Access to this full-text is provided by Frontiers.
Content available from Frontiers in Surgery
This content is subject to copyright.
EDITED BY
Ronald Mark Gillies,
Medical Device Research Australia, Australia
REVIEWED BY
Rongshan Cheng,
Shanghai Jiao Tong University, China
Ye Ye,
Luoyang Orthopedic Traumatological Hospital,
China
*CORRESPONDENCE
Yeon Soo Lee
biomechanics.yslee@gmail.com
Seung-Hoon Baek
sbaek@knu.ac.kr
SPECIALTY SECTION
This article was submitted to Orthopedic
Surgery, a section of the journal Frontiers in
Surgery
RECEIVED 19 July 2022
ACCEPTED 21 November 2022
PUBLISHED 26 December 2022
CITATION
Min Lee J, Baek S-H and Soo Lee Y (2022) Vital
protocols for PolyWare™measurement
reliability and accuracy.
Front. Surg. 9:997848.
doi: 10.3389/fsurg.2022.997848
COPYRIGHT
© 2022 Min Lee, Baek and Soo Lee. This is an
open-access article distributed under the terms
of the Creative Commons Attribution License
(CC BY). The use, distribution or reproduction in
other forums is permitted, provided the original
author(s) and the copyright owner(s) are
credited and that the original publication in this
journal is cited, in accordance with accepted
academic practice. No use, distribution or
reproduction is permitted which does not
comply with these terms.
Vital protocols for PolyWare™
measurement reliability and
accuracy
Jong Min Lee1, Seung-Hoon Baek2*and Yeon Soo Lee1*
1
Department of BioMedical Engineering, School of BioMedical Science, Daegu Catholic University,
Gyungbuk, South Korea,
2
Department of Orthopedic Surgery, School of Medicine, Kyungpook
National University, Kyungpook National University Hospital, Daegu, South Korea
Background and objective: PolyWare™software (PW) has been exclusively
used in the majority of polyethylene wear studies of total hip arthroplasty
(THA). PW measurements can be significantly inaccurate and unrepeatable,
depending on imaging conditions or subjective manipulation choices. In this
regard, this study aims to shed light on the conditions needed to achieve the
best accuracy and reliability of PW measurements.
Methods: The experiment looked at how PW fluctuated based on several
measurement conditions. x-ray images of in-vitro THA prostheses were
acquired under a clinical x-ray scanning condition. A linear wear rate of
6.67 mm was simulated in combination with an acetabular lateral inclination
of 36.6° and anteversion of 9.0°.
Results: Among all the imported x-ray images, those with a resolution of
1,076 × 1,076 exhibited the best standard deviation in wear measurements as
small as 0.01 mm and the lowest frequencies of blurriness. The edge
detection area specified as non-square and off the femoral head center
exhibited the most blurriness. The x-ray image that scans a femoral head
eccentrically placed by 15 cm superior to the x-ray beam center led to a
maximum acetabular anteversion measurement error of 5.3°.
Conclusion: Because PW has been the only polyethylene wear measurement
tool used, identifying its sources of error and devising a countermeasure are
of the utmost importance. The results call for PW users to observe the
following measurement protocols: (1) the original x-ray image must be a
1,076 × 1,076 square; (2) the edge detection area must be specified as a
square with edge lengths of 5 times the diameter of the femoral head,
centered at the femoral head center; and (3) the femoral head center or
acetabular center must be positioned as close to the center line of the x-ray
beam as possible when scanning.
KEYWORDS
total hip arthroplasty (THA), PolyWare (PW), polyethylene wear, anteversion, lateral
inclination
1. Introduction
Wear debris-induced osteolysis and implant loosening are the primary causes
limiting implant longevity after total hip arthroplasty (THA) (1,2). Additionally,
proper acetabular cup (AC) placement in THA is essential to reduce implant wear
and dislocation. Thus, early detection of the complications via accurate measurement
TYPE Original Research
PUBLISHED 26 December 2022
|
DOI 10.3389/fsurg.2022.997848
Frontiers in Surgery 01 frontiersin.org
of wear rate and AC alignment during routine check-ups is of
paramount clinical value (3–8).
Previous studies have demonstrated the high accuracy of
PolyWare™software (PW) in measuring wear rate or cup
orientation (9). Even though reliable interactive computerized
methods for measurements based on 2D AP x-ray images or
2D-3D registration methods have been proposed (7,10), the
majority of them have not been commercialized. In contrast,
for decades PW has been the only commercially available tool
to quantify THA polyethylene wear, due to its ease of use and
lack of need for bead insertion or dual x-ray scanners.
Because PW matches 3D sphere models representing the AC
and femoral head (FH) onto the silhouettes of the AC and
FH on x-ray images, it can measure the anteversion and the
lateral tilt of the AC alongside polyethylene wear.
