured before dislocation is that it is a constant reference to measure leg length, and thus must be reproducible [16]. Once the femoral head is dislocated, the femoral neck is cut manually with a power saw and removed. Then the ligaments, labrum and other soft tissue surrounding the acetabulum are excised to provide adequate exposure. A pelvic checkpoint is recorded with the navigation probe to ensure that the optical array has not moved during or after registration. For acetabular registration, there are 3 initial points taken, followed by another 32 points both in and around the acetabulum. To ensure a more reproducible registration, it is suggested that surgeons aim for sclerotic bone, rather than osteophytes or cysts. Registration is verified by moving the probe along 8 surface points to verify that the the probe is within 1mm of the bone surface [16],  See Figure 4. At this point the robotic arm is brought into position anterior to the patient. The probe confirms the robotic arm, and therefore the reamer, and the reamer is brought into the acetabulum. Anteversion and abduction angles are set with the guidance of navigation. Haptic-guided reaming proceeds when the computer screen displays green bone that needs to be reamed. The underlying bone that does not need to be reamed is white on the screen. This occurs when the reamer is within 1mm of the planned depth. The underlying bone will appear red if reaming occurs beyond this point, and the haptic system will prevent further reaming through resistance, and it will switch off the reamer if it deviates more than 2.6 mm in any direction from the surgical plan [16]. Once reaming is finished, the reamer is removed and replaced with the acetabular impaction handle and acetabular implant. The robotic arm maintains the planned anteversion and abduction, whilst the surgeon uses a mallet for impaction to insert the acetabular implant. The haptic system will again engage during this step to ensure real time feedback on cup orientation is given to the surgeon, and that the cup is impacted to reamed depth. The final position of the cup is verified with navigation. This is necessary as this information is used later to calculate offset and leg length. This can be done by verifying 5 points around the implanted cup to determine the position of the cup [16]. The advantage of robotic control is that it creates an excellent pressfit with line-to-line reaming. With acetabular dysplasia, haptic-guided reaming and cup impaction is particularly useful as slip before engagement, and therefore compromises the peripheral fit of the implanted cup. Manual broaching and trialling of the femur is performed, and the trial head attached. Trial reduction is then performed. Standard manual techniques are used by the surgeon to assess leg length, and navigation is used in conjunction to augment these techniques. The femoral checkpoints are identified with the navigation probe. If the information provided by the from these points conflict with the surgeons intraoperative assessment, then the pelvic checkpoint can be verified to ensure that the pelvic array has not moved. If there is still a discrepancy, then the surgeon can compromise between their clinical assessment and the information provided by the system. Although not a part of the manufacturers recommended workflow, the lesser-to-centre distance from the proximal aspect of the lesser trochanter to the centre of the femoral head can be measured by the software, and it was found to be an excellent double check against both the surgeon’s clinical assessment on leg length and offset and the navigation system [16]. If the surgeon is satisfied that the pre-operative plan has been achieved, then the trials are removed, the definitive femoral stem is impacted. The trial head can be assessed once more for confirmation of the planned outcome. The final femoral head is then impacted, and the hip joint reduced. For the enhanced workflow, a modular array is placed on the proximal femur near the greater trochanter, along with the femoral checkpoint. A femoral broach array is used to measure the anteversion, as well as position of the broach relative to the pre-operative plan, and anticipated leg length and offset. The array is then disassembled, with the anchor pin left in place on the femur [16]. The same acetabular workflow is followed as is in the express workflow. The femoral stem is then impacted, and the modular array is reassembled. Changes in leg length and offset are recorded with navigation, as well as combined anteversion, and changes are made as needed. The ACROBOT Sculptor is another semiactive robot which uses CT data as input, and assists with bone resection in unicompartmental knee replacements, see Figure 5. It comprises of a surgeon-operated, high-speed cutting apparatus, which is attached to a robotic arm. Bone resections are constrained to a predetermined zone based on the pre-operative plan [17]. A CT image is used to identify the pathology of the patient and map out a preoperative surgical plan. This CT scan is segmented by the Acrobot’s own software. It generates the surface structure of specific bones, and in turn diagnoses the pathology. The surface model is then loaded in to the ACROBOT system for preoperative planning. Intraoperatively, the patient’s anatomical landmarks are registered and matched to the imaging data, allowing any points in the plan to be known to the surgeon. The ACROBOT has a cutting burr attached to its arm, which can move in three degrees of freedom, and can sculpt the bone based on the preoperative plan. To ensure accuracy, a tracking arm is pinned to the bone so that the system is aware of the 3D position of the bone relative to the robot at any time [18]. Following the attachment of the bone to the tracking arm, registration of points on the bone surfaces is also required for the procedure to begin. The ACROBOT system uses a mechanical digitizer, which is also a secondary use of the robotic arm, to register the surface. The tip of the cutter, which is ball point in nature, is used as a probe, and has a 2 mm diameter, and the data is transferred to the computer platform for modelling. After the selection of appropriate implants, the cutting coordinates are determined and then transferred to the computer that controls the robotic arm [14].  The system has a built in active damping/constraint system. This restricts the movement of the robotic arm outside the safe zone with force. Other robotic systems use a similar constraint algorithm [14]. The iBlock robotic cutting guide was previously known as Praxiteles, an example of which is shown in Figure 6. It is a “motorized, bone-mounted cutting guide that positions the saw guide for all femoral resections according to the surgeon’s plan, allowing the surgeon to then complete the resections with a standard oscillating saw” [12]. Utilized with an instrument called the NanoBlock (a separate, adjustable, resection block used for tibial resection), the system is an imageless robotic arthroplasty platform. The system’s computer station uses bone morphing technology to generate a unique 3D digital model of the patient’s knee. All anatomic data is acquired intraoperatively through registration. The system allows for planning of intraoperative implant positioning and sizing and visualizing planned bone cuts before they are made. The iBlock system is limited by having no haptic feedback, it is only available for total knee replacements, is a closed platform and has limited kinematic assessment after implantation of trials and/or implants.


