Secondary Logo

Journal Logo


Assessment of Internal and External Prosthesis Kinematics during Strenuous Activities Using Dynamic Roentgen Stereophotogrammetric Analysis

Papaioannou, George PhD; Mitrogiannis, Christos BEE; Nianios, George BEE; Fiedler, Goeran MSc

Author Information
JPO Journal of Prosthetics and Orthotics: April 2010 - Volume 22 - Issue 2 - p 91-105
doi: 10.1097/JPO.0b013e3181cca7bb
  • Free

Appropriate socket fitting is a complicated and a laborious process. The residual limb-socket-prosthesis complex can be considered as a dynamic human machine interface with intricate biomechanical behavior. A satisfactory socket interacts dynamically with the residual limb and delivers the desired results according to patient-specific load-transmission profile, enhanced dynamic stability, and efficient control for mobility.

It is apparent that improved socket fitting and comfort of lower-limb prostheses is essential for optimum functionality. Till date, many lower limb amputees are reporting dissatisfaction with socket comfort, residual limb pain, and/or skin breakdown particularly when engaging their prosthesis in dynamic strenuous tasks.1–7

Recent reviews of state-of-the-art techniques suggest that the development of successful sockets depends on the labor and cost intensive work of skilled prosthetists who are relatively scarce compared with the number of amputees. The literature shows that the majority of socket evaluation studies use only indirect measurement techniques for in-vivo assessment of mechanical linkage of all participating components in lower amputee limb dynamics, namely socket, soft tissue, and residual bone. Some of these indirect measurement methods include socket motion analysis techniques using externally fixed skin and socket markers,8–10 pressure measurements,11,12 computational modeling,13,14 friction-slip related phenomena,15,16 and limb tissue responses to external mechanical loads17,18 at the socket interface. The consensus is that improved direct measurement techniques for the understanding of stresses experienced at the residual limb have not yet led to enough clinical acceptances that could fundamentally alter clinical practice.19,20

Different sensing technologies have been used for the quantification of shear/slippage of soft tissue within the socket and the telescoping motion of residual bone with interesting but questionable and controversial results. These sensors demonstrate many disadvantages such as accuracy, hysteresis, signal drift,21 the response to curvature,22 inherent obtrusiveness,23 spatial resolution,24 temperature sensitivity,25 and unknown shear-coupling effects.11,12,26

Other methods offer the capability of incorporating patient-specific residual limb-socket geometry assessed by computed tomography (CT or spiral x-ray CT [SXCT]) into a computer-assisted design/controlled alignment method system. This allows software-based modifications and an inexpensive customized fabrication of a plaster positive likeness. Despite being static analyses, these and other socket rectification studies27 offered the benefit of seeing below the skin with high image quality. However, significant disadvantages from movement artifacts were reported,15,28,29 and it was concluded that the SXCT scanner is sufficiently precise and accurate only for static distance and volumetric quantitative socket-fitting studies. In addition, ultrasound and roentgenological-based measurements of the relative motion between the skeleton and the prosthetic socket30–34 have been used, but their 2D resolution, accuracy, and dynamic range cannot reliably assess dynamic tasks. The next level of static analysis is the use of biplane (stereo) Roentgen Stereophotogrammetric Analysis (RSA) to completely characterize the motion between socket and residual limb in three dimensions (3D).33,35 RSA socket-residual limb motion studies have reported variation of 10 to 30 mm in the vertical motion and 0 to 15 mm in the anterioposterior (AP) motion.36 Femur movements within transfemoral sockets was also measured by ultrasound techniques.30 All the aforementioned methods cannot differentiate between the contribution of frictional slippage/shear to the load transfer during 3D dynamic movement and the exact mechanism of residual tissue deformation that might be associated with skin damage. This is due to their limited accuracy and associated resolution in the study of dynamic activities, such as running, pivoting, stepping, sudden stop, and changing direction abruptly. Currently, there is no method of addressing the direction of 3D socket slippage and its rate during the performance of dynamic tasks.

This biomechanical analysis of transtibial amputee kinematics becomes more complicated if the patient is an existing total knee replacement recipient. In this complicated case, the altered mechanical loading at the knee joint will add to the complexity of residual limb loading and might result in further unpredicted effects on socket-residual limb kinematics.

In this particular case, the knee can be considered as a fundamental mechanical linkage that exhibits complex motions in response to applied loads that result from socket-residual limb and residual bone-internal prosthesis interaction. Accurate measurements of internal/external prosthesis kinematics during functional activities of amputees would provide the basis for assessing their performance. Past research on indirect techniques reveals that they are unable to precisely measure high-speed total knee replacement (TKR) kinematics. They simply rely on skin-mounted markers or fixtures for the assessment of bone motion.37–40 Although some progress has been reported in quantifying the effects of soft tissue,37 it remains to be seen whether accurate measurements of bone motion can be obtained using skin-mounted markers.41–43 Some methods apply direct attachment of markers to bone and reveal accurate results44,45 but cannot be considered with TKR subjects due to the associated patient discomfort and fear of infection. Data acquisition rate of high-speed CT and magnetic resonance imaging for motion measurement46–48 is still too low to make them suitable for dynamic tasks.

It should be noted that several (>20 currently registered) different designs of sockets are available in the clinic. Similarly, >30 different designs of knee replacements are available with variations in design, customization concepts, and sizes. The artificial prosthesis and socket-residual limb complex must postoperatively mimic the in vivo performance of the healthy joint. It is expected that both the large number of surgeries currently performed worldwide and the available implant designs would indicate that a fundamental understanding of the design, implantation, and dynamic behavior of the prosthesis is already available. Although some studies have been trying to address this problem,41–43 the methods available in the clinic offer almost no information on the detailed in vivo dynamic mechanical performance of joint replacements. Most designs are developed based on theory, mechanical simulation, and computational modeling studies, and the majority of objective performance data is derived from implants retrieved after clinical failure. Although TKR is already recognized as an extremely effective treatment for arthritic pain and functional disability, further improvements in TKR design would result in enhanced patient function and improved device longevity.