However, we found that PW measurement results can be
significantly inaccurate depending on factors such as the
observer’s technical preferences and the features of x-ray
images. Various error messages have frequently been
encountered during our PW measurements due to unknown
causes and PW spontaneously shutting down during
measurements. The authors have categorized these errors into
intrinsic and extrinsic, according to their dependency on PW
performance. We believe that some errors can be reduced by
optimizing the observer’s choices or skill: Ext1)PW’s extrinsic
error as a result of the original x-ray images being imported
at an improper size; Ext2)PW’s extrinsic error as a result of
the object’s eccentric location away from the x-ray source-to-
detector center line; Int1)PW’s intrinsic error, i.e., PW’s
functional limitation which is unable to fix the measurement
error due to the asymmetrical specification of the edge
detection area.
Because PW has been the only polyethylene wear
measurement tool used, identifying the sources of its errors
and developing a countermeasure is critical for THA research.
In this regard, the current study has two aims. The first is to
experimentally assess PW’s extrinsic and intrinsic errors (Ext1,
Ext2,andInt1). The second is to provide three technical
empirical guidelines that clinicians or researchers can use.
2. Materials and methods
2.1. Study design
The experiments parametrically investigated the effects of
three potential error-causing factors: the size of the original
x-ray image (S), the eccentric placement of the THA implants
with respect to the x-ray source-to-detector center line (E),
and the geometric characteristics of edge detection area
definition (G). The S,E, and Gfactors correspond to Ext1,
Ext2, and Int1, respectively. Figure 1 shows the overall layout
of the current study. To ensure the highest level of reliability
for PW measurements, the three best parameters for S,E, and
Gwere ultimately identified.
2.2. Materials
2.2.1. THA prosthesis
The employed THA prosthesis set was composed of a
Biolox® Delt ϕ28 mm femoral head (CeramTec®, Plochingen,
Germany), a Trilogy® ϕ58 mm acetabular cup (Zimmer
Biomet®, Warsaw, IN, USA), a Bencox® stem (CorenTec®,
Cheon-An, Korea), and a Longevity® liner (Zimmer Biomet®,
Warsaw, IN, USA). According to the authors’experience with
PolyWare measurements, the edge detection of the prostheses
in x-ray images was independent of the prosthesis size. The
majority of THA femoral heads have sizes between 26 and
36 mm, large enough to accurately detect the edge of the
prostheses and locate the femoral head and acetabular
component centers.
2.2.2. Wear measurement software
A software called PolyWare™, v.8 (Draftware Inc., IN, USA)
for radiographic measuring was used for evaluation. A PW
measurement compares the analysis results of any two follow-
up times. Figure 2 shows the measurement process for PW.
The follow-up times can be postop (1–14 days from THA),
intervals of 3 months, 6 months, 1 year, and annual
increments after that. The results of the analysis include the
polyethylene liner wear and the anteversion and lateral
inclination of the AC. The liner wear is calculated as the
difference in distance between the FH center and the AC
center from the initial to the final follow-up times. The initial
and final follow-up times in a PW measurement correspond
to earlier and later, respectively.
2.2.3. x-ray images
The images of the THA prostheses were obtained using a
clinical x-ray scanner (Innovision SH, DongKang Co., Rep.
Korea). The perpendicular distance from the x-ray beam source
to the detector panel was fixed at 115 cm. These scanning
conditions were maintained because nonuniformity in the
distance or scanning direction of the beam source to the
detector can lead to different results. All x-ray images were first
acquired in DICOM format at a resolution of 3,020 × 3,020
pixels. They were converted to TIFF format because PW
software v.8 only analyzes TIFF images or converts DICOM
images into TIFF ones automatically inside the software.
2.2.4. Computers
The incidence of errors in PW work may be affected by
computer performance. In this regard, a laptop PC and a
desktop PC with different performance levels were tested
(Table 1).
Min Lee et al. 10.3389/fsurg.2022.997848
Frontiers in Surgery 02 frontiersin.org
2.2.5. Experimental simulation setup for
polyethylene wear and AC alignment
Wear was replicated by translating the femoral component. The
initial position of the prosthesis matched the condition in which the
FH fully contacts the AC, while the final position was intended as a
translation of the FH by 6.67 mm along the normal direction to the
equatorial plane of the AC. x-ray images were collected before
(initial) and after (final) the translation of the FH component
FIGURE 1
Overall process scheme of the current study.
FIGURE 2
PolyWare measurement workflow.
Min Lee et al. 10.3389/fsurg.2022.997848
Frontiers in Surgery 03 frontiersin.org
(Figure 3). To secure the spatial link between the FH and the AC at
the initial and final positions during the x-ray, alginate, an irreversible
hydrocolloid, was used. Alginate powder and water were mixed in a
plastic case. The mixture was left at room temperature up until the
alginate started to solidify. The components of the hip prosthesis
were then positioned over the alginate. The alginate foam
hardened into the native shape of the prosthetic frame in 1 min.
2.2.6. Measurement of true polyethylene wear
and AC alignment
A CAD measurement was used to determine the true
translation of the simulated wear. The original x-ray images of
resolution 3,020 × 3,020, with the FH center located at their
center, are imported into CAD software, Solidworks (Dassault
Systèmes, Vélizy-Villacoublay Cedex, France). The change in the
intercenter distance between the femoral head and the acetabular
cup was used to calculate polyethylene wear with respect to the
known diameter of the FH. Additionally, the lateral tilt of the
acetabular cup was calculated as the angle between the horizontal
line (also known as the medial-lateral line) and the line
connecting the medial-most and lateral-most points (Figure 4).
AC anteversion was calculated with the Lewinnek method (11).
ThetruetranslationoftheFHwas6.67mm,andthetruelateral
inclination and AC anteversion were 36.6° and 9.0°, respectively.
2.3. Compatibility of x-ray image sizes
with PW
2.3.1. Image loading error
When loading the x-ray images into PW, all x-ray images with
a resolution of 3,020 × 3,020 or higher led to an error message.
This was known as an “image loading error.”Imagesizeis
determined by several parameters, such as file format, level of
color/gray expression, and resolution. Because all of the x-ray
images in our study were in TIFF format with a 256 grey level,
the only parameter affecting image size was resolution. Various
image resolutions were tested to assess their compatibility with
PW during the image loading process. The original x-ray image
had a resolution of 3,020 × 3,020 and captured the FH at its
center. It was subsequently shrunk to several lower-resolution
images, the lowest being 1,024 × 1,024 (Table 2).
2.4. Effect of spatial eccentricity of the
objects in the original x-ray images
2.4.1. Test setups for spatial eccentricity modes
The distance from the x-ray beam source to an object grew
as it moved away from it on a transverse plane, yet the
TABLE 1 Specifications of the laptop and the desktop personal computers (PCs).
Manufacturer, model OS RAM CPU Memory Graphics
Laptop PC Laptop PC NT270E5R, Samsung Electronics
Co., Ltd., Suwon, South Korea.
Windows 7
(32bit)
8 GB Intel Core i5
4200U
DDR 3 8 GB Intel HD Graphics 4400, Shared
memory
Desktop
PC
Desktop PC, Custom-built Windows 10
(64bit)
16 GB Intel Core i7
4930K
DDR 3 16 GB NVIDIA GeForce GTX 750, 1GB
FIGURE 3
x-ray images of the initial (left) and final (right) positions, simulating cup wear by a 6.67 mm translation of the femoral stem normal to the equator
plane of the AC.
Min Lee et al. 10.3389/fsurg.2022.997848
Frontiers in Surgery 04 frontiersin.org
perspective viewing angle of the object field decreased (12). As a
result, the object’s silhouette shape was projected differently on
a detector plane, and PW measurements would give different
results. We defined spatial eccentricity as the translational
deviation of the FH center from the original x-ray image’s
middle on the same plane normal to the vector passing the x-
ray source and detector centers.
Nine spatial eccentricity modes were set up via
translating the THA prosthesis on the x-ray detector. With
respect to the central location mode (O), other eight
modes were specified via translating the prosthesis by
15 cm in left, right, superior, and/or inferior directions
relative to the center placement mode (O)(Figure 5). The
central mode (O) indicates the location of the center of
the FH within the x-ray beam. All of the x-ray images
used for the eccentricity tests had a resolution of 1,076 ×
1,076. Without applying any rotation, the same wear of
6.67 mm was reproduced in each of the nine modes. The
angular alignments of AC and acetabular liner wear should
be measured at the same values because the prosthesis was
only translated without rotation at all nine eccentricity
modes.
2.5. PW compatibility of geometric
features of the user-specified edge
detection area
The pre-processing step termed “a pre-processing
anteroposterior (AP) image”removes the superfluous region
from the initially loaded AP x-ray images for measurements
in a set of PW analyses. When a user assigns a rectangular
area by dragging the cursor from a point to its matching
diagonal point, PW magnifies the interior of the rectangle to
the size of a full working window. This step only assigns the
regions required for FH and AC edge detection, allowing for a
more accurate, quicker analysis. Following this, PW performs
edge detection for this rectangular area.
2.5.1. Blurring of the edge detection area
Even though images were loaded into PW without any
errors, PW occasionally returned a blur in the selected
region during the pre-process AP step. The blur was
intuitively recognizable, as in Figure 6. However, the
condition in which the image blur occurs is not revealed.
Standard imaging did not change the gray expression of the
original x-ray image. By contrast, the blurred imaging
rendered the entire edge detection area of the gray
expression considerably whiter and blurrier. It was necessary
to prevent the circumstances leading up to the blur.
Numerous tests indicated that the placement of the user-
specified edge detection area significantly affected the
blurring. The frequency of the blur decreased when the
center of the detection area was set as being closer to the FH
center. Consequently, we hypothesized that the image blur is
directly affected by the location of the FH in the edge
detection area. Therefore, the following three configurations
of the edge detection area were set up (Figure 7).