Belleman reported 21 cases being performed with the CASPAR system between 2000 and 2002. Despite an intense learning curve, full leg x-rays showed an alignment within 1 degree of neutral alignment, with only 3 cases having been aborted due to technical difficulties. Siebert et al. used the same system for 70 cases, with “a mean difference between planned and obtained tibiofemoral alignment of 0.8 degrees” [4], whilst some outliers were noted, the most being 4 degrees. The average operating time was 135 minutes. The results reported by Borner et al. for the ROBODOC system for 100 cases showed alignment was within 3 degrees for all cases. 5% of cases were abandoned, switching to a conventional technique, and an operating time between 90 and 100 minutes after the initial learning curve. Nakamura, Nishii and Miki showed that, although using a model (JOA) that had not been validated (but used universally in Japan), patient satisfaction outcomes in patients who had been operated on with the ROBODOC system had increased when compared to patients being operated on using manual instruments in the first 2-3 years, but was not significant at 5 year follow up. The average operating time for the robotic group was 120 minutes, compared to 108 minutes for the manual instrument grasp. It was noted however though that the operating time in the robotic group was reduced by 17 seconds per procedure from the initial operating time of 140 minutes, indicating that there was a learning curve associated with the adoption of new technology. Clinically, leg length discrepancy was better at 5 year follow up. Although the differences were not significant, the variances in the robotic group were considerably less. Despite all the positive reports on clinical outcomes, there were other reports of technical complications, such as “a procedure stop due to bone motion during cutting requiring registration, femoral shaft fissures requiring wire cerclage, acetabular rim damage during milling, milling of a defect off the greater trochanter, and registration failures” [10]. The complications of active systems suggests that these systems should in fact be thought of as autonomous, with the implied ability on the part of the robotic system to make decisions, and that the surgeon ultimately decides to start, pause, or stop a preoperatively programmed milling procedure [10]. These systems showed excellent accuracy, and yet they were not generally adopted. The main reasons were the complex nature of the systems, as well as the cost, which was approximately 500,000 Euros. It is for these reasons that they were considered unjustifiable, and why most research groups and companies in orthopaedic robotics have started focusing more on semi-active robot systems [4]. Semi-active systems are more affordable and much more user friendly, and as such, a number of groups are utilising semi-active robots in arthroplasty and exploring their potential for future applications. Cobb et al. reported their results in a randomized controlled trial using the Acrobot system in unicompartmental knee replacements. They found that “the operation lasted on average 16 minutes longer in the Acrobot group, but the alignment was much better, with 100% of cases within 2 degrees deviation from the planned alignment” [4]. The conventional group only reported 40% of cases within 2 degree deviation. This same study also noticed a rise in patient reported outcomes, with Western Ontario and McMaster Universities Osteoarthritis index and the American Knee Society Scores improving at 6 weeks and 3 months. In a study by Roche et al., 43 patients had post-operative radiographs, and 344 individual radiographic measurements were taken. Of these, only 4 (1%) were identified as outliers. Postoperative American Knee Society Scores and Western Ontario and McMaster Universities Osteoarthritis index scores significantly improved from 95 to 150 and 41 to 21, respectively. Quality of life, as measured by the SF-12 physical summary, also significantly improved from 30 to 39. Robot-guided unicompartmental knee replacement procedures improved every measured clinical outcome. Masjedi et al. went an additional step further to not only validate the outcomes of the cuts made, but to validate the software that segments the bone against other software, the reproducibility of the registration process and the accuracy of the constraints. The results showed that for segmentation validation, the difference was considerably less than 0.5mm and that for the registration reproducibility, the maximum error among all subjects was found to be 1.8mm [18]. Additionally, the results for the placement of the femoral component was more prone to error with a maximum error of 5.3 degrees around the – and -axes [18]. Overall, the authors found that the results of segmentation, registration, and cuts made by robot were satisfactory using the ACROBOT system, and that it is possible to apply the protocol in the study to other robotic products on the market to better understand and validate robotic technology [18]. In a comparative study, Coon et al. studied a cohort of 33 patients that had received a robotic assisted knee to 44 standard unicompatrmental knee replacement patients. The RMS error of the tibial slope in the manual group was 3.5 degrees, compared 1.4 degrees in the robotic group. In the coronal planer, the average error was 3.3 degrees more in the manual group, compared to 0.1 degrees in the robotic group. At 12 weeks, the average increase in the combined Knee Society Score was 83.6 in the conventional group and 79.7 in the haptic-guided group. In a study of the first 20 MAKOplasty patients, Sinha et al. reported that using the robotic arm for total knee replacements resulted in extremely accurate reconstruction of the patient’s anatomy. In a post-operative assessment, all femoral components matched the pre-operative coronal and sagittal alignments [4]. The joint line changed only 0.4 mm. For the tibia, bone preparation matched pre-operative alignment in varus and posterior slope. However, final implant position had a slightly higher error rate, suggesting care must be taken when implanting the tibia.  Cobb and Pearle examined a cohort of 159 MAKOplasty patients. These patients were measured preoperatively and at follow-ups of 6 weeks, 3 months, and 1 year. MAKOplasty was shown to significantly improve all measured clinical outcomes. In a study for total hip replacements, Domb et al. found that all robotically placed cups were more likely to be in the safe zones compared with conventionally placed cups. 100%(50/50) of cups in the robotic THA group were in the safe zone of Lewinnek et al., compared with 80% (40/50) of cups in the conventional THA group [6].

