Journal of Computer Engineering & Information TechnologyISSN : 2324-9307

Reach Us +1 850 754 6199

Research Article, J Comput Eng Inf Technol Vol: 5 Issue: 1

The Construction of the Anterior Pelvic Plane using Landmarks Accessible in the Lateral Decubitus Position during Computer Assisted Hip Arthroplasty

Davis ET1,2, Palaparthy P1*, Smith GM1, Schubert M3, Wegner M3 and Haimerl M3
1The Dudley Group NHS Foundation Trust, Pensnett Road, Dudley, West Midlands, UK
2The Royal Orthopaedic Hospital NHS Foundation Trust, Bristol Road South, Northfield, Birmingham, UK
3Brainlab AG, R&D Surgery, Feldkirchen, Germany
Corresponding author: Prakash Palaparthy, MRCS
The Dudley Group NHS Foundation Trust, Pensnett Road, Dudley, West Midlands, UK
Tel: +447809609470
E-mail: [email protected]
Received: January 18, 2016 Accepted: March 01, 2016 Published: March 08,2016
Citation: Edward TD, Prakash P, Gemma MS, Mario S, Melanie W, et al. (2016) The Construction of the Anterior Pelvic Plane using Landmarks Accessible in the Lateral Decubitus Position during Computer Assisted Hip Arthroplasty. J Comput Eng Inf Technol 5:1. doi:10.4172/2324-9307.1000142

Abstract

The Construction of the Anterior Pelvic Plane using Landmarks Accessible in the Lateral Decubitus Position during Computer Assisted Hip Arthroplasty

Construction of the anterior pelvic plane during computer-assisted hip arthroplasty.

200 pelvic CT scans were used to validate a new methodology to construct the Anterior Pelvic Plane (APP), using anatomical landmarks easily palpated in the lateral decubitus position. Scans were also analysed to simulate inaccuracies of obtaining the APP through soft tissue.

Comparing the new methodology to the APP, error in acetabular inclination was 0.69° (SD=2.96) and anteversion was 1.17° (SD=3.53). This compared favourably to the error when the APP was registered through soft tissue; error in inclination of -0.92° (SD=0.26), anteversion of -5.24° (SD=2.09). Using this new methodology, acetabular placement was within the ‘safe zone’ >99.6% of cases. This study appears to show that by identifying anatomical constants the APP can be constructed using this new methodology and provide more accurate acetabular component placement.

Keywords: Computer assisted surgery; Hip Arthroplasty; Acetabular position

Keywords

Computer assisted surgery; Hip Arthroplasty; Acetabular position

Introduction

Acetabular component orientation in total hip arthroplasty and hip resurfacing is now of paramount importance. Numerous studies have shown the detrimental effects of abnormal component orientation in metal on metal hip arthroplasty [1-3]. However, all other bearings used in hip replacement have also been shown to be highly dependent on component orientation in their production of wear particles [4,5] as well is the risk of dislocation [6]. Traditional mechanical alignment guides have been shown to be highly variable in their ability to correctly orientate the acetabular component. In a review of total hip replacement and hip resurfacings [7], only 49% were found to be within the acceptable limit. More concerning is that this was only raised to 51% in the hands of high-volume arthroplasty surgeons [7]. These figures are not inconsistent with other published studies [8-11]. It is clear that there is a desperate need for technology to assist the surgeon in placing the components in the correct orientation. Unfortunately, computer navigated hip arthroplasty has been slow to gain acceptance. Reasons for this include the necessity to place fixation pins into the bone, line of sight issues related to optical tracking technology and the requirement to directly access the APP. Access to the APP can be a major obstacle when performing hip arthroplasty in the lateral decubitus position as the patient often requires repositioning. There also exists concern that the APP registration points are taken through adipose tissue, potentially introducing error, particularly in anteversion measurements [12,13].
Thus, there is high demand for robustly identifiable reference planes on the pelvis based on landmarks, which can be acquired purely on the treatment side. This requires constant relationships between the landmarks and the reference planes. If appropriate anatomical pelvic constants could be defined then it would be possible to reconstruct the APP using a point acquired on the ipsilateral ASIS, a mid-sagittal point (obtained through the drapes) and points acquired during the normal surgical exposure. This may encourage more surgeons to adopt this technology with improved component positioning, as all points can be accessed while the patient is finally prepared in a lateral decubitus position. It may also improve the accuracy when performing the surgery in the supine position as it eliminates the need for pubic point registration.
In this study, our objective was to validate anatomical constants from data obtained from routine CT scans of the pelvis and analyse their variation including anatomical as well as inter-individual variation in the identification of the measurements. We hypothesise that these constants can be used to construct new reference planes that can be defined by the surgeon when operating in the lateral decubitus position and can also improve the accuracy when operating in a supine position. The APP can still be reconstructed from this new reference planes to ensure a consistent understanding of cup orientation values.