Accurate Dynamic Roentgen Stereogrammetric Analysis (DRSA)-based measurement techniques for tracking embedded small markers that are implanted in bone adjacent to the artificial joint have been demonstrated.41,42,49–62 Markerless DRSA methods are also available for different joints.63,64 However, the techniques have not yet demonstrated frame rates or translational/rotational accuracy adequate for analysis of dynamic (high-speed impact) motions when applied to combined internal and external prosthesis kinematics assessment.

This study presents a new method of assessment of three-dimensional (3D) high-speed socket-residual limb slippage and knee joint replacement kinematics using Biplane DRSA instrumentation. This direct combination of the “marker-based and model-based” techniques has been previously reported to perform the 6D searches necessary to obtain the required 3D kinematics61,64–66 with accuracy that is one order of magnitude or greater of improvement from that commonly used in conventional motion analysis techniques. The proposed method intends to provide a more accurate in-vivo socket-joint replacement kinematics assessment to the prosthetist and therefore enable improved socket fitting with less discomfort for the amputee.


A 65-year-old woman transtibial amputee (Table 1 and Figure 1) wearing a patellar tendon bearing (PTB) socket with silicon liner was selected for this study, approved by our institutional review board. At the initial interview, the subject complained about the performance of her socket identifying problems with frequent discomfort and pain that required volumetric adaptations and resulted in limited prolonged use and occasional pressure ulcer incidents. We rigidly fixed several 4-mm tantalum markers on the external surface of the socket according to a recognizable pattern for easier tracking (Figure 2). Stripes of tantalum pigment paint in the shape of orthogonal meshes were placed on the skin around the perimeter of the residual limb to cover the most critical areas where possible skin deformation might occur (Figures 1, 2). A high-resolution laser-based 3D digitizer (Konica Minolta, VIVID 910 Noncontact 3D Digitizer, Color image data for 300,000 [640 × 480] points per scanning; Konica Minolta Sensing Americas, Inc., NJ) was then used to scan the 3D geometry of the residual limb-skin surface and internal and outer surface geometry of the patient's socket. The 3D distances between the skin markers on the residual limb during the unloaded (before donning) residual limb and the 3D distances of the socket surface markers were then possible to scan and register. The carefully donned amputee residual limb-socket complex with all the markers was then scanned with a CT imager (slice spacing, 0.5 mm; 1000 × 1000 pixels matrix; GE, St. Luke's Hospital, Aurora Health System, Milwaukee, WI). These protocols enhance the accuracy for co-registration of the CT-based reconstructed geometry of the residual bone with the 3D geometry of the residual limb (using the tantalum paint skin mesh markers) and the 3D socket surface data (Figure 2 and Table 1).

Table 1:
Anthropometry measurements of the amputee participating in the study
Figure 1.:
a, Transtibial amputee in the Biplane Dynamic Stereoradiography instrument; (b) fusion of all the imaging modalities into one entity with 3D co-registration of socket-skin markers and bone geometry. All the coordinated systems (CS) are shown: Global CS, computed tomography (CT) data CS, and socket CS: socket-based embedded coordinate system was obtained from three markers placed at standard locations in the socket. c, d, DRSA images showing the socket-residual limb complex at two different time instants (impact: t 1 = 0.0 seconds) and (t 2 = 1.46 seconds). Note the cluster of skin-socket markers and the bone edges; also note the initial impact phase (t 1) and the impact absorption phase (t 2) showing the characteristic residual bone vertical telescoping motion during the step-down task.
Figure 2.:
The protocol for all marker placements at both the residual limb and socket regions. The socket for the purpose of data presentation is divided into several sectors, so that markers can be labeled from combinations of four letter codes: So stands for socket and sk stands for skin; P, M, and D stand for proximal, middle, and distal part of the socket/residual limb, respectively, followed by L and M for lateral and medial side; and finally Ant and Pos stand for anterior and posterior side.

The patient was then asked to perform several sudden fast-stop movement trials (n = 3) in the direction of progression during running and the task of coming down from stairs (while landing on the prosthesis [stair height, 22 cm], n = 3), which were assessed using DRSA (IMEROSIN™, Bioimerosin Laboratories SA, Milwaukee, WI) and force plate instrumentation.58–60,67 This study did not precisely control the direction of the laboratory-based coordinate system (Figure 1), because the laboratory-based coordinate system orientation is not relevant when results are reported in an anatomical-based coordinate system defined by bone landmarks. Past DRSA studies have reported on its capability to measure the displacement of the tantalum markers during high-speed motion with high resolution (1440 × 1200 pixels; Figure 1) and with ±0.1 mm translational and 0.8 degrees rotational accuracy.61,65,66,68 In this study, the cameras were optically coupled to image intensifiers with lenses that allowed aperture of 1.0 to enhance the sensitivity of the system. The cameras were shuttered at 1/2000 seconds to eliminate motion blur. Radiographic parameters were optimized to identify the implanted beads and considerable reduction in exposure (55 kV, 60 mA) was possible due to the camera, lens, and intensifier optical coupling69. Maximum entrance exposure from the biplane radiographic system using these parameters was 0.09 R per trial as it was assessed with a dose real-time monitoring (Polimaster PM1203M, Minsk, Republic of Belarus) method that was always performed during operation. Tracking of the high-speed kinematics of the orthogonal skin meshes, the residual bone edges, the knee joint prosthesis, and all the tantalum markers was possible using marker-based and markerless tracking software (Figures 3, 4).57,61,64,66 The method has been shown to assess efficiently in vivo joint motion from biplane radiographic images without implanted tantalum beads.64,70 This technique, referred to as the model-based tracking technique (MBT), can track the position of bones based on their 3D shape and texture. To validate this technique, we 1) measured the position of the socket by tracking the embedded tantalum beads at the socket of our participating transtibial amputee using the marker- based method, 2) measured the position of the socket using the model-based tracking system, and 3) compared the results of the two techniques. The DRSA data were used as the “gold standard.”

Figure 3.:
A pair of digitally reconstructed radiographs (DRRs) is constructed from the CT residual bone/prosthesis and socket model. The position and orientation of the models is refined to optimize the correlation between the two DRRs and the two biplane x-ray images (actual radiographs).
Figure 4.:
a, Single x-ray image close up of a skin orthonormal tantalum paint mesh from the DRSA sequence. b, c, Gaussian filter to enhance the skin marker signature, identifying relative skin-tissue strain by assessing the elongation/deformation of the skin mesh at different time instants. d, e, Similar estimation of relative engineering shear strain by assessing the 3D skewness (“departure from orthonormality”) of the skin mesh. f, Image processing of the grey signature of the crosspoint at the skin mesh in terms of grey-level distribution assists in identifying shear by comparing the mesh at rest (unloaded socket, top) and the skewed mesh (loaded socket, bottom).

Given the geometry of the biplane x-ray system and a 3D bone, joint prosthesis and socket model from a CT scan, a pair of digitally reconstructed radiographs (DRRs) can be produced via ray-traced projection through the 3D bone model (Figure 3).64 In the next step, the similarity between the two DRRs and the actual 2D biplane radiographic images is tuned using optimization, so that the in vivo position and the orientation of the socket or residual bone can be estimated. Laplacian enhancement masks, Gaussian blur filters, Canny edge, and Sobel edge-detector output are added to the base images for both the DRRs and the radiographs to improve the matching process. The quality of matching is measured by calculating the correlation coefficient of each DRR with its corresponding radiograph and then multiplying the two view correlations to get total system correlation.

The first step in the model-based tracking involved developing and separating the 3D volumetric bone, prosthesis, and socket model. First, the residual bone and joint prosthesis CT images were manually segmented from soft tissue and the socket information. The implanted tantalum beads were manually removed from the CT images so that the presence of implanted beads did not improve the model-based tracking. The CT volume was then interpolated using a feature-based interpolation technique and scaled to have cubic voxels with dimensions similar to the 2D pixel size in the biplane x-ray system images.

The model-based tracking process uses a homegrown operator-friendly interface of graphical tools. This graphical interface includes the following tools: 1) a visual overlay of the DRRs on the radiographs that empowers the operator's initial guesses and provides contrasting enhancements to help the operator match position and orientation; 2) an array of six position buttons that control the position and orientation of the model; 3) a low-resolution 2D search tool that performs a broad exhaustive search by translating and rotating each DRR to maximize the correlation with its radiograph; 4) a high-resolution but finer narrower local six-axis search tool to refine alignments in position and orientation; 5) similarity between the DRR and the radiographic image is determined with a correlation and an optimization71–73 iterates motion parameters until the maximum similarity is obtained. Estimating 3D kinematics requires simultaneous optimization of all the six motion parameters (three positions and three rotations). Once 6 df of the center point of the bone, prosthesis, or socket model are estimated from each single-plane system, the absolute 3D position and orientation of the bone in the reference coordinate system are determined using a 3D line intersection method and the known imaging geometry of the biplane system; and 6) visualization tools for evaluation of quality of the motion.

In addition, the new MBT and markerless tracking algorithms apply image-processing routines (Laplacian enhancement masks, Gaussian blur filters, and Canny edge detection; Figure 4), so that enhanced tracking of the orthogonal skin-mesh based on its geometrical properties is possible.74 This allows characterization of 3D displacement/deformation and skewness, i.e., relative strain and relative engineering shear strain of the skin mesh from the sequences of DRSA images. Two different methods were used to estimate the relative “resting” initial length (3D distance) between the skin-mesh markers. In the unloaded (without the socket) mode, the 3D distance was obtained from the laser scan measurements. In the unloaded but donned socket mode, we averaged the 3D distances from DRSA data collected from a total of 10 to 20 frames. This “skin resting initial length” is then combined with the DRSA skin mesh deformation data to produce relative strain/shear assessments. This is repeated for all critical regions of the residual limb during the impact phase of the strenuous tasks. The kinematic relationships of the socket, residual bone edges, TKR prosthesis, and several skin-meshes were analyzed using data from the two tasks from touchdown to up to 0.4 seconds (or up to the time point when data existed). Clinically relevant joint kinematics was calculated from each of the marker-based and model-based tracking results. Translation was defined as the three-dimensional distance between the femoral and tibial prosthesis component's most distal edge and expressed in the tibia anatomical coordinate system. Previously described anatomic coordinate systems were adopted.73 Segmental rotations were calculated using ordered rotations.75 Accuracy measures (bias, precision, and root mean square [RMS] error) for clinically relevant kinematic variables were calculated as described earlier for the 2.4 seconds after foot impact. One-sample, two-tailed, t-tests with alpha set at 0.05 were used to test for bias. 3D pathways or proximity maps (using software tools from MATLAB [The Mathworks Inc., MA], Autodesk 3ds max studio 2009, Autodesk Autocad 2008 [Autodesk Inc., CA], Geomagic Studio 10 [Geomagic US Corporate, NC], and Mimics [The Materialise Group, Leuven, Belgium]) between the socket-skin-TKR-bone-edge markers at critical socket areas were plotted for the tasks in question. Areas of kinematic importance were further animated and presented with visualization paradigms that the clinician is trained to interpret.


Reassessing the static and dynamic accuracy method of DRSA of the marker-based method was possible by estimating the difference between the known 3D distance of two rigidly fixed socket markers and the respective measured 3D distances during the static standing and dynamic tasks (from all trials). Static accuracy was 0.03 ± 0.06 mm and dynamic accuracy was 0.09 ± 0.05 mm. Minimum and maximum dynamic errors in determining the 3D distance were between 1% and 2.3% of the original known distance between the two socket markers (SoMMAnt and SoMLPos).

Estimation of the efficiency of the model-based tracking technique relative to dynamic RSA was established by quantified accuracy in terms of bias and precision as reported elsewhere.62,65,76 Measurement bias was assessed by determining the average difference in 3D tantalum marker locations between the two techniques across all trials. It was assumed that the tantalum markers were rigidly fixed in all structures in question. Therefore, a direct estimate of the uncertainty in the model-based tracking measurements independent of residual bone, prosthesis, or socket motion was possible. The standard deviation of the difference between the two techniques during the motion trials represents the dynamic precision. We also assessed the overall dynamic accuracy of the model-based tracking technique by computing the RMS error between the two techniques across all trials.