Bitar et al. found some very promising results with the MAKO system. In their study, “97.6%  of robotic-measured values were within 10° of radiographic-measured values for inclination angle, 98.4% were within 10° for anteversion angle, 100% were within 10 mm for leg-length change, and 91.8% were within 10 mm for global offset change” [19]. All cases had radiographic leg-length discrepancy less than 10 mm and a radiographic leg length change less than 5 mm. Roche and colleagues found that at two years postoperatively, patients had an increased average range of motion of 129 degrees of flexion compared with a preoperative range of motion of 123 degrees. Postoperative Knee Society Scores also increased from 43.8 preoperatively to 96.8 postoperatively for knee scores and 63.9 preoperatively to 80 postoperatively for their function [8]. Coon and colleagues reported patient outcomes for 36 of their initial robotic-arm–assisted unicompartmental knee replacements and compared those results with the 45 cases performed with manual instruments. They saw no significant difference in average knee society scores at postoperative follow-up [8]. In a few studies, iBlock’s automated cutting guide resulted in more efficient and more accurate femoral cuts in comparison to conventional navigation methods in a cadaveric model. A cadaveric study compared the automated cutting guide of the iBlock to the conventional computer-assisted arthroplasty technique for femoral preparation. The mean femoral preparation time was shorter with the automated cutting guide than the conventional method (5.5 vs 13.8 minutes). The average deviation in the final bone resections was more accurate with iBlock’s automated cutting guide in all planes (frontal/rotational, sagittal, and cut height direction). The adjustable cutting block was found to provide equal or better component alignment, while there was a decrease in postoperative mechanical alignment and tourniquet time compared with conventional navigated instrumentation. There is very limited clinical data available for this system. A retrospective review of the first 100 cases with the imageless computer-navigated robotic cutting guide at a single institution was performed, allowing one surgeon to make bone resections within three degrees of neutral alignment in 98% of cases, but radiographic limb alignment was less precise, consistent with the known limitations inherent to this measurement technique [12]. The economic impact of robotics has been reported in numerous papers. Goradia has shown that economic analyses demonstrate cost benefits and savings only for high-volume centres [20]. There are some small potential savings for the hospital, such as reduction of inventory and less use of conventional instruments, but these soft costs are difficult to measure accurately [7]. There is also the additional operating time, however, for the usual surgeon doing only a few cases each week, this additional time may not be significant [7]. Although preliminary results are promising, long term clinical results are needed to really appreciate robotics benefits [9]. Clinical study reports with follow up at 10, 15 and 20 years are needed to gain a full understanding of the benefits and pitfalls of robotics, including implant survivorship, patient reported outcomes and economic factors.