Methods and Materials

CT data collection
Full ethical approval was obtained from the West Midlands Ethics Committee. Two hundred CT scans were retrospectively reviewed. The images were obtained from patients undergoing pelvic CT for conditions unrelated to their hips. For inclusion into the study, it was required that patients were greater than 20 years of age to ensure they had reached skeletal maturity. Patients were excluded if any of the following criteria were met:
Acute or previous pelvic fracture
Acute or previous acetabular fracture
Previous surgery causing distortion of the bony pelvis
Patients with hip arthroplasty in-situ
Incomplete scans not including the ASIS points and the pubic tubercles
Data analysis
Each CT scan was imported into the iPlan 3.0 software (Brainlab AG, Feldkirchen, Germany). In 20 randomly selected scans, three observers independently identified the anatomical landmarks to analyse inter-observer variation for the definition of the new planes as well as the standard APP. The observers consisted of one consultant arthroplasty surgeon and two surgical trainees. Each observer performed the planning in these scans. The remaining 180 CT scans were only processed by one observer to analyse the statistical variation of the landmarks and anatomical measurements. Each of the landmarks was initially identified on the 3D reconstruction and then more accurately positioned on the bony anatomy using the axial, sagittal, and coronal slices.
Gold standard definition
The first set of anatomical landmarks, comprising ASIS points on both sides and the anterior aspect of the symphysis pubis, was taken to define the APP which is considered as a standard reference for hip arthroplasty surgeries [6]. The midsagittal plane was defined as the plane perpendicular to the connecting line between both ASIS points. This coordinate system was then used as the “gold standard” with which to measure the accuracy of the new planes. The depth of soft tissue over the APP was also recorded. It was defined as the anteriorposterior distance between the skin and the bony landmarks, i.e. ASIS points and symphysis pubis. This information was used to estimate errors in clinical landmark acquisition on soft tissue. We assumed that the soft tissue over the pubis is compressed by 70%, i.e. the originally planned pubis points were virtually shifted anteriorly by 30% of the recorded soft tissue thickness. Additionally, we assumed that the soft tissue above the ASIS is compressed by 90% according to the lower amount of fat above this landmark. These values are consistent with the results about deviations in APP acquisition in the literature [13,14]. These soft tissue corrections were taken into account for the all types of registrations. We acknowledge that the errors in palpation will occur in all 3D degrees of freedom and not isolated to the AP plane.
Defining landmarks for new method
For defining the new planes and reconstructing the APP the following additional landmarks were taken. The spinous process of the L5 vertebra was used to define the position of the mid-sagittal plane. The deepest point of the acetabular fossa on both hips (fossa points) and the centre of rotation (CoR) of the hip were taken. The CoR was determined by the centre of a sphere placed over the femoral head. Additionally, the anterior rim of the acetabulum as well as the distance between both ASIS points was recorded.
These measurements enabled the calculation of the fossa to fossa distance, i.e. medial-lateral distance between the fossa points, and angles between the APP and alternative reference planes. These alternative planes only differed from the APP in a sagittal projection as shown in Figure 1. They still referred to the ASIS points as a basic reference for the medial-lateral direction; but the points at the symphysis pubis were replaced by information around the acetabulum, i.e. CoR (plane A1), deepest point at the fossa acetabuli (plane A2), and distance between an anterior rim point and the APP (d3) to define their sagittal orientation. The statistics of these anatomical constants were analysed including gender-specific statistics to reflect systematic differences between male and female pelvises.
Figure 1: Alternative reference planes and distance information shown in a sagittal projection. Plane A1 defined by ASIS points and CoR and A2 by ASIS points and deepest point at the fossa acetabuli. The third reference plane is defined by the sagittal distance d3 between the APP and the anterior rim of the acetabulum.
Definition of planes P1 to P4
Based on the taken landmarks, the following registration planes were created to analyse the accuracy of standard APP registrations and an indirect reconstruction of the APP based on the statistics of the analysed relationships, which were considered as anatomical constants. Subsequently, the planes are used synonymously with the associated coordinate systems, i.e. registrations. P1 was defined as the bony APP or the “gold standard” with which to compare the other planes to (Figure 2). This was the two ASIS and the symphysis pubis point directly defined on bone.
Figure 2: Landmarks which were used to define the gold standard (P1), i.e. ASIS and pubis points defined on bone and represented by small and dark dots, and the epicutaneous registration of the APP (P2), i.e. ASIS and pubis points defined on soft tissue and represented by big and light dots.
P2 is the plane that is normally acquired during a computer navigated hip arthroplasty. This plane is the two ASIS points and the symphysis pubis point acquired epicutaneously. Soft-tissue compression ratios were applied as described above to simulate the acquisition of soft tissue landmarks. We refer to P2 by using the name epicutaneous APP registration (Figure 2).
The amalgamation of the different reference points and anatomical constants were then combined to provide an algorithm, which reconstructs the APP using landmarks accessible, when the patient is in the lateral decubitus position. These points consisted of the ipsilateral ASIS (including soft tissue errors), the L5 spinous process, the anterior acetabular rim point, the acetabular fossa and the centre of hip rotation. This plane will be defined as P3 (Figure 3) and called lateral registration. The registration algorithm used a gender-specific value for the fossa to fossa distance to reflect differences between male and female pelvises.
Figure 3: Landmarks which were used to perform the lateral registration (P3), i.e. treated ASIS defined on soft tissue (red), spinous process of L5 vertebra (green), centre of hip rotation (light green), deepest point of the acetabular fossa (blue), and anterior acetabular rim point (purple).
A further plane can be defined in a situation where the patient is operated on in the supine position with access to both ASIS (including soft tissue errors) but where the use of the anterior acetabular rim point, centre of hip rotation and fossa points are used, instead of the symphysis pubis in an attempt to remove inaccuracies from the adipose tissue covering the symphysis pubis. This plane will be defined as P4 (Figure 4) and called (pubic-free) supine registration. No gender-specific parameters were used for P4.
Figure 4: Landmarks which were used to perform the pubic-free supine registration (P4), i.e. ASIS points defined on soft tissue (red), centre of hip rotation (light green), deepest point of the acetabular fossa (blue), and anterior acetabular rim point (purple).
Comparison to Gold Standard P1
For all types of registrations (i.e.P2-P4), the error in potential acetabular orientation was calculated in terms of the deviation of inclination/antiversion for a standard cup position. For this purpose, the cup was virtually placed in a position of 40° inclination and 15° anteversion [6] according to the APP defined on bone. The cup orientation was adapted according to the differences between the gold standard and the analysed registration. The resulting cup orientation was compared with the originally intended orientation. These values were then used to quantify the difference in component position from the “gold standard” (bone APP) to the virtual position, which would be obtained using the different registration methods. The calculation of registration results was performed for each treatment side on each patient, resulting in 400 test cases. The inter-observer variation in the definition of the “gold standard” was obtained in the same way. All values for inclination/anteversion were specified in the radiographic definition [15].
For the evaluation of the registration results, the anatomical constants in the registration algorithm were taken from a different data set to ensure independence between the study data and the constants in the registration algorithm [16].
Descriptive statistics, i.e. mean, standard deviation (SD), and 95% confidence intervals (CI), were calculated for each of the measurements including anatomical constants (i.e. fossa to fossa distance as well as deviation between alternative reference planes and APP), inter-observer variability, and deviations between the gold standard and the registration methods. Additionally, an estimation of the number of cases, which would have had cup positions outside the Lewinnek safe zone [6], was performed. The actual numbers of outliers as well as statistical intervals assuming normally distributed deviation values were determined. Student’s two sample t-tests assuming equal variances (p-value 0.05) were applied to determine statistical differences between male and female groups as well as differences between the distinguished registration types. The square root of the mean squared differences (RMSD) was used for analysing interobserver variability. Additionally, modified correlation and Bland- Altman type diagrams were used to summarize the inter-variation of the gold standard between all three observers in one graph. All statistics were calculated by using the R software package (version 2.13.0, R Foundation for Statistical Computing, Vienna, Austria). The values in Table 1, were then utilized to indirectly reconstruct the APP according to the lateral (P3) and pubic-free supine registration (P4).
Table 1: Statistics of anatomical constants and measurements.

Results

Patient population
One hundred and one of the patients were male (50.5%) and 99 were female (49.5%). The mean age was 69 years (range 22-95 years). The majority of the patients were of Caucasian origin (92%) with 5% Afro-Caribbean and 3% of Asian origin. The images demonstrated osteoarthritis in 17 patients (9%) and the presence of developmental dysplasia in 2 patients (1%) (Defined, as a centre edge angle of <20 degrees).
Comparison to gold standard
Table 1 shows the relationship between the APP (P1) and the alternative reference planes defined by the two ASIS points and the centre of hip rotation (A1) as well as deepest point in the fossa acetabuli (A2). No statistically significant differences could be found between male and female pelvises. The distance d3 from the anterior acetabular rim and the APP appeared to be relatively constant with a SD of only 3.8 mm. There was a statistically significant difference between male and female pelvises. But the difference between the mean values was below 1mm and the SD was almost equal. The table also demonstrates the fossa to fossa distance in males and females which was significantly different. But, it was consistent within gender with a SD of 8.13 mm and 8.16 mm.
The soft-tissue thickness above pubis was in males and females significantly different, but above the ASIS points consistent within gender (see Table 1). After compression, the resulting mean soft tissue thickness over the ASIS was 2.9 mm (SD 2.2 mm) and over the pubis it was 13.1 mm (SD 5.6 mm).
Inter-Observer Variation
The variation between observers (RMSD) for the measured anatomical constants is also reported in Table 1 (third data column). The measurements could be performed consistently with the angle between the APP and the plane A2 (defined by the ASIS and deepest fossa points) showing the highest inter-observer variation (RMSD=2.19°). The inter-observer variation (RMSD) of the gold standard APP was 0.97° for inclination and 1.20° for anteversion. The case-by-case results for this gold-standard are shown graphically in Figure 5 by using correlation and Bland-Altman type diagrams. The RMSD for rotations around the left-right axis (i.e. rotations in a sagittal plane) was 1.68°.
Figure 5: Interobserver variation of the gold standard (P1). Top row: correlation between the observers for inclination (left) and antiversion (right). The straight line represents an ideal correspondence. Bottom row: Bland-Altman type comparison between the measurements for inclination (left) and anteversion (right).Differences in the measurements of the observers are shown on the y-axis, average values from all three observers are given on the x-axis. The limits of agreement in terms of a 95% CI (horizontal lines) were calculated assuming that the SD of the error distribution was given by the RMSD. In all diagrams,each data series (cross, triangle, circle) represents one pair of observers. The results for each particular data set are linked with a line to provide a visual representation of this connection.
Simulation of clinical data (soft tissue simulation)
Table 2, first row, demonstrates the error that can be expected when defining the APP through soft tissue (P2) in an attempt to replicate the intra-operative scenario. The error was defined as the difference in acetabular inclination and anteversion when compared to the bony APP (P1). Table 2, second and third row, describes the statistics of the deviations for P3 and P4 in comparison to the gold standard (P1). The differences for the epicutaneous APP (P2) compared to both new types of registration (P3 and P4) was found to be statistically significant for inclination and anteversion with p-values <0.001. The lateral (P3) and pubic-free supine (P4) registration were not significantly different with p-values of 0.105 for inclination and 0.132 for anteversion.
Table 2: Simulation of clinical data (soft tissue simulation).

Discussion

This study demonstrates that it is possible to reconstruct the anterior pelvic plane with registration points obtained with the patient fully prepared in the lateral decubitus position. This is achieved by using anatomical constants which allow to reconstruct the APP only from these landmarks.
Other authors have examined for constant geometries within the human pelvis. Obstetricians have shown that the diameter of the pelvic cavity is relatively constant [17]. They found similar results for the transverse diameters, where the definition of these measurements differed from our approach using the fossa-to-fossa distance, which was directed to the clinical purpose. In particular, a statistically significant difference between genders was found analogously. Thornberry et al. [18] described the construction of a plane utilising the two anterior superior iliac spines and the hip centre of rotation in an attempt to develop a form of pelvic registration without the need for registering pubis points. They compared this plane with the traditional APP in 36 patients using CT scans. They found the angle to be 36.5° ± 3.5°. This is slightly lower than the angle found in our study, which was 39.83° ± 4.46°. However, the use of this plane still relies on the ability to identify the contralateral ASIS, which can be challenging when the patient is fully draped in the lateral decubitus position. Relying on solely the centre of rotation of the hip also raises concern, when dealing with arthritic joints causing deformity of the acetabular cavity. In this study, we have combined the centre of hip rotation with the anterior acetabular rim point and the fossa points in an attempt to provide a more robust algorithm, which will be more resistant to errors introduced by acetabular deformity.
The inter-observer variation for the differences between the alternative reference planes and the APP was found to be between 1.5° and 2.2° (RMSD). It was similar to the RMSD of the sagittal alignment of the gold standard (1.68°). For the distance between the anterior rim of the acetabulum and the APP as well as the fossa to fossa distance, the RMSD was below 1.8 mm. This shows a high consistency between the observers. An inter-observer variation of approximately 1° for inclination and 1.2° for anteversion was determined for the definition of the gold standard APP directly defined on bone. This limitation should be considered when assessing the accuracy of a registration method, i.e. a registration method cannot be more accurate than these limits.
Our results show that using the combined acetabular landmarks along with the ipsilateral ASIS and defining the mid-sagittal point; the mean error in acetabular inclination for the lateral registration is 0.69° (SD 2.96°) and in acetabular anteversion 1.17° (SD 3.53°). Additionally, the data also shows that in more than 99.6% of cases the acetabular placement would have been within a ±10° safe zone as described by Lewinnek [6]. This study also appears to demonstrate that a combination of acetabular registration points can be used when undertaking hip arthroplasty in the supine position to potentially improve the accuracy of computer navigation. The mean error in acetabular inclination was 0.44° (SD 0.77°) and the mean error of acetabular anteversion 1.51° (SD 2.94). Statistically, 99.8% of the results were in the safe zone.
The comparison between the different types of registrations showed, that the number of outliers was highest in the epicutaneous registration of the APP. This applied to the numerical count of outliers as well as the statistical calculations for the accuracy intervals. Based on the high number of analysed CT data sets, statistically significant differences could be found between the epicutaneous APP registration (P2) and the other registration types. For anteversion measurements, these differences were basically caused by the differences in the mean values (bias) for the deviation between the particular registrations and the gold standard. For example, a bias of more than 5° was observed for anteversion measurements according to P2 whereas this bias could be reduced to 1.2° and 1.5° when using P3 and P4. The systematic bias according to P2 is basically caused by the soft tissue above the symphysis pubis and well known in literature [19,13,14]. For example, Sendtner et al. [14] reported a mean error of 0.37° (SD 3.26) for inclination and -5.61° (SD 6.48) for anteversion and Ybinger et al [13] 3.5° (SD 4.4°) for inclination and -6.5° (SD 7.3°) for anteversion.
The standard deviations for P2 in our study appeared to be considerably better than they are reported in these studies. For the inclination measurements, this seemed to be caused by the fact, that we exactly used the same medial-lateral reference line (i.e. connecting line between both ASIS points) for P2 and the gold standard. This also applies to P4. For anteversion measurements in P2, the assumption of fixed soft tissue compression ratios of 30% obviously produces better results in our theoretical simulation. Higher variations in the compression ratio would be expected in a clinical scenario and would increase the standard deviation for anteversion errors in P2 as e.g. reported in the referenced studies [13,14]. This also inhibits options to automatically correct anteversion measurements in P2, which could be considered to compensate the systematic bias. For example, techniques using stab incisions for the pubis points as used by Dorr et al. [20] would be degraded. Additionally, Ybinger et al. [13] showed that the correlation between soft tissue thickness above the pubis and errors in anteversion measurements is only moderate. Thus, compensation could not fully address the dependency between soft tissue and registration accuracy for P2.
Of course, variations in the clinical landmark acquisition will also decrease the accuracy of the measurements for P3 and P4 to a certain degree. However, most of the landmarks for these registration methods are acquired directly on bone (acetabular points) or close to the bone (ASIS points) and can presumably be detected with a lower deviation. Additionally, the L5 landmark is only used to define a point on the midsagittal plane. Thus, a point acquisition away from the bone because of soft tissue above the landmark does not directly impact the final accuracy. In summary, the new registration methods seem to have the potential to perform favourable to the epicutaneous registration of the APP, which is currently used in most navigation systems. In particular, this applies to anteversion measurements.
This study does have inherent limitations in that the CT scans were obtained from patients with relatively normal anatomy and that the registration procedures were simulated by using anatomical landmarks defined in the CT data sets directly on bone. Thus, individual variations of the landmark acquisition steps in a clinical setup could not be covered. Osteoarthritis is known to cause anterior osteophytes and deform the femoral head and acetabulum which may lead to registration errors. In the clinical setting, the centre of rotation will have to be determined using points on the acetabular surface instead of defining a sphere around the femoral head. The overall accuracy will therefore need to be verified in further clinical studies on patients with osteoarthritis, dysplasia and acetabular protrusion which include these effects.
Further studies will be required to identify the potential error in the calculation of the centre of rotation and L5 points in the clinical setting. Additionally, the compression ratios for the soft tissue above the pubis and ASIS points could not directly be measured. However, the used compression ratios match with clinical results regarding the inaccuracy of epicutaneous APP registrations [13,14]. Our estimations slightly underestimates the relation between the soft tissue at the pubis in relation to the ASIS and thus prefers the P2 results, as the bias for the anteversion errors is higher in the clinical studies in comparison to our results. This can also be directly derived from the statistics of soft tissue thickness in our study where a 3.5:1 (instead of a 3:1) relationship could finally be established to obtain the average bias of 6° as identified in the clinical studies [13,14]. Thus, the compression ratios were considered as cautious estimates which should assure that the new registration methods should not inadequately be favoured.
In summary, this study appears to provide evidence suggesting that with the use of anatomical landmarks that are accessible at the time of hip arthroplasty surgery with the patient in the lateral decubitus position, it may be possible to define a plane that can more accurately position an acetabular component when compared to conventional mechanical guides. It also suggests that the use of acetabular landmarks and both ASIS may provide an improvement in the accuracy of computer navigated hip arthroplasty when compared to the traditional epicutaneous registration of the APP. Further studies were performed to validate these theoretical constants in the clinical setting. Initial results obtained in a series of cadaver studies and a prospective clinical study [21,22] confirmed that it is possible to accurately reconstruct the APP based on the analysed anatomical relationships also in a clinical environment.

Acknowledgement

We would like to acknowledge the work by two of our colleagues, Poitzsch L who helped us in evaluations and Kling S who supervised the overall research project on lateral pelvis registration.
We would also like to acknowledge the support provided by Russell’s Hall Hospital, Dudley Group NHS Foundation Trust in carrying out this project.

References























Track Your Manuscript

Share This Page