Excellent agreement was found between the results produced by the model-based tracking and the DRSA technique. Quantitatively, measurement bias ranged from 0.012 to 0.11 mm (depending on coordinate axis) for the femoral prosthesis and ranged from 0.004 to 0.048 mm for the socket (Table 6). Dynamic measurement precision was better than 0.49 mm for the femoral prosthesis and 0.37 mm for the socket. The assessment of overall dynamic accuracy indicated that RMS errors in any one direction were <0.50 mm for the femoral prosthesis and <0.40 mm for the socket. Finally, the assessment of overall dynamic rotational accuracy indicated that RMS errors in flexion extension, ab-adduction, and internal-external rotation were <1.03, 0.29, and 1.12 degrees, respectively.

Table 6:
Measurement bias (i.e. average difference between model-based tracking and dynamic RSA) and measurement precision (i.e., standard deviation of the difference between model-based tracking and DRSA)

Dynamic skin slippage was assessed by plotting the relative 3D displacement of a skin-marker with respect to the socket. Figure 5 shows the respective proximity 3D pathway for characteristic skin markers and the 3D kinematics (telescoping motion) of the femoral/tibia prosthetic components at four time instants (t1: 0.0 seconds, t2: 1.46 seconds, t3: 1.8 seconds, and t4: 2.14 seconds) after impact with the ground during the step-down trial 2. It was observed that the characteristic slippage patterns of skin markers skDAnt, skDL, skDPos, and the femur and tibia prosthesis 3D displacement were the most amplified immediately after impact. All skin markers inside the socket demonstrated the largest displacement in the X-direction of progression (3.51 to 5.6 mm), whereas vertical displacement ranged between 1.2 and 4.3 mm. The skin marker skPPos that is outside the socket and just at the edge of the socket's upper rim demonstrated the highest magnitudes in the displacements in all three directions (X: 11.3 mm, Y: 5.073 mm, Z: 7.883 mm). There was a significant amount of perturbation in these areas of the skin that occurs between time instant t1 that the foot impacts the ground and time instants t2t4 when the skin settles and the impact has been absorbed by the skin-residual limb-socket complex. In addition, the femoral/tibia prosthesis components moved 25 mm vertically (a clear telescoping type of motion) with respect to the socket (Figures 1, 5). The femoral component translated almost twice as much in the X and Y direction when compared with the tibia motion. These plots can be used as visualization assistants to demonstrate large dynamic slippage of the skin tissue at critical regions. The 3D maximum, minimum, and average distance (millimeter) as well as relative strain and engineering shear strain values between selected residual limb-socket-markers and the residual bone-edge for fast-stop and step-down trials are given in Tables 2 to 5. The maximum 3D distance between several socket-skin-marker pairs ranges from 3 to 16 mm for the fast-stop task and from 2 to 8 mm for the step-down task. Marker pair soDMAnt-skDPos demonstrated the maximum 3D distance of 16 mm in fast-stop trial: this is the pair representing the distal medial-anterior side of the residual limb-socket complex. It should be noted that for all trials, the highest maximum 3D distance is observed between the residual femoral prosthesis component and socket-or-skin-marker pairs located at the distal residual limb. The highest telescoping motion of the residual bone with respect to the skin marker skDAnt reached a value of 28 mm for the step-down trial 2. Higher magnitude ranges were observed between the most proximal skin markers that were outside the socket. Skin-marker pairs reach relative strain values that range between 0.01 and 0.1 for step-down and fast-stop trials, respectively (Table 4). The relative engineering shear (γ) between selected marker clusters that form orthonormal meshes (skPAnt, skMAnt, skDAnt, skDL, and skDPos) ranged between 81.5 and 129 degrees. Table 5 demonstrates the highest relative engineering shear (γ) values from all skin markers measured for marker skDAnt.

Figure 5.:
a-c, 3D slippage (millimeter) of skin markers and femoral/tibial prosthesis 3D telescoping motion with respect to socket at four time instants after heelstrike for a step-down trial (trial 2). There is a significant amount of perturbation in selected areas of the skin that occurs between time instant t 1 = 0.0 seconds that the foot impacts the ground and time instants t 2t 4 (t 4 = 2.4 seconds) when the skin settles and the impact has been absorbed by the skin-residual limb-socket complex.
Table 2:
Three-dimensional maximum, minimum, and average distance (mm) between selected skin (Sk) and socket (So) markers and the residual bone-edge (BE) for the worst (in terms of max distance values) fast-stop trials (n = 3 trials per task)
Table 3:
Three-dimensional maximum, minimum, and average distance (mm) between selected skin (Sk) and socket (So) markers and the residual bone-edge (BE) for the worst (in terms of max distance values) step-down trials (n = 3 trials per task)
Table 3:
Table 4:
Maximum minus minimum 3D distance (mm) during the performance of a task, and relative engineering strain relationships for selected skin-to-skin marker pairs (markers and trials with the maximum strain are shown)
Table 5:
Relative engineering shear strain (γ) between selected marker clusters that form orthonormal meshes around a skin marker (data from marker skDAnt is shown here): data from the worst case trials

The results demonstrate a downward, anterior-posterior shift of a whole group of distal, and a lateral and posterior region skin markers after impact. Figures 6 to 8 present the worst-case scenarios of overall “deformation” between adjacent pairs of skin-to-socket to bone edge (joint prosthesis) and skin-to-skin markers for both tasks. Maximum deformation of up to 12.5% is observed for the fast-stop trials and step-down trials between skin-to-skin-marker pairs (markers with highest deformations for both tasks are shown in Figure 6). The respective deformation between skin-to-socket marker pairs reached maximum values of almost 22% (markers with highest deformations for both tasks are shown in Figure 7). The deformation between the tibia prosthetic component and skin/socket marker pairs reached maximum values of almost 100% (Figure 8).

Figure 6.:
Overall “deformation” between adjacent pairs of skin-to-skin markers (also average data from all trials [n = 6] ±SD are shown). Plots show step-down and fast-stop data for the worst trials (those with maximum deformation). Note that depending on the location of the marker, deformation from “resting or unloaded intermarker distance” can range between 2% and 12%.
Figure 7.:
Overall “deformation” between adjacent pairs of socket-to-skin markers (also average data from all trials [n = 6] ±SD are shown). Plots show step-down and fast-stop data for the worst trials (those with maximum deformation). Note that depending on the location of the marker, deformation from “resting or unloaded intermarker distance” can range between 1% and 22%.
Figure 8.:
Overall “deformation” between adjacent pairs of socket-to-skin markers and joint prosthesis/residual bone edge (also average data from all trials [n = 6] ±SD are shown). Plots show step-down and fast-stop data for the worst trials (those with maximum deformation). Note that depending on the location of the marker, deformation from “resting or unloaded intermarker distance” can range between 2 and almost 100%.


Socket fitting and evaluation is a skill that remains an art today, depending on the accumulated expertise of the practitioner/prosthetist. A new method for the evaluation of dynamic fitting of prosthetic sockets is presented here. Marker- based assessment of dynamic socket-residual limb and bone-socket telescoping motion with as much as ±0.03 mm translational and 1.3 degrees rotational accuracy (one order of magnitude higher than current techniques) using DRSA was demonstrated. The in-vivo dynamic accuracy for the markerless tracking method was further improved from that reported previously.61,64,77 Excellent agreement was found between the results produced by the model-based tracking and the DRSA technique. In particular, the position and orientation of the femoral prosthesis component and socket component were qualitatively acceptable when superimposed over the original biplane images (Figure 3). Quantitatively, measurement bias ranged from −0.012 to −0.11 mm (depending on coordinate axis) for the femoral prosthesis and from 0.004 to 0.048 mm for the socket (Table 6). This is due a) to the enhanced out-of-plane CT resolution used in this study (an improvement in slice thickness from 1.5 to 0.5 mm) and b) to the improved DRSA image resolution (1440 × 1200 pixels matrix). This in turn was the result of improved x-ray tube focal spot (0.3 to 0.8mm) and increased image intensifier size (19"). These hardware and camera improvements resulted in almost complete reduction in blurriness due to motion and significantly more information on the image for the tracking algorithms.

No operator bias data are available because only one operator processed and analyzed the data. It should be noted that the marker placement protocol must be kept identical as much as possible so that differences in the geometry and shape-morphology of socket/residual limb complex will not introduce discrepancies when comparing data from different patients. Another critical challenging aspect of this method is that in very dynamic tasks such as the fast-stop task the critical socket regions of interest might get out of the field of view (40 cm intensifier diameter). They might also get occluded by the other limb while trying to perform the task naturally and without familiarization. It is recommended that large image intensifiers and alternative x-ray beam penetration angle orientations be in use.

Our results showed that deformations (from an initial “resting unloaded length”) between selected skin-socket pairs of markers and joint prosthesis-bone-edge exceeded 100% for both strenuous trials. These deformations relate to 3D distances that reached values of up to 25 mm for some marker pairs for both tasks. These tasks demonstrate a large envelope of magnitude in socket marker and bone kinematics as well as an abrupt slope immediately after the impact phase. Deformations of up to 12.5% and 22% were observed for both trials between skin-to-skin marker pairs and between skin-to-socket marker pairs, respectively. These deformations succeed a huge vertical downward displacement (telescoping motion) of the joint prosthesis immediately after the impact followed predominantly by an AP movement of all skin markers. This information on the telescoping behavior of the residual bone and slippage mapping is critical in one's effort to further rectify the socket toward improved AP stability. This is further supported from the fact that the prosthetist now has more detailed information demonstrating that the most affected skin areas are those at the distal posterior part of the socket for this subject.

Tracking of 3D skin surface mesh markers indicated that skin deformed (through a characteristic skewness of the mesh) during the support/impact phase of both tasks (Figure 4). Maximum strains were observed for skin-marker pair skDM-skDPos and were 0.07 and 0.1 for the step-down and fast-stop tasks, respectively. These are the markers placed distally and posteriorly on the residual limb that reflect the vertical (distal-to-proximal movement) axes and signify the topology of the more severe tissue deformations occurring during these strenuous activities. Strains of all other skin marker pairs range from 0.01 to 0.1. Relative engineering shear strain for skin-marker meshes associated with marker skDAnt (worst case) reached values ranging from 81 to 130 degrees (range for other marker clusters: 81.5 to 129 degrees). Engineering shear strain is defined as the change in the angle between two material line elements initially perpendicular to each other in the nondeformed or initial configuration. In this study, this change in the initial configuration was represented by the change in the orthonormal lines of the skin mesh. Because complete initial orthonormality in the skin-mesh was not always met in this study after socket donning, the shear strain data should be viewed with precaution. It is apparent that for this protocol special attention must be given to the size and density of the skin meshes and their position on critical residual limb locations.

Assessing the dynamic skin 3D slippage was possible by plotting the relative displacement of a skin marker with respect to the 3D inner-socket mesh of the area around an adjacent socket marker. The method is capable of describing with the use of a 3D visualization tool the large dynamic slippage of skin tissue at critical regions. In the fast-stop and step-down trials, skin slippage of up to 16 mm and residual bone telescoping motion of up to 25 mm was observed. Figure 5 shows the 3D direction of skin slippage with respect to the patient-specific inner-socket geometry. The maximum slippage of 16 mm occurs in the fast stop-trial from a skin marker pair representing the distal medial and anterior side of the residual limb-socket complex. It should also be stressed here that for this subject and for all trials, the highest deformation was observed between the residual bone edge and tibial prosthesis component and socket-or-skin marker pairs located at the distal residual limb. It is important to note that this skin slippage is associated and synchronous to the significant telescoping motion of both components of the internal prosthesis. Information obtained from a proximal posterior-medio-lateral skin-marker above and outside the posterior socket rim indicates that in the step-down trials, the skin marker is moving downward and in the AP direction during impact and then away from the socket rim as the load is absorbed and the residual limb settles in the new position. Overall displacement in the outside the socket markers is twice that of the markers inside the socket. The above findings are interesting knowing that the PTB socket is expected to be a characteristically tight-fit socket design. It should be noted however that this particular amputee was wearing a silicon liner and had excessive residual skin and adipose tissue on the residual limb. The subject was capable of performing these activities but complained for discomfort and pain. This might be an explanation for the increased residual limb mobility, slippage, and instability even at the given tight-fit socket design.

We are in the process of further calibrating patient-specific, whole lower limb (socket-residual limb-prosthesis) finite element (FE) models driven by DRSA high-accuracy kinematics. In the past, the preferred protocol used for most socket rectification simulation studies relies on prescribing the displacement boundary conditions at the nodes on the outer surface of the socket or liner.17,78–80 Displacement boundary conditions corresponding to the shape of a given socket design are applied to deform the residual limb soft tissue or the liner to conform to the rectified socket shape. Data available from this study provide accurate dynamic in-vivo patient-specific displacement boundary conditions at the nodes on the inner surface of the socket. Adding tantalum meshes at the liner would also improve our current protocol and provide information on the liner dynamics. In a very complicated residual bone-residual limb-socket FE model, this information can be critical in improving contact solution and the explicit or implicit model convergence.

The new marker and markerless socket-residual limb-internal prosthesis kinematics assessment method a) is one order of magnitude or greater of improvement in accuracy from existing conventional motion analysis techniques, b) allows the use of the markerless and marker-based methods interchangeably to track the different hard and soft segments/components in a completely unobtrusive way, c) uses new representations of this hard-to-soft tissue interaction with 3D visualization paradigms that the prosthetist is trained to interpret, and d) provides information that is not possible with other in-vivo dynamic tracking techniques. This highly accurate in-vivo patient-specific unobtrusive assessment of the dynamic socket-residual limb 3D slippage and residual bone telescoping motion can significantly impact the iterative cycle of socket fitting and evaluation.


1. ACA. Available at: Accessed April 8, 2009.
2. Dillingham TR, Pezzin LE, MacKenzie EJ, Burgess AR. Use and satisfaction with prosthetic devices among persons with trauma-related amputations: a long-term outcome study. Am J Phys Med Rehabil 2001;80:563–571.
3. Ephraim PL, Dillingham TR, Sector M, et al. Epidemiology of limb loss and congenital limb deficiency: a review of the literature. Arch Phys Med Rehabil 2003;84:747–761.
4. Frost & Sullivan. U.S Lower Extremity Prosthetics Markets; 2007:58.
5. Miller AL, McCay AJ. Summary and conclusions from the academy's sixth state-of-the-science conference on lower limb prosthetic outcome measures. Am Acad Orthot Prosthet 2006;18:2–7.
6. P&O. Report on the State-of-The-Science meeting in prosthetics and orthotics. Research in P&O: Are we addressing clinically-relevant problems. Northwestern University Feinberg School of Medicine, Chicago, IL; 2006.
7. Pezzin LE, Dillingham TR, Mackenzie EJ, et al. Use and satisfaction with prosthetic limb devices and related services. Arch Phys Med Rehabil 2004;85:723–729.
8. Huang GF, Chou YL, Su FC. Gait analysis and energy consumption of below-knee amputees wearing three different prosthetic feet. Gait Posture 2000;12:162–168.
9. Schmalz T, Blumentritt S, Jarasch R. Energy expenditure and biomechanical characteristics of lower limb amputee gait: the influence of prosthetic alignment and different prosthetic components. Gait Posture 2002;16:255–263.
10. van der Linde H, Hofstad CJ, Geurts AC, et al. A systematic literature review of the effect of different prosthetic components on human functioning with a lower-limb prosthesis. J Rehabil Res Dev 2004;41:555–570.
11. Cork R. XSENSOR technology: a pressure imaging overview. Sensor Rev 2007;27:24–28.
12. Pramanik C, Saha H, Gangopadhyay U. Design optimization of a high performance silicon MEMS piezoresistive pressure sensor for biomedical applications. J Micromech Microeng 2006;16:2060–2066.
13. Torres-Moreno R, Jones D, Solomonidis SE, Mackie H. Magnetic resonance imaging of residual soft tissues for computer-aided technology applications in prosthetics—a case study. J Prosthet Orthot 1999;11:6–11.
14. Zachariah SG, Sanders JE. Finite element estimates of interface stress in the transtibial prosthesis using gap elements are different from those using automated contact. J Biomech 2000;33:895–899.
15. Commean PK, Smith KE, Cheverud JM, Vannier MW. Precision of surface measurements for below-knee residua. Arch Phys Med Rehabil 1996;77:477–486.
16. Zhang M, Mak AFT. In vivo friction properties of human skin. Prosthet Orthot Int 1999;23:135–141.
17. Reynolds DP, Lord M. Interface load analysis for computer-aided design of below-knee prosthetic sockets. Med Biol Eng Comput 1992;30:419–426.
18. Vannah WM, Drvaric DM, Hastings JA, et al. A method of residual limb stiffness distribution measurement. J Rehabil Res Dev 1999;36:1–7.
19. Mak AF, Zhang M, Boone DA. State-of-the-art research in lower-limb prosthetic biomechanics-socket interface: a review. J Rehabil Res Dev 2001;38:161–174.
20. Report on the State-of-The-Science meeting in prosthetics and orthotics. Research in P&O: Are we addressing clinically-relevant problems, Northwestern University Feinberg School of Medicine, Chicago, IL; 2006.
21. Papaioannou G, Protopappas CV, Tsopelas P, et al. A new method for pressure sensor equilibration and conditioning. Braz J Biomotr 2008;2:176–195.
22. Papaioannou G, Demetropoulos KC, King HY. Predicting the effects of knee focal articular surface injury with a patient-specific finite element model. The Knee 2009;17:61–68.
23. Hadcock L, Stevenson J, Morin E, et al. Pressure Measurement Applications for Humans. Kingston, ON: Queen's University; 2007.
24. Brimacombe JM, Anglin C, Hodgson AJ, Wilson DR. Validation of calibration techniques for tekscan pressure sensors. XXth ISB Congress—29th ASB Annual Meeting, Cleveland, Ohio, USA, 2005.
25. Buis AW, Covery P. Calibration problems encountered while monitoring stump/socket interface pressures with force sensing resistors: techniques adopted to minimise inaccuracies. Prosthet Orthot Int 1997;21:179–182.
26. Polliack AA, Sieh RC, Craig DD, et al. Scientific validation of two commercial pressure sensor systems for prosthetic socket fit. Prosthet Orthot Int 2000;24:63–73.
27. Engsberg JR, Sprouse SW, Uhrich ML, et al. Comparison of rectified and unrectified sockets for transtibial amputees. J Prosthet Orthot 2008;18:1–7.
28. Smith KE, Commean PK, Vannier MW. In Vivo 3D measurement of soft tissue change due to lower limb prostheses using spiral computed tomography. Radiology 1996;200:843–850.
29. Vannier MW, Commean PK, Smith KE. 3D lower- extremity residua measurement systems error analysis. J Prosthet Orthot 1997;9:67–76.
30. Convery P, Murray KD. Ultrasound study of the motion of the residual femur within a trans-femoral socket during gait. Prosthet Orthot Int 2000;24:226–232.
31. Erikson U, Lemperg R. Roentgenological study of movements of the amputation stump within the prosthesis socket in below-knee amputees fitted with a PTB prosthesis. Acta Orthop Scand 1969;40:520–529.
32. Grevsten S, Erikson U. A roentgenological study of the stump-socket contact and skeletal displacement in the PTB-Suciton prosthesis. Ups J Med Sci 1975;80:49–57.
33. Lilja M, Johansson T, Oberg T. Movement of the tibial end in a PTB prosthesis socket: a sagittal X-ray study of the PTB prosthesis. Prosthet Orthot Int 1993;17:21–26.
34. Long I. Normal shape-normal alignment (NSNA) above-knee prosthesis. Clin Prosthet Orthot 1988;9:9–14.
35. Bocobo CR, Castellote JM, MacKinnon D, Gabrielle-Bergman A. Videofluoroscopic evaluation of prosthetic fit and residual limbs following transtibial amputation. J Rehabil Res Dev 1998;35:6–13.
36. Soderberg B, Ryd L, Person BM. Roentgen stereophotogrammetric analysis of motion between the bone and the socket in the transtibial amputation prosthesis: a case study. J Prosthet Orthot 2003;15:95–99.
37. el Nahass B, Madson MM, Walker PS. Motion of the knee after condylar resurfacing—an in vivo study. J Biomech 1991;24:1107–1117.
38. Kadaba MP, Ramakrishnan HK, Wootten ME. Measurement of lower extremity kinematics during level walking. J Orthop Res 1990;8:383–392.
39. Stein A, Fleming B, Pope MH, Howe JG. Total knee arthroplasty kinematics. An in vivo evaluation of four different designs. J Arthroplasty 1988;3 (suppl):S31–S36.
40. Verstraete MC, Soutas-Little RW. A method for computing the three-dimensional angular velocity and acceleration of a body segment from three-dimensional position data. J Biomech Eng 1990;112:114–118.
41. Banks SA, Hodge WA. Accurate measurement of three-dimensional knee replacement kinematics using single-plane fluoroscopy. IEEE Trans Biomed Eng 1996;43:638–649.
42. Dennis DA, Komistek RD, Walker SA, et al. Femoral condylar lift-off in vivo in total knee arthroplasty. J Bone Joint Surg Br 2001;83:33–39.
43. Hirokawa S, Ariyoshi S, Hossain AM. A 3D kinematic measurement of knee prosthesis using x-ray projection images. JSME Int J. Series C 2005;48.
44. Lafortune AM, Cavanagh RP. The measurement of normal knee joint motion during walking using intracortical pins. In: Whittle M, Harris D, eds. Biomechanical Measurements in Orthopaedic Practice. Oxford, U.K.: Clarendon; 1985;234–243.
45. Murphy CM, Zarins B, Jasty M, Mann WR. In vivo measurement of the three dimensional skeletal motion at the normal knee. Trans Orthop Res Soc 1985:142.
46. Stehling MK, Turner R, Mansfield P. Echo-planar imaging: magnetic resonance imaging in a fraction of a second. Science 1991;254:43–50.
47. Thomuson OW, Thaete LF, Fu HF, Dye FS. Tibial meniscal dynamics using three-dimensional reconstruction of magnetic resonance images. Am J Sports Med 1991;19:210–216.
48. Woltring JH, Roy VP, Hebbelinck M, et al. 3-D knee joint kinematics by magnetic resonance imaging. J Biomech 1990;23:384.
49. Berthonnaud E, Herzberg G, Zhao KD, et al. Three-dimensional in vivo displacements of the shoulder complex from biplanar radiography. Surg Radiol Anat 2005;27:214–222.
50. Bingham J, Li G. An optimized image matching method for determining in-vivo TKA kinematics with a dual-orthogonal fluoroscopic imaging system. J Biomech Eng 2006;128:588–595.
51. Blankevoort L, Huiskes R, de Lange A. The envelope of passive knee joint motion. J Biomech 1988;21:705–720.
52. Hanson G, Suggs J, Freiberg A, et al. Investigation of in vivo 6dof total knee arthoplasty kinematics using a dual orthogonal fluoroscopic system. J Orthop Res 2006;24:974–981.
53. Lemieux L, Jagoe R, Fish DR, et al. A patient-to-computed-tomography image registration method based on digitally reconstructed radiographs. Med Phys 1994;21:1749–1760.
54. Li G, Wuerz T, DeFrate L. Feasibility of using orthogonal fluoroscopic images to measure in vivo joint kinematics. J Biomech Eng 2004;126:314–318.
55. Nilsson KG, Karrholm J, Ekelund L. Knee motion in total knee arthroplasty. A roentgen stereophotogrammetric analysis of the kinematics of the Tricon-M knee prosthesis. Clin Orthop Relat Res 1990;(256):147–161.
56. Papaioannou G, Fiedler G, Mitrogiannis C, Nianios G. Using high accuracy biplane dynamic roentgen stereogrammetric analysis to assess interface proximity residual bone-stump-skin-socket kinematics of above knee amputees during strenuous activities. International Society of Biomechanics, XXII Congress, Cape Town, South Africa, July 5–9, 2009.
57. Papaioannou G, Mitrogiannis C, Nianios G, Fiedler G. The use of dynamic biplane roentgen stereophotogrammetric analysis (DRSA) as a new diagnostic paradigm in orthopedics. Fifth International Conference on Health Care Systems, Milwaukee, WI, October 13–15, 2008.
58. Papaioannou G, Mitrogiannis C, Nianios G, Fiedler G. A new improved tracking technique for assessment of high resolution dynamic radiography skeletal kinematics. 55th Annual Meeting Orthopaedic Research Society, Las Vegas, NV, February 22–25, 2009.
59. Papaioannou G, Mitrogiannis C, Nianios G, Fiedler G. A new method for assessing residual limb skin-tissue strain during above-knee amputee high-speed movement. 55th Annual Meeting Orthopaedic Research Society, Las Vegas, NV, February 22–25, 2009.
60. Papaioannou G, Mitrogiannis C, Nianios G, Fiedler G. Assessing residual bone-stump-skin-socket interface kinematics of above knee amputees with high accuracy biplane dynamic roentgen stereogrammetric analysis. 55th Annual Meeting Orthopaedic Research Society, Las Vegas, NV, February 22–25, 2009.
61. Papaioannou G, Nianios G, Mitrogiannis C, Fiedler G. An improved tracking technique for assessment of high resolution dynamic radiography kinematics. Comput Model Eng Sci 2008;392:931–936.
62. Tashman S, Anderst W. In vivo measurement of dynamic joint motion using high speed biplane radiography and CT: application to canine ACL deficiency. J Biomech Eng 2003;125:238–245.
63. Bey MJ, Zauel R, Brock SK, Tashman S. Validation of a new model-based tracking technique for measuring three-dimensional, in vivo glenohumeral joint kinematics. J Biomech Eng 2006;128:604–609.
64. You B-M, Siy P, Anderst W, Tashman S. In-vivo measurement of 3D skeletal kinematics from sequences of biplane radiographs: application to knee kinematics. IEEE Trans Med Imag 2001;20:514–525.
65. Anderst W, Zauel R, Bishop J, et al. Validation of three-dimensional model-based tibio-femoral tracking during running. Med Eng Phys 2009;31:10–16.
66. Papaioannou G, Nianios G, Mitrogiannis C, et al. Patient-specific knee joint finite element model validation with high accuracy kinematics from biplane dynamic roentgen stereogrammetric analysis. J Biomech 2008;41:2633–2638.
67. Papaioannou G, Tashman S. Validation of lower limb model using 3D in vivo femoral kinematics from biplane radiorgaphy. In: Dassios G, Fotiadis DI, Kiriaki k, Massalas cV, eds. Scattering Theory and Biomedical Engineering Modeling and Applications. New Jersey: World Scientific; 2001:179–209.
68. Tashman S. Comments on “validation of a non-invasive fluoroscopic imaging technique for the measurement of dynamic knee joint motion”. J Biomech 2008;41:3290–3291.
69. Papaioannou G, Mitrogiannis C, Nianios G, Fiedler G. Tracking high speed arthrokinematics using a new and high resolution biplane dynamic roentgen stereogrammetric method. Int J Imag 2009;2:66–85.
70. Bey MJ, Kline SK, Zauel R, et al. Measuring dynamic in-vivo glenohumeral joint kinematics: technique and preliminary results. J Biomech 2008;41:711–714.
71. Arora SJ. Idesign ARORA. Introduction to Optimum Design. 2nd ed. Orlando, FL: Academic Press; 2004.
72. Jonkers I, Spaepen A, Papaioannou G, Steward C. Contribution of forward simulation techniques to the understanding of muscle activation patterns in mid-stance phase of gait. J Biomech 2002;35:609–619.
73. Papaioannou G. A Three Dimensional Mathematical Model of the Knee Joint. Ph.D Thesis, Glasgow, Scotland, UK: University of Strathclyde Bioengineering Unit; 1999.
74. Parker JR. Algorithms for Image Processing and Computer Vision. New York: Wiley, Paper/Cdr Edition; 1996.
75. Grood ES, Suntay WJ. A joint coordinate system for the clinical description of three-dimensional motions: application to the knee. J Biomed Eng 2007;105:136–144.
76. ASTM. ASTM E 177-90a—Standard Practice for Use of the Terms Precision and Bias in ASTM Test Methods. West Conshohocken, PA: American Society for Testing and Materials; 1996.
77. Papaioannou G, Nianios G, Mitrogiannis C, et al. Patient specific knee joint finite element model validation with high accuracy kinematics from biplane dynamic radiography. Comput Model Eng Sci 2008;392:891–896.
78. Silver-Thorn MB, Childress DC. Parametric analysis using the finite element method to investigate prosthetic interface stresses for persons with trans-tibia amputation. J Rehabil Res Dev 1996;33:227–238.
79. Zhang M, Lord M, Turner-Smith AR, Roberts VC. Development of a nonlinear finite element modelling of the below-knee prosthetic socket interface. Med Eng Phys 1995;17:559–566.
80. Zhang M, Mak AFT. A finite element analysis of the load transfer between an above-knee residual limb and its prosthetic socket - Roles of interfacial friction and distal-end boundary conditions. IEEE Trans Rehabil Eng 1996;4:337–346.

amputee; TKA; internal prosthesis; socket-residual limb kinematics; dynamic radiography

© 2010 American Academy of Orthotists & Prosthetists