Robotic-assisted unicompartmental knee replacements have gained popularity in the past 5-10 years, and now the major challenge to widespread implementation is cost. Results have demonstrated the accuracy and precision with which preoperative plans can be delivered. However, despite numerous studies demonstrating improved alignment, the clinical benefits are still yet to be proven. Robotics in orthopaedic surgery assists the surgeon in achieving an improved level of accuracy in implant placement. This is done through preoperative three-dimensional planning and intraoperative implementation via robotic delivery, and bone surface preparation errors are reduced due to the elimination of manual instrumentation. It has been argued that coronal and sagittal alignment goals may not be related to postoperative outcomes. Should this be shown to be true, then if and when alignment goals are ever changed, then robotic systems will be well placed to help surgeons achieve these new goals.

Robotic-assisted total hip replacements are consistent in placing the acetabular components in the safe zone of Lewinnek et al., but whether radiographic improvements observed will turn into clinical benefits for patients has yet to be reported. The economic impact of robotics within healthcare is potentially both a positive and a negative effect. By using a technology that improves the accuracy and precision of arthroplasty procedures, it may attract more patients to a centre utilising the technology. The downside is that, at present, these procedures require a longer operating time. This can have a negative economic effect if the surgeon is doing many joint replacements in a single day. There is also additional time required for preoperative planning for robotic surgery. Hospitals are most interested in cost and return on investment. Robotic systems can cost the hospital more than $1 million. These systems also require continuous calibration of hardware and software upgrades, which incur additional costs. The implant manufacturers are the real beneficiaries in the reduction of inventory and instruments, and in the future hospitals and area health service may be able to negotiate a lower price for implants used in robotic cases. Current designs are focusing on decreasing outliers and improving accuracy in arthroplasty radiographic outcomes. Whilst it will be difficult to predict the array of technological innovations that will be used to transform robotic-assisted arthroplasty procedures, the critical areas to focus will be preoperative analysis and intraoperative sensors, as well as robotic instrumentation. The next step for robotic-assisted arthroplasty will be to go beyond imaging to appreciate the kinematics of the operative knee before being altered by the pathology of arthritis, and then replicate the kinematics intraoperatively. Preoperative plans will be used to re-create the anatomic and kinematic framework that the surgeon strives to achieve. Intraoperative sensors will be essential for the implementation of the preoperative plan. Forces will be measured and contact points across the joints and implants through range of motion will quantify soft tissue balancing that would be otherwise be attempted through surgeon intuition. The surgeon will be given intraoperative real-time feedback to improve bone cuts, soft tissue tension, and ligamentous balancing to obtain the perfect outcome. The idea that implants need to be trialled will be redefined, as instead of verifying that the components are sized correctly, the re-creation of knee kinematics will be assessed. These changes will be implemented using haptics to provide feedback for the surgeon through devices controlled by robotics. Intraoperative sensors will quantify forces across the joint to verify whether restoration of normal kinematics has been achieved. Device companies continue to invest in robotic systems for orthopaedic surgery. Not surprisingly, surgeons who have had success with robotics in orthopaedic surgery believe the technology is here to stay. The consensus is that reduction in human error and consistent results possible with robotics is a significant benefit. In the future, robotic-assisted arthroplasty has the potential to become a valuable tool to assist the surgeon in delivering an optimized, patient-specific surgical implant plan, which will be designed to fit the knee or hip, rather than the current philosophy of having the knee or hip fit the implant.

Cite This Work

To export a reference to this article please select a referencing stye below:

Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.

Related Services

View all

DMCA / Removal Request

If you are the original writer of this dissertation and no longer wish to have your work published on the UKDiss.com website then please: