591091

research-article2015

FAIXXX10.1177/1071100715591091Foot & Ankle InternationalBarg et al

Article

Influence of Ankle Position and Radiographic Projection Angle on Measurement of Supramalleolar Alignment on the Anteroposterior and Hindfoot Alignment Views

Foot & Ankle International® 1­–10 © The Author(s) 2015 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/1071100715591091 fai.sagepub.com

Alexej Barg, MD1,2, Richard L. Amendola1,2, Heath B. Henninger, PhD1,2, Ashley L. Kapron, PhD1, Charles L. Saltzman, MD1, and Andrew E. Anderson, PhD1,2,3,4,5

Abstract Background: Using digitally reconstructed radiographs (DRRs), we determined how changes in the x-ray beam projection angle from the horizon, tibiotalar joint angle, and axial rotation of the foot influenced measurements of the medial distal tibial angle (MDTA) on the anteroposterior (AP) and hindfoot alignment views (HAV). Methods: Seven cadaver foot-ankle specimens were scanned by computed tomography (CT) at fixed tibiotalar joint positions, ranging from 15 degrees of dorsiflexion to 25 degrees of plantarflexion. DRRs were created from each CT scan to simulate alterations in the horizontal projection angle (0 to 25 degrees) and foot axial rotation (–30 to 30 degrees). The MDTA was measured on each DRR and compared with that quantified on the baseline HAV and AP view. Results: Altering the horizontal projection angle by ≥5 degrees and >10 degrees significantly altered the MDTA for the AP view and the HAV, respectively. Shifting dorsiflexion and plantarflexion caused minor changes in the MDTA that were only statistically significant for the HAV. Axial rotation significantly changed the MDTA on both views, but deviations were more pronounced for the HAV. Conclusions: Compared with the HAV, the MDTA on the AP view was less sensitive to changes in foot-ankle position. However, increasing the tilt of the x-ray beam from the horizon altered the MDTA on the AP view substantially. Clinical Significance: To avoid misinterpretation of the MDTA, we recommend using the AP view to quantify supramalleolar alignment as it is less sensitive to changes in positioning of the foot-ankle. When acquiring an AP film, the x-ray beam should be directed along the horizon to ensure consistent assessment of the MDTA across patients. Keywords: ankle, alignment, medial distal tibia angle, radiographs, digitally reconstructed radiographs Osteoarthritis (OA) is a common problem in the ankle joint.46,52,56 Concomitant varus or valgus deformities frequently accompany cartilage damage in osteoarthritic ankles.19,20 Surgical strategies to treat OA address both cartilage damage and possible malalignment. Treatment can be divided into joint-preserving (eg, supramalleolar realignment osteotomy5,17,28,37,48) and nonpreserving (eg, ankle arthrodesis,12,14,18,40,41 total ankle replacement15,23,24,31,42,45) procedures. Preoperative planning is critical to the success of ankle surgery. Weight-bearing radiographs, in particular, are essential to evaluate concomitant foot and ankle deformities.11,32,50,51 For this purpose, the medial distal tibial angle (MDTA) serves to quantify supramalleolar ankle alignment in the coronal plane.4,49 The MDTA can be measured on the anteroposterior (AP) view as well as the hindfoot alignment view (HAV).4,27,49 In the AP view, the MDTA has been measured as 92.4 ± 3.1 degrees (range, 88.0-100.0

degrees) in a cohort of 93 asymptomatic control subjects27 and 93.3 ± 3.2 degrees (range, 88.0-100.0 degrees) in a cadaver study.22 1

Department of Orthopaedics, University of Utah, Salt Lake City, UT, USA 2 Harold K. Dunn Orthopaedic Research Laboratory, University of Utah, Salt Lake City, UT, USA 3 Department of Bioengineering, University of Utah, Salt Lake City, UT, USA 4 Department of Physical Therapy, University of Utah, Salt Lake City, UT, USA 5 Scientific Computing and Imaging Institute, University of Utah, Salt Lake City, UT, USA Corresponding Author: Andrew E. Anderson, PhD, Department of Orthopaedics, Harold K. Dunn Orthopaedic Research Laboratory, University of Utah, 590 Wakara Way, Salt Lake City, UT 84108, USA. Email: [email protected]

Downloaded from fai.sagepub.com at University of Otago Library on July 20, 2015

2

Foot & Ankle International 

The x-ray beam projection angle is inherently different between the HAV and AP view, which likely explains why the MDTA differs between the two views (up to 12 degrees as reported by Barg et al4). However, within a given view, there may be subtle changes in the manner in which the x-ray beam is aligned across patients. There may also be variation in how the patient’s foot and ankle are positioned during the acquisition of the radiograph. Since these measures guide the magnitude of realignment osteotomies, it is critical to understand how these spatial factors may influence the clinical interpretation of supramalleolar alignment. Ideally, the MDTA on the AP view and the HAV could be measured following known, incremental changes in the projection angle and degrees of ankle dorsiflexion and plantarflexion as well as axial rotation of the foot. In practice, it may be difficult to implement an experiment, with cadavers or live subjects, capable of achieving highly controlled, repeatable orientations of the x-ray equipment and footankle. To fully understand how incremental changes in spatial alignment and projection influence the MDTA could require hundreds of radiographs, which is not practical for live subjects. Alternatively, using digitally reconstructed radiographs (DRRs) generated from computed tomography (CT) data sets, one can create an equivalent radiograph for any projection by applying computer-coordinated and exact movements of the object of interest.7,36 In the context of scientific research, the primary advantage of using DRRs is that they are generated from exact, user-defined perspectives. In theory, an analyst can generate an infinite number of simulated radiographs from a single CT scan, which could reduce the burden to x-ray equipment and the time required to generate a large number of films. As DRRs require a CT scan, they are not a practical alternative to conventional radiographs for clinical care. However, DRRs have served as a valuable scientific tool to study the hip,16,21,53 knee,3,47,54,55 and ankle.29 The suitability of DRRs as surrogates for standard radiographs was demonstrated in two studies.16,21 To the best of our knowledge, DRRs have not been used to elucidate how measurements of the MDTA may be influenced by projection angle and patient positioning in the AP view and the HAV. Furthermore, the use of DRRs as a surrogate to conventional radiographs for measurement of the MDTA has not been demonstrated. We undertook the present study to better understand whether measurements of the MDTA are influenced by radiographic technique. First, we set out to validate the use of DRRs as a surrogate for standard radiographs by comparing measurements of the MDTA between DRRs and conventional radiographs. Second, we wanted to understand whether MDTA measurements of the AP view and the HAV were influenced by the x-ray projection angle from the horizon and degree of foot-ankle axial rotation as well as dorsiflexion and plantarflexion of the tibiotalar joint. We hypothesized that the projection angle and position of the

foot-ankle specimen would alter the MDTA relative to the baseline values in both the AP view and the HAV.

Materials and Methods Specimens and Multidetector CT Imaging To avoid exposure to ionizing radiation to patients or research subjects, we chose to execute this research as a combined in vitro and computational simulation study. Seven fresh-frozen, cadaveric, male foot-ankle specimens were acquired with a mean age of 65.0 ± 16.5 years (range, 52-89 years). Specimens were stored at -20°C and thawed for 24 hours prior to use. On radiographic examination, as assessed by an orthopedic surgeon (A.B.), all ankles were absent of gross degenerative changes and joint space narrowing that would be indicative of OA. A custom radiolucent acrylic fixation device was built to hold the foot and ankle in a static position at known dorsiflexion and plantarflexion angles (Figure 1). All CT scans were performed without external loads applied to the cadaver. Axial CT images were acquired with a Siemens Sensation (Siemens Medical, Malvern, PA) 16 CT scanner. The CT settings were the same for all specimens at 100 kV, 35 mAS, 512 × 512 acquisition matrix, 0.6-mm slice thickness, 0.5-mm increments, and 0.6 pitch. Scans were performed on all specimens in neutral position and at increments ranging from 15 degrees of dorsiflexion through 25 degrees of plantarflexion. CT scans were stored in DICOM (Digital Imaging and Communications in Medicine) format for later processing.

Digitally Reconstructed Radiographs DRRs of the foot-ankle specimens were created from CT data using Amira (version 5.3, Visage Imaging, San Diego, CA). High-resolution DRR alpha scale transparency and gamma values were set to 0.0194805 and 1.0, respectively, in Amira. AP views for the DRRs were generated from an orthographic coronal view. Hindfoot alignment view DRRs were created by axially rotating the CT images 180 degrees to a coronal posteroanterior view. For each ankle, a focal landmark in which to rotate the tibiotalar joint center (Figure 2A) was required. The focal landmark was manually determined (Figure 2B) by two independent observers (A.B., R.L.A.) trained in image processing. CT image stacks were then rotated axially about the Cartesian coordinates of the focal landmark at 10-degree increments from 30 degrees of external rotation to 30 degrees of internal rotation (to assess the influence of axial rotation) and sagittally from 0 degrees to 25 degrees at 5-degree increments (to assess the influence of projection angle from the horizon). DRRs were extracted after each rotation or combination of rotations. Forty-two DRRs were generated to study each view (AP or HAV) of each specimen (6 for dorsiflexion and plantarflexion, 7 for axial rotation).

Downloaded from fai.sagepub.com at University of Otago Library on July 20, 2015

3

Barg et al

Figure 1.  Sagittal (left) and anterior (right) views of custom acrylic device that allowed for fixed, user-defined dorsiflexion and plantar flexion angles. The device was calibrated using a digital goniometer, with tick marks added to indicate 5-degree increments.

Medial Distal Tibial Angle Measurement The MDTA was measured using a semiautomatic, computer-assisted approach as described previously.4 Briefly, a circle was automatically drawn and positioned to fit between the medial and lateral cortex at the most proximal location of the tibial shaft. The center of this circle was the most proximal point of the longitudinal axis of the tibia. Next, another circle was drawn over the distal tibia and positioned to fit between three cortices: the medial, lateral, and tibial plafond. The center of this circle was placed on the longitudinal axis of the tibia, where it passed through the center of the talus. The joint orientation line was drawn across the flat subchondral line of the tibial plafond.38 The MDTA was then calculated automatically as the angle between the longitudinal axis of the tibia and the joint orientation line.4 All measurements of two specimens were performed by two observers (A.B., R.L.A.) with different levels of training to assess interobserver reliability. The MDTA was measured again six weeks after the initial reading to assess intraobserver reliability. In two different specimens, a radiological control study was performed to assess the ability of DRRs to serve as a surrogate for conventional radiographs. Specifically, the AP view and the HAV were acquired at 10-degree increments from 30 degrees of external rotation to 30 degrees of internal rotation with the foot-ankle in neutral dorsiflexion (Figure 3), and the MDTA was measured as described above. The film focus distance was held constant at 40 inches for the AP view. The beam (55 kV, 6 mAS) was

centered on the tibiotalar joint. For the HAV, the film focus distance was constant at 40 inches. The beam (62 kV, 6 mAS) was centered on the ankle with the beam angle of 20 degrees to the horizontal, with a field of view that included the midshaft of the tibia to just below the calcaneus.44

Statistical Analysis Bland-Altman plots were generated to quantify agreement between the DRR and conventional radiographic methods to measure the MDTA.2,8,9 The Bland Altman plots included the bias (mean of the differences in MDTA between DRR and conventional films) as well as the upper and lower 95% limits of agreement (estimated as 1.96 times the standard deviation of the differences). The intraclass correlation coefficients (ICCs) and associated 95% confidence intervals (CIs) of the ICCs quantified inter- and intraobserver repeatability. ICC values were interpreted as follows: ICC = 1, perfect agreement; 0.81 to 0.99, excellent agreement; and 0.61 to 0.80, substantial agreement.13 The Shapiro-Wilk test was performed to verify whether DRR and conventional film measurements of the MDTA were normally distributed. With normality confirmed, MDTA measurements of the DRRs from the first set of observations of the primary assessor were compared using paired t tests. Here, paired t test, with significance set at P < .05, compared the MDTA measured on the neutral AP view DRR (0-degree x-ray beam angle from the horizon, 0 degrees of axial rotation, 0 degrees of tibiotalar joint dorsiflexion) separately with the MDTA measured on the AP

Downloaded from fai.sagepub.com at University of Otago Library on July 20, 2015

4

Foot & Ankle International 

Figure 2.  Method used to verify placement of focal landmark. The focal landmark (yellow dot) was created manually by estimating the midpoint of the joint about the 3 computed tomography (CT) planes (top: shown for a single specimen). Soft tissue structures were then visualized as semitransparent media relative to the osseous anatomy to help the observer determine the final position of the focal point (bottom: shown for a single specimen).

view DRR after it was altered by an incremental change in x-ray beam projection angle, tibiotalar joint position, and foot-ankle axial rotation. The same approach was used to compare the MDTA measured on the baseline neutral HAV DRR (20-degree x-ray beam angle from the horizon, 0 degrees of axial rotation, 0 degrees of tibiotalar joint dorsiflexion). With such a statistical approach, multiple paired t tests were required. When multiplicity is present, the usual approach is to adjust P values to control for type I error. However, in this study, we sought to compare one endpoint, a single incremental change in projection angle/joint position, to the baseline MDTA of the AP view and the HAV to better understand what level of change induces statistically significant alterations to the MDTA. If P values were adjusted, it could lead to a loss of significance for a single change in position or projection angle, which could be otherwise clinically important to highlight. Therefore, despite

having multiple t tests present, the P values were not adjusted.43 All data were analyzed using IBM SPSS Statistics version 22.0 (IBM, Armonk, NY) and SigmaPlot version 12.0 (Systat Software Inc, San Jose, CA).

Results For the AP view, the Bland-Altman plot indicated strong agreement between the MDTA measured on the DRR to that from the conventional radiograph, with no obvious directional preference (Figure 4, left). Specifically, the MDTA could be measured to a bias of 0.16 degrees. The limits of agreement for the AP view indicated that DRR measurements would be measured to within approximately ±1.5 degrees of that from the conventional film in 95% of the cases. Excellent agreement was also observed between the MDTA measured on the conventional HAV to that on

Downloaded from fai.sagepub.com at University of Otago Library on July 20, 2015

5

Barg et al

Figure 3.  Photographs and radiographs demonstrating the various views acquired for the digitally reconstructed radiograph (DRR) validation study. An anteroposterior (AP) view was acquired at neutral position and 30 degrees of external rotation as indicated in the approximate location of the cadaver specimen shown (top). Conventional films and corresponding DRRs of the specimen in neutral rotation are shown for the AP view (lower left) and hindfoot alignment view (HAV) (lower right).

the corresponding DRR (Figure 4, right). The DRR was found to slightly underestimate the MDTA, with a bias of -1.20 degrees. The limits of agreement for the HAV indicated that DRR measurements would be made to within approximately ±2.0 degrees of that from the conventional film in 95% of the cases. The ICCs representing interobserver and intraobserver repeatability of the MDTA measurements using DRRs were excellent at 0.912 (95% CI, 0.881-0.952) and 0.882 (95% CI, 0.792-0.916), respectively. Overall, the MDTA decreased when the x-ray projection angle from the horizon was increased (Figure 5). All changes in the tilt of the x-ray beam significantly altered

MDTA measurements for the AP view (Table 1). However, for the HAV, tilt of the x-ray beam only imparted significant changes when the disparity was greater than 10 degrees from the baseline value of 20 degrees. There was a clear increase of the MDTA on the HAV when the foot and ankle went from 30 degrees of external rotation to 30 degrees of internal rotation; only a gradual increase was noted for the AP view (Figure 6). More specifically, for the HAV, significant changes in the MDTA were observed for all axial rotations analyzed (Table 2). Conversely, for the AP view, axial foot-ankle rotation only significantly altered MDTA measurements at 20 and 30 degrees of external rotation.

Downloaded from fai.sagepub.com at University of Otago Library on July 20, 2015

6

Foot & Ankle International 

Figure 4.  Bland-Altman plots showing level of agreement between medial distal tibial angle (MDTA) measurements acquired from digitally reconstructed radiographs (DRRs) to those from conventional films (solid line represents the bias; dashed lines represent the upper and lower bounds of the 95% limits of agreement). Left: For the anteroposterior (AP) view, the DRR overestimated the MDTA measured on the conventional film by only 0.162 degrees (solid line). The limits of agreement for the AP view indicated that DRR measurements would be measured to within approximately ±1.5 degrees of that from the conventional film in 95% of the cases. Right: For the hindfoot alignment view (HAV), the DRR underestimated the MDTA measured on the conventional film by 1.202 degrees. The limits of agreement for the HAV indicated that DRR measurements would be measured to within approximately ±2.0 degrees of that from the conventional film in 95% of the cases.

Overall, the MDTA appeared to decrease slightly when the tibiotalar joint progressed from dorsiflexion to plantarflexion for both the HAV and the AP view. However, significant changes due to position of the tibiotalar joint were only observed at 25 degrees of plantarflexion for the HAV; at 20 degrees of plantarflexion, the change in MDTA approached statistical significance (Table 3). Significant changes in the MDTA were not found for the AP view as the tibiotalar joint progressed from dorsiflexion to plantarflexion (Table 3).

Discussion In this study, we first demonstrated that DRRs could be used as a surrogate to conventional films for the measurement of the MDTA. Next, using DRRs, we found that the MDTA as measured on the AP view did not change substantially due to axial rotation or tibiotalar joint position. However, measurements of the MDTA on the AP view were clearly sensitive to the tilt of the x-ray beam from the horizon. For the HAV, measurements of the MDTA changed substantially as the foot-ankle experienced axial rotation. Conversely, only minor changes in the HAV MDTA were observed as the x-ray beam was altered from the horizon.

Also, significant changes in the MDTA for the HAV only occurred at 25 degrees of plantarflexion. Projections at computer-controlled angles through volumetric CT data have been used to generate DRRs of the hip,16,21,53 knee,3,47,54,55 and ankle.29 Use of DRRs can be advantageous as they reduce the time required to achieve a large number of equivalent radiographs and also limit equipment use and personnel costs (burden to x-ray equipment, need for radiology technicians, etc). Most important, DRRs are constructed from computer-controlled projections about a well-defined, constant center. In this study, it would have been difficult to ensure that exact projection angles were obtained using standard x-ray equipment. Therefore, use of DRRs in this study effectively eliminated bias that would otherwise be present if an experimental approach were used. We found that DRRs served as a strong surrogate to conventional radiographs for measurement of the MDTA, with mean differences on the order of 1 to 2 degrees. Although minor, two sources could be responsible for the observed difference between the two methods. First, although we strived to obtain a radiographic projection that was nearly identical to a corresponding projection, it is possible that experimental bias was introduced. For example, the goniometer used to align the cadaver in axial rotation is

Downloaded from fai.sagepub.com at University of Otago Library on July 20, 2015

7

Barg et al

Figure 5.  Mean medial distal tibial angle (MDTA) as an influence of the projection angle from the horizon during neutral rotation. As the projection angle was increased, there were significant decreases in the MDTA (Table 1). Graphs denote the anteroposterior (AP) view and hindfoot alignment view (HAV) at 0 degrees and 20 degrees, respectively. Bars represent standard deviations.

Figure 6.  Mean medial distal tibial angle (MDTA) as an influence of foot-ankle rotation about the same x-ray projection angle for the anteroposterior (AP) view and hindfoot alignment view (HAV). There were significant increases in MDTA for the HAV over the entire range of axial rotation analyzed. Changes in the AP view were only found for 20 and 30 degrees of internal rotation (Table 2). Negative angles indicate external rotation. Bars represent standard deviations.

Table 1.  P Values to Indicate Significant Findings When the Baseline Anteroposterior (AP) View and Hindfoot Alignment View (HAV) Were Compared With Incremental Changes in X-ray Beam Tilt From the Horizona.

Table 2.  P Values to Indicate Significant Findings When the Baseline Anteroposterior (AP) View and Hindfoot Alignment View (HAV) Were Compared With Incremental Changes in Axial Rotation of the Foot-Anklea.

View

View

AP (0 degrees)

HAV (20 degrees)

X-ray Beam Tilt vs 5 degrees vs 10 degrees vs 15 degrees vs 20 degrees (HAV) vs 25 degrees vs 0 degrees (AP) vs 5 degrees vs 10 degrees vs 15 degrees vs 25 degrees

P Value .007 .007 .021 .016 .005 .016 .027 .083 .286 .084

a

Significant values indicated in boldface.

not as accurate as the computer-controlled rotation used to create the corresponding DRR. The small discrepancy between DRRs and conventional films could also be the result of MDTA measurement errors, which would be within the narrow 95% confidence interval of the ICCs. Nevertheless, because the discrepancies were very minor, we can conclude that DRRs serve as an excellent surrogate to conventional films for measurement of the MDTA in both the AP view and HAV. Previous studies demonstrated that the measurement of the MDTA depended on the radiographic technique.4,49

Axial Rotation

AP (0 degrees)

vs 30 degrees external rotation vs 20 degrees external rotation vs 10 degrees external rotation vs 10 degrees internal rotation vs 20 degrees internal rotation vs 30 degrees internal rotation HAV (20 degrees) vs 30 degrees external rotation vs 20 degrees external rotation vs 10 degrees external rotation vs 10 degrees internal rotation vs 20 degrees internal rotation vs 30 degrees internal rotation

P Value .023 .134 .912 .530 .604 .081

Influence of Ankle Position and Radiographic Projection Angle on Measurement of Supramalleolar Alignment on the Anteroposterior and Hindfoot Alignment Views.

Using digitally reconstructed radiographs (DRRs), we determined how changes in the x-ray beam projection angle from the horizon, tibiotalar joint angl...
688KB Sizes 1 Downloads 10 Views