- Head-centered 5D
h
×5D
h
square: the first configuration
involves assigning the area as a square with edge lengths
corresponding to five times the diameter of the FH (D
h
)
and centered at the center of the FH component.
- Head-centered 7D
h
×7D
h
square: the second configuration
has the same profile as that of the first method, although
its edge lengths are seven times the diameter of the FH
component (D
h
).
- Not head-centered, non-square: the final configuration is a
random specification because it is neither square-shaped
nor centered at the FH center. The non-square
specification indicates that the observer specifies the areas
in non-squared rectangles and improvised sizes.
FIGURE 4
Measured values for PolyWare evaluation. (A) AC liner wear, (B) AC lateral tilt, and (C) AC anteversion.
Min Lee et al. 10.3389/fsurg.2022.997848
Frontiers in Surgery 05 frontiersin.org
For this edge detection area specification test, x-ray images
with a resolution of 1,076 × 1,076 were used. The image
resolution of 1,076 × 1,076 was selected because it was found
to be the most compatible resolution with PW (presented in
the “Results”section).
3. Results
3.1. Image loading error vs. loaded
image size
Concerning the image loading error, images with a
resolutionequaltoorhigherthan 1,800 × 1,800 frequently
failed while loading the initial or final x-ray images
(Table 2). Each resolution image was tested ten times. All
the images of resolutions corresponding to 2,494 × 2,494,
2,780 × 2,780, or 3,020 × 3,020 failed at being loaded into
PW, i.e., the loading error rate was 10/10 = 1. The image
loading error rate for 1,800 × 1,800 resolution images
was 8/10. Conversely, all images with a resolution of
1,500 × 1,500 or lower were successfully loaded into PW
with no errors.
In terms of occurrence rate, the image loading error was
identical for the desktop PC and laptop PC (Table 2).
Therefore, the PW image loading error did not depend on
computer performance.
3.2. Blurring of the edge detection image
vs. loaded image size
Only images that had been successfully loaded in PW could
be used in the edge detection process. For all the images
successfully loaded into PW, the edge detection area was
specified in the head-centered 5D
h
×5D
h
square.
When the edge detection area specified an error, all two
successfully loaded 1,800 × 1,800 resolution images became
blurry (Table 2). In contrast, images with a resolution of
1,076 × 1,076 exhibited a 2/10 blur ratio, which
corresponded to the lowest blur occurrence rate among
all resolutions.
TABLE 2 Polyware compatibility tests of multiple TIFF x-ray image sizes.
Image Resolution 1,024 ×
1,024
1,076 ×
1,076
1,200 ×
1,200
1,300 ×
1,300
1,400 ×
1,400
1,500 ×
1,500
1,800 ×
1,800
2,494 ×
2,494
2,780 ×
2,780
3,020 ×
3,020
Gray bits 8 8 8 8 8 8 8 8 8 8
Size (KB) 1,060 1,220 1,499 1,742 1,994 2,253 3,433 5.978 7,202 26,721
Loading Error ratio
(in the desktop PC)
0/10 0/10 0/10 0/10 0/10 0/10 8/10 10/10 10/10 10/10
Loading Error ratio
(in the laptop PC)
0/10 0/10 0/10 0/10 0/10 0/10 8/10 10/10 10/10 10/10
Blur ratio in the edge
detection image (identical
in both PCs)
5/10 2/10 4/10 6/10 5/10 5/10 2/2 NA NA NA
Wear (mm) True = 6.67
of All cases
6.88 (0.50) 6.79 (0.01) 6.42 (0.42) 6.64 (0.52) 6.70 (0.14) 6.61 (0.29) 6.49 (0.66) NA NA NA
Wear (mm) True = 6.67
of Non-blur cases only
6.60 (0.00) 6.79 (0.00) 6.24 (0.17) 6.27 (0.28) 6.68 (0.01) 6.46 (0.25) NA NA NA NA
Lateral tilt (
)
True = 36.70
of All cases
36.5 (0.8) 36.0 (0.5) 36.5 (0.5) 36.2 (0.4) 36.2 (0.4) 36.4 (0.6) 36.5 (0.9) NA NA NA
Lateral tilt (
)
True = 36.70
of Non-
blur cases only
36.3 (0.7) 36.0 (0.6) 36.6 (0.6) 36.3 (0.3) 36.0 (0.4) 36.3 (0.4) NA NA NA NA
Anteversion (
)
True = −9.0
of All cases
−8.7 (0.4) −8.6 (0.7) −8.3 (0.3) −8.7 (0.8) −8.8 (0.5) −8.7 (0.7) −8.5 (0.1) NA NA NA
Anteversion (
)
True = −9.0
of Non-blur
cases only
−8.5 (0.2) −8.6 (0.8) −8.5 (0.2) −8.9 (0.9) −9.0 (0.5) −8.5 (0.7) NA NA NA NA
NA, not available since none of the measurement trials were successful or possible. The wear, lateral tilt, and anteversion were obtained from the only successful
measurements without any blur phenomenon in both the initial and final images. These tests were performed for the x-ray image whose midpoint coincides with
the center of the femoral head (Oin Figure 5).
Min Lee et al. 10.3389/fsurg.2022.997848
Frontiers in Surgery 06 frontiersin.org
FIGURE 5
Eccentricity comparison test setup, i.e., nine spatial eccentricity modes. With respect to the center of the x-ray detector, nine spatial eccentricity
locations of the THA prostheses were set up to figure out how the eccentricity of the component location affected PolyWare measurement results.
FIGURE 6
The blur of the edge detection area. For the same x-ray image, different specifications of rectangular edge detection areas result in different image
sharpness. The left one is normal, but the right one is blurred. In the normal case, the rectangular edge detection area is specified such that its center
is at the very center of the femoral head. In the blurred case the rectangular edge detection area is specified so that its center is considerably off the
center of the femoral head, causing the edge detection area to blur.
Min Lee et al. 10.3389/fsurg.2022.997848
Frontiers in Surgery 07 frontiersin.org
3.3. PW-compatible geometric
features of the edge detection area
specification
The effects of the edge detection area's geometric feature
were assessed with only the X-ray images with a resolution of
1076×1076, because all the images with this resolution were
successfully loaded into PW and exhibited the least blur in
the edge detection process. The incidence across 10 trials
served as a measure of the blur’s occurrence rate. The blur
indicates that the original image is degraded by the blur
created while specifying the edge detection area, and edge
detection will be processed for the degraded image.
The reliability of measurements was evaluated by the
incidence of blurs or unexpected errors, as shown in Table 3.
When the edge detection area was specified as a square with its
center in the center of the FH on x-ray images, PW
measurements showed more reliability as opposed to when the
center of the area was described being as randomly located off
the center of the FH. The edge detection operation is terminated
by an unexpected error, which indicates that the edge detection
procedure returned an error message without any explanation.
Ten trials of the not-head-centered, non-squared
specification resulted in three unexpected errors and five blurs
at the edge detection procedure. When it comes to blurring,
the 7D
h
×7D
h
square specification showed two blur incidents
in ten trials, whereas the 5D
h
×5D
h
square specification
showed one blur incident in 10 trials. The wear values of both
square specifications (including all the blur and non-blur
situations) corresponded to 6.79 (0.00) mm, which was
extremely close to the true value of 6.67 mm. In comparison,
10 trials with not-head-centered, non-squared specifications
produced three unexpected errors and five blurs during the
edge detection procedure. The wear of the non-head-centered
random non-square specification was 6.92 (0.15) mm, which
was less accurate and precise than the squared specifications.
3.4. Effect of the prosthesis’s eccentric
placement at the time of the x-ray
scanning
The eccentricity tests were performed with only the images
with a resolution of 1,076 × 1,076, and their edge detection area
specification was the head-centered 5D
h
×5D
h
square. The PW
measurements for each eccentricity mode were averaged from
ten trials. Table 4 shows the wear amount and alignment
measurement results for the nine different eccentricity modes.
The spatial eccentricity of the prosthesis from the original
x-ray image center led to inaccurate results in wear
measurement. L15,R15 S15,R15 I15, and L15I15 eccentricities
resulted in an error of approximately 0.42 mm, and the
I15 eccentricity resulted in an error of approximately 0.67 mm.
L15S15 and R15 resulted in an error of 0.50 mm.
AC anteversion measurements were considerably inaccurate
due to any eccentricity in all directions, and, in particular, the
maximum error appearing at L
15I
15
mode by 5.4° (=14.4°–9.0°).
S
15
and I
15
eccentricity modes resulted in anteversion
measurement errors of 4.3° and −3.6°, respectively.
4. Discussion
In the study, we are faced with the very uncomfortable fact
that some of the published PW measuring studies may not be
valid if they did not acknowledge and fix the errors our
research revealed. In light of our findings, we advise
FIGURE 7
Three ways of specifying the edge detection area. The edge detection area was assigned as a rectangle whose edge lengths were 5 times (5D
h
)
square, 7 times (7D
h
) square of the diameter of the femoral head component (D
h
), or non-square. The square areas specified were centered in
the middle of the FH, whereas non-square ones were off the FH center.
Min Lee et al. 10.3389/fsurg.2022.997848
Frontiers in Surgery 08 frontiersin.org
polyethylene wear researchers to use the following three PW
measurement protocols.
4.1. Finding 1: Optimal size for original
x-ray images (Sbest)
Regarding the image loading problem, all images with a
resolution of 1,500 × 1,500 or lower were successfully loaded
into PW without any errors. Particularly, images with a
resolution of 1,076 × 1,076 showed a two-to-ten (2/10) blur
occurring ratio that was the lowest among all image
resolutions. In practical situations, an original image
transferred from a medical modality may be non-square
(1,076 × 1,500 or 1,200 × 1,100, for example). In this instance,
we recommend cropping it into a square with the original
image’s center at the center, changing its pixel size to 1,076 ×
1,076. Therefore, an x-ray image with a resolution of 1,076 ×
1,076 is optimally compatible with the PW measurement, i.e.,
Sbest = 1,076 × 1,076.
4.2. Finding 2: Optimal location of the
THA prosthesis on the original x-ray
images (Ebest)
The eccentricity of the FH location from the x-ray beam
center line significantly reduced the accuracy of the liner wear
and AC anteversion measurements. The errors in Figures 8,9
are mean deviations from the true wear and anteversion
values recalculated from Table 3, respectively. It is clear that
an eccentric placement of the prosthesis with respect to the
x-ray beam center line leads to errors in the liner wear and
AC anteversion. Because the prosthesis was placed superiorly
or inferiorly in relation to the x-ray beam source, the
anteversion specifically showed a greater inaccuracy. Unless
the FH was placed extremely close to the central x-ray
TABLE 3 The measured wear for different area specifications (true wear = 6.67 mm).
Head-centered Not head-centered
5D
h
×5D
h
square 7D
h
×7D
h
square Non-square
Trials Trouble Wear (mm) Trouble Wear (mm) Trouble Wear (mm)
1 No 6.79 No 6.79 No 7.05
2 Blur 7.76 No 6.79 Error NA
3 No 6.79 No 6.79 Blur 7.75
4 No 6.79 No 6.79 No 6.79
5 No 6.79 No 6.79 Blur 6.12
6 No 6.79 Blur 6.12 Error NA
7 No 6.79 Blur 6.12 Blur 7.98
8 No 6.79 No 6.79 Blur 6.12
9 No 6.79 No 6.79 Error NA
10 No 6.79 No 6.79 Blur 6.12
Total Error: 0
Blur: 1
6.89 (0.31) of all 6.79 (0.00)
of 9 N-blurs 7.76 of 1 blur
Error: 0
Blur: 2
6.66 (0.28) of all 6.79 (0.00) of 8
N-blurs 6.12 (0.00) of 2 blurs
Error: 3
Blur: 5
6.85 (0.79) of all 6.92 (0.18) of 2
N-blurs 6.82 (0.96) of 5 Blurs
NA, not available since none of the measurement trials were successful or possible. These tests were performed for the x-ray image whose midpoint coincides with
the center of the femoral head (O in Figure 5). The symbol D
h
denotes the diameter of the femoral head component.
TABLE 4 PolyWare measurement results for nine spatial eccentricity
modes.
Eccentricity
mode
Liner wear,
mm
True = 6.67
Lateral tilt,
True = 36.7
Anteversion,
True = −9.0
O6.79 (0.00) 36.3 (0.3) 9.0 (0.6)
L15 6.25 (0.00) 36.2 (0.6) 11.1 (0.7)
L15S15 6.52 (0.00) 36.2 (0.2) 7.9 (0.6)
S15 6.79 (0.00) 36.7 (0.3) 4.7 (0.5)
R15S15 6.25 (0.00) 37.2 (0.3) 1.5 (0.2)
R15 6.52 (0.00) 37.2 (0.3) 5.9 (0.6)
R15I15 6.25 (0.00) 37.2 (0.3) 9.8 (0.4)
I15 6.00 (0.00) 37.1 (0.5) 12.6 (0.4)
L15I15 6.25 (0.00) 37.2 (0.4) 14.4 (0.3)
L,R,S, and Iin the eccentricity mode represent left, right, superior, and inferior,
respectively. The subscript 15 in the eccentricity mode indicates a translational
distance of 15 mm.
Min Lee et al. 10.3389/fsurg.2022.997848
Frontiers in Surgery 09 frontiersin.org
beamline at the x-ray scanning instant, the anteversion
measurement by PW was unreliable.
To determine why the eccentric prosthesis placement
significantly affected the anteversion, we measured the
anteversion of the virtual x-ray images generated by
simulating a projection of the hip prosthesis 3D CAD
model in a perspective view. The perspective view
simulation was made with 3D CAD software, i.e.,
Rapidform 2006® (INUSTechnology, Seoul, Korea). With a
source-to-detector distance of 394 cm, Rapidform 2006
creates a virtual perspective image in which the proximal
edge of a 100 cm 100 cm 100 cm cube is projected as
130 cm on the detector plane. Figure 10 demonstrates
changes in liner wear based on superior and inferior
eccentricity modes. The anteversion was calculated via the
Lewinnek method (11). The superior and inferior 15 cm
eccentricity modes were 4.3° and 3.6° of over- and under-
anteversion, respectively. From our CAD measurement
using Rapidform, it is postulated that PW uses the
Lewinnek method to calculate acetabular anteversion. It is
concluded that the acetabular measurement is only valid
when the center of the FH (or similarly, the center of the
AC) is placed very close to the center line of the x-ray
beam. As a result, eccentricity significantly impairs the
accuracy of measurements of wear and acetabular
anteversion; thus, the FH center should be positioned along
the center line of the x-ray beam (Ebest =O).
4.3. Finding 3: Optimal specification of
the edge detection area (Gbest)
The pre-process AP image in PW measurements required
cutting out unnecessary portions from the originally loaded
image. The image remaining after the pre-processing was
used for edge detection of the FH and AC. The occurrence
of image blur was influenced by the geometric
characteristics of the region that users had specified for the
pre-processing. The geometric features of the selected area
include size and symmetry with respect to the center of the
FIGURE 8
The error (in absolute values) in the wear of the femoral head’s spatial eccentricity modes in the original x-rays.
Min Lee et al. 10.3389/fsurg.2022.997848
Frontiers in Surgery 10 frontiersin.org
FH. In the current study, the asymmetry of the specified area
increased the likelihood of a blur. Because the magnified
process image can be more accurately analyzed for edge
detection, the head-centered 5D
h
×5D
h
square is preferable
to the head-centered 7D
h
×7D
h
square. In this sense, we
postulate that a head-centered 3D
h
×3D
h
square would also
be preferable. The optimal geometric specification mode of
the image processing for edge detection corresponds to the
head-centered 5D
h
×5D
h
square, i.e., Gbest = head-centered
5D
h
×5D
h
square or probably the head-centered 3D
h
×3D
h
square.
The current research presents several limitations. First,
because the study only used prostheses rather than
including real tissues such as bones and soft tissues, the x-
ray images used here are clinically impractical. There
should be a small occlusion when tissues are absent around
the prostheses; as a result, the outline of the prostheses will
be more visible than when tissues are present. However, the
current study aims to evaluate measurement accuracy. To
assess accuracy, the true wear rate was translated into a
precise simulation, and we compared the measured values
to that true rate. Real clinical patient hip images cannot
provide a true wear value since we are not allowed to
measure the true AC wear of living individuals by surgically
opening them and taking direct measurements.
Additionally, the accuracy was also hindered by the
difficulty of standardizing complex human tissue shapes
and material compositions around THA prostheses during
each x-ray scanning. Hence, in the current study, x-ray
images were obtained without considering human tissues, to
control accurately wear simulation by translating the
femoral component. In future studies, a simulation may be
developed to represent tissues around the prostheses.
Second, the resolution and aspect ratio of the original x-ray
images that were tested did not cover all possible variations.
Clinical x-ray images may have a variety of resolutions or
aspect ratios. Additionally, although PW automatically
squared the imported images, practically obtained original
x-ray images may not be. However, the aspect ratio will be
irrelevant if the x-ray image has a resolution of 1,500 ×
FIGURE 9
Errors (in absolute values) in the acetabular anteversion for the femoral head’s spatial eccentricity modes in the original x-rays.
Min Lee et al. 10.3389/fsurg.2022.997848
Frontiers in Surgery 11 frontiersin.org
1,500 or lower. Thirdly, the current study investigated only
one type of THA prosthesis, i.e., THA using fourth-
generation ceramic-on-polyethylene articulations. When it
comes to opacity in x-ray scanning, fourth-generation
ceramic-on-polyethylene and metal-on-polyethylene are
comparable since their liners are made of polyethylene.
However, if the liners are made of radio-opaque materials
like metal or fourth-generation ceramics, it can be difficult
to identify the outline of the femoral head. It must be
noted that PW compares patient x-ray images to measure
the volume of polyethylene material worn away from the
bearing surfaces of orthopedic hip implants over time
(http://www.draftware.com/html/polyware.htm). Hence, PW
can only be used the measure polyethylene wear.
The authors are aware of no published research that has
investigated the error sources and their solutions in
PW measurements. Recent literature has reported that
manual measurements of the digital x-ray screen and PW
measurement are comparable when it comes to measuring
AC anteversion (9,13). However, it should be highlighted
that since there is no way for them to measure true
polyethylene wear in living THA patients, their study only
reports repeatability and not accuracy. When it comes to
wear, comparing our findings with existing literature is
quite limited.
5. Conclusion
Because PW has been the only polyethylene wear
measurement tool used, identifying its sources of error and
devising a countermeasure is of the utmost importance. For
the best accuracy and reliability in PolyWare™measurements,
this study strongly recommends following the methodology
proposed. Otherwise, the validity of the PW measurements
cannot be reliably determined.
FIGURE 10
Measurement of acetabular anteversion using CAD to investigate the effect of the eccentricity of the prosthesis from the center of the x-ray beam on
the acetabular anteversion. The same x-ray images used for polyethylene measurements were also used for the measurement using CAD software,
i.e., Rapidform 2006
®
(INUSTechnology, Seoul, Korea). The superior and inferior placements of the prosthesis bring about errors in acetabular
anteversion by the nature of perspective x-ray imaging.
Min Lee et al. 10.3389/fsurg.2022.997848
Frontiers in Surgery 12 frontiersin.org
Data availability statement
The original contributions presented in the study are
included in the article/Supplementary Material, further
inquiries can be directed to the corresponding author.
Author contributions
JM carried out the experiments and data analysis, and S-HB
conceived the idea for the study and participated in the
manuscript writing. Additionally, YS conducted experiments,
interpreted their results and wrote the manuscript. All authors
contributed to the article and approved the submitted version.
Conflict of interest
The authors declare that the research was conducted in the
absence of any commercial or financial relationships that could
be construed as a potential conflict of interest.
Publisher’s note
All claims expressed in this article are solely those of the
authors and do not necessarily represent those of their
affiliated organizations, or those of the publisher, the editors
and the reviewers. Any product that may be evaluated in this
article, or claim that may be made by its manufacturer, is not
guaranteed or endorsed by the publisher.
References
1. Bozic KJ, Kurtz SM, Lau E, Ong K, Vail TP, Berry DJ. “The epidemiology of
revision total hip arthroplasty in the United States”.J Bone Joint Surg . (2009) 91
(1):128–33. doi: 10.2106/JBJS.H.00155
2. Shen C, Tang Z-H, Hu J-Z, Zou G-Y, Xiao R-C, Yan D-X. “Does cross-linked
polyethylene decrease the revision rate of total hip arthroplasty compared with
conventional polyethylene? A Meta-Analysis”.Orthop Traumatol: Surg Res.
(2014) 100(7):745–50. doi: 10.1016/j.otsr.2014.07.015
3. Biedermann R, Tonin A, Krismer M, Rachbauer F, Eibl G, Stöckl B.
“Reducing the risk of dislocation after total hip arthroplasty: the effect of
orientation of the acetabular component”.J Bone Joint Surg Br. (2005) 87
(6):762–9. doi: 10.1302/0301-620X.87B6.14745
4. D’lima DD, Urquhart AG, Buehler KO, Walker RH, Colwell Jr CW. “The
effect of the orientation of the acetabular and femoral components on the range
of motion of the hip at different head-neck ratios”.J Bone Joint Surg . (2000)
82(3):315–21. doi: 10.2106/00004623-200003000-00003
5. Dumbleton JH, Manley MT, Edidin AA. “A literature review of the
association between wear rate and osteolysis in total hip arthroplasty”.
J Arthroplasty. (2002) 17(5):649–61. doi: 10.1054/arth.2002.33664
6. Kennedy J, Rogers W, Soffe K, Sullivan R, Griffen D, Sheehan L. “Effect of
acetabular component orientation on recurrent dislocation, pelvic osteolysis,
polyethylene wear, and component migration”.J Arthroplasty. (1998) 13
(5):530–4. doi: 10.1016/S0883-5403(98)90052-3
7. Kerrigan CS, Mckenna SJ, Ricketts IW, Wigderowitz C. “Automated
assessment of polyethylene wear in cemented acetabular components using
anteroposterior radiographs of total hip replacements”.Comput Med Imaging
Graph. (2008) 32(3):221–38. doi: 10.1016/j.compmedimag.2007.12.002
8. Wan Z, Boutary M, Dorr LD. “The influence of acetabular component
position on wear in total hip arthroplasty”.J Arthroplasty. (2008) 23(1):51–6.
doi: 10.1016/j.arth.2007.06.008
9. Shin W, Lee S, Lee K, Cho H, Lee J, Suh K. “The reliability and accuracy of
measuring anteversion of the acetabular component on plain anteroposterior and
lateral radiographs after total hip arthroplasty”.Bone Joint J. (2015) 97(5):611–6.
doi: 10.1302/0301-620X.97B5.34735
10. Zheng G. “Effective incorporating spatial information in a mutual information
based 3D–2D registration of a CT volume to x-ray images”.Comput Med Imaging
Graph. (2010) 34(7):553–62. doi: 10.1016/j.compmedimag.2010.03.004
11. Lewinnek GE, Lewis J, Tarr R, Compere C, Zimmerman J. “Dislocations
after total hip-replacement arthroplasties”.J Bone Joint Surg Am. (1978) 60
(2):217–20. doi: 10.2106/00004623-197860020-00014
12. Kerr DA. “The proper pivot point for panoramic photography”.Pumpkin.
(2008) 2(8):1–15.
13. ParkYS,ShinWC,LeeSM,KwakSH,BaeJY,SuhKT.“The best method for
evaluating anteversion of the acetabular component after total hip arthroplasty on
plain radiographs”.JOrthopSurgRes. (2018) 13(1):66. doi: 10.1186/s13018-018-0767-4
Min Lee et al. 10.3389/fsurg.2022.997848
Frontiers in Surgery 13 frontiersin.org
Available via license: CC BY
Content may be subject to copyright.
