Validation of a Novel Software Measurement Tool for Total Hip Arthroplasty

Background Preoperative planning and postoperative evaluation of component position in total hip arthroplasty (THA) utilize specialized software that must be able to provide measurements that are both accurate and precise. A new software program for use in THA has recently been developed. We sought to evaluate the accuracy of this new software in comparison with two current, widely used software programs. Methodology Postoperative anteroposterior (AP) pelvic radiographs from 135 THA patients were retrospectively reviewed. Reference values for acetabular anteversion, inclination, and leg length were established using validated software programs (TraumaCad® as the primary reference value [PRV] and OsiriX LiteTM as the secondary reference value [SRV]). Measurements from the new software program (Intellijoint VIEWTM) were compared with reference values using Student’s t-test and chi-square test. Results For anteversion, mean values for the PRV (27.34° ± 7.27°) and the new software (27.29° ± 7.21°) were not significantly different (p = 0.49). The new software differed from the PRV by a mean of 0.05° ± 0.93°. Similar results were noted for inclination, where the new software differed from the PRV and SRV by -0.13° ± 0.65° and 0.25° ± 1.26°, respectively (mean values: PRV: 43.62° ± 6.02°; SRV: 43.99° ± 6.27°; new software: 43.74° ± 6.17°; p = 0.87), and for leg length, where the new software differed from the PRV and SRV by 0.05 mm ± 0.46 mm and 0.22 mm ± 0.52 mm, respectively (mean values: PRV: 10.61 mm ± 11.60 mm; SRV: 10.77 mm ± 11.70 mm; new software: 10.56 mm ± 11.61 mm; p = 0.98). Measurements were highly correlated across multiple reviewers (intraclass correlation coefficient ≥0.987). Conclusions The new software measurement tool is accurate and precise for assessing the acetabular component position and leg length measurements following THA in AP pelvic radiographs compared to currently used image measurement software.


Introduction
The accurate radiographic assessment of component position and orientation in total hip arthroplasty (THA) relies in large part on the ability of analytic software to correctly measure these parameters. Accuracy in measurement is critical in both the preoperative planning and postoperative verification of component positioning, as positioning errors are known to contribute to poor outcomes [1], poor patient satisfaction [2,3], and can lead to instability, loosening, and dislocation [4,5]. The use of radiographs as the primary imaging modality in THA is accompanied by an acceptance of the inherent error associated with radiographs due to patient positioning errors, artifacts, or distortion [6][7][8]. As such, ensuring that measurement software is accurate, i.e., does not itself introduce additional sources of error, is a critical step in minimizing potential error when evaluating component parameters in THA.
Digital templating and measurement software for THA is widely available and various programs have been validated for accuracy in the measurement of acetabular component position [9,10] and leg length discrepancy [11,12]. These existing products provide important information for both the planning and 1 1 2 3 4 1 5 evaluation phases of procedures; however, as they focus on measurements from a single radiograph view, the data they provide is somewhat limited. With the recent advancements in implant technology and an increased focus on the biomechanical relationship between the hip and spine in THA [13][14][15][16], software that expands the data available and considers these factors would be a valuable asset to THA planning and evaluation. Likewise, advancements in intraoperative technology such as computer-assisted navigation, which can measure component position relative to a number of anatomic planes, have highlighted the limitations of traditional radiographic measurement software, which measures component position in only one plane [17][18][19].
Regardless of advanced features, any new software designed to assist with THA planning and evaluation must first demonstrate the accuracy of measurement when compared with existing, validated options. The present study sought to validate the accuracy of such a new measurement tool for acetabular component and leg length evaluation on standing anteroposterior (AP) radiographs. Measurements from the novel software were compared to two widely used and clinically accepted software platforms.

Study design and patient eligibility
This observational study was a retrospective review of postoperative, standing AP pelvic radiographs of patients who underwent primary THA between February 2016 and November 2017. Only radiographs where the implanted acetabular component and all anatomical landmarks required for leg length measurement (e.g., lesser and greater trochanter, ischial tuberosities) were clearly visible were included in the analysis. Acetabular component position (anteversion and inclination) and leg length were measured using a new software tool and compared to reference values established from two currently available and regularly used software programs for evaluation of pre and postoperative radiographs during primary and revision THA. Informed consent and ethics approval (South County Hospital Institutional Review Board) were obtained prior to data collection.

Data collection and outcome variables
Radiographs were analyzed in each digital software program by multiple trained, independent observers. For all radiographs, measurements were made in triplicate and the results were averaged. Reference values were established for each radiograph by measuring acetabular component orientation and leg length discrepancy using the two existing software programs. Measurements derived from the new software were then compared with these reference values.
For each image and software program, manual calibration was performed using a standard 25 mm scaling ball or, in cases where no marker was used, the known diameter of the femoral head implant. Acetabular component position and leg length were measured using the inter-ischial line method [11,20]. For all software programs, leg length measurements were defined as the perpendicular distance between the interischial line and the most prominent medial point of the lesser trochanter. In cases where the lesser trochanter was not visible, the most lateral point of the greater trochanter was used. Acetabular inclination was defined as the angle between the horizontal reference line and a line bisecting the center of the acetabular cup face through the medial and lateral apexes (i.e., along the major axis) of an oval overlaying the cup face. This oval was used to calculate the acetabular anteversion according to the method discussed in Lewinnek et al. [21], where version is calculated using the formula: V = arcsin(b/a), where V is acetabular component version, b is the radius of the minor ellipse, and a is the radius of the major ellipse.

Measurement software
Intellijoint VIEW TM (Intellijoint Surgical, Kitchener, CA) is a new web-based, digital measurement software tool that has capabilities for functional acetabular component and leg length measurement. It features simultaneous viewing of three radiographic views: standing AP, standing lateral, and seated lateral ( Figure  1), allowing for in-depth analysis and simultaneous comparison of the relative positioning of implants during sitting and standing postures. Acetabular components placed in each view during planning move simultaneously as adjustments are made, reflecting the relative position of the cup component in each of the standing and seated views. The software functionality mirrors that of standard programs, where an anatomical reference line is identified and measurements are made relative to that reference ( Figure 2). For the purposes of this validation study, only standard AP pelvic radiographic views were evaluated. Sample AP pelvic radiograph with acetabular component position and leg length measurements, as measured using Intellijoint VIEW.

AP: anteroposterior
TraumaCad® (BrainLab, Chicago, IL, USA) is a validated, commonly used, multifunctional imaging software used for preoperative planning, templating, and postoperative evaluation of radiographs following total joint arthroplasty and other common orthopedic procedures. Measurements from TraumaCad constituted the primary reference value (PRV) for this study. When used during THA, a horizontal reference line is established by the user and the change in leg length and acetabular inclination are measured relative to this line ( Figure 3). The acetabular version is measured using the method outlined in Lewinnek et al. [21]. Westacott et al. have validated the use of TraumaCad for acetabular cup measurement [22].

FIGURE 3: Primary reference value.
The primary reference value was obtained from radiographic measurement using TraumaCad.
OsiriX TM (Pixmeo SARL, Geneva, Switzerland) is an image display program developed for viewing multidimensional data. Measurements from OsiriX served as the secondary reference value (SRV) for this study. In THA, leg length and acetabular inclination are measured relative to a horizontal reference line demarcated by the user (Figure 4). OsiriX does not measure acetabular anteversion. Marques et al. have previously validated its use for angle and distance measurements [23].

FIGURE 4: Secondary reference value.
The secondary reference value was obtained from radiographic measurement using Osirix.

Statistical analysis
Data are presented as mean (SD) or mean ± SD. Alpha was set a priori at 0.05. The mean values in each software were compared using a paired t-test and one-way analysis of variances. Inter-rater reliability between programs was assessed with Pearson's correlations. Chi-square or Fisher's exact test was performed to evaluate the proportion of new software measurements within 1° and 5° of the absolute mean difference (ABS) between program measurements of acetabular component position and within 1 mm and 5 mm of the ABS between program measurements of leg length. Intra-rater reliability among triplicate measurements was assessed with intraclass correlation coefficients (ICCs).

Study cohort
A total of 135 radiographs were included in the analysis. The patient population was 50% (68/135) female with a mean age at the time of surgery of 64 years (SD: 9.98, range: 29−91). A right THA was represented on 65 (48%) images, 62 (46%) images represented a left THA, and eight (6%) images represented bilateral THA. A total of 143 hips were measured for the acetabular component position. A total of 131 radiographs were available for measurement of leg length, producing 262 measurements of leg length for analysis.
Overall, 100% of measurements from the new software were within 5° of the PRV, and 94.4% were within 1°o f the PRV.  The PRV for inclination was 43.62° (SD: 6.02°) versus 43.74° (SD: 6.17°) as measured by the new software (p = 0.87, Table 1). The SRV was measured at 43.99° (SD: 6.27°). The mean differences in inclination between the new software and the PRV and SRV measurements were -0.13° ± 0.65° (ABS: 0.49° ± 0.44°) and 0.25° ± 1.26° (ABS: 0.79° ± 1.01°), respectively. Measurements with the new software were highly correlated with both the PRV (r = 0.99) and the SRV (r = 0.98). The chi-square test showed no significant difference in the proportion of inclination measurements within 1° (p = 0.88) or 5° (p = 1.00) of the ABS, respectively. When compared to reference values, 95.8% and 100% of measurements were within 1° and 5° of the PRV, respectively. Overall, 99.3% of measurements were within 5° of the SRV, with 90.2% within 1° of the SRV.

Leg Length
The PRV for leg length was 10.61 mm (SD: 11.60 mm) versus 10.56 mm (SD: 11.61 mm) as measured by the new software. The SRV was 10.77 mm (SD: 11.70 mm; p = 0.98, Table 1). The mean differences in leg length between the new software and the PRV or SRV measurements were 0.05 ± 0.46 mm (ABS: 0.29 ± 0.36 mm) and 0.22 ± 0.52 mm (ABS: 0.36 ± 0.43 mm), respectively. Measurements with the new software were highly correlated with the PRV (r = 0.99) and the SRV (r = 0.99). The chi-square test showed no significant difference in the proportion of leg length measurements within 1 mm of the ABS (p = 1.00). Overall, 100% of leg length measurements were within 5 mm of both reference values, with 97.7% within 1 mm of both the PRV and SRV.

Precision
ICCs were calculated for each set of triplicate measurements of acetabular component position and leg length ( Table 2). All correlations were shown to be excellent at ≥0.987.

Discussion
Radiographs remain the standard tool for preoperative and postoperative imaging in THA due to their simple and cost-effective nature. Analytic software using radiographs for preoperative planning and postoperative evaluation must be accurate and repeatable to minimize potential sources of measurement error. The current study demonstrated the capacity of a new software measurement tool to accurately and reliably measure acetabular component anteversion, inclination, and leg length compared to two widely used software measurement programs.
The new software evaluated in our study provides the ability to simultaneously view AP, lateral, standing, and seated images and to observe changes in component position in real-time on all views. Thus, it may be a valuable tool for surgical planning in complex cases such as those involving imbalances in spinopelvic movement. However, the goal of this initial study was to validate the fundamental accuracy and precision of a new software measurement tool in assessing acetabular component position (anteversion and inclination) and leg length. In order to be of value in clinical or research efforts, new software must be accurate in comparison with existing offerings and must demonstrate repeatability across multiple users. We were able to demonstrate both of these requirements. In our study, angles of acetabular component position and measurement of leg length using the new software tool were as accurate as of the two widely accepted reference software platforms. When compared relative to a clinically relevant threshold of ±5° or ±5 mm [24], proportional analyses revealed that for acetabular component orientation and leg length, 100% of the new software's measurements fell within this threshold, with the only exception being inclination relative to the SRV, which saw 99.3% of inclination measurements fall within 5° of the reference value. This study is not without limitations. The software program serving as the SRV does not include an ellipse tool to allow measurement of anteversion, thus limiting that parameter to only one comparator. That program, however, is widely used and trusted for radiographic analysis, and the fact that our study showed excellent correlation with the PRV measured from TraumaCad, considered among the gold standard in the field, is an important finding. The use of nonsurgical personnel to evaluate radiographs may be viewed as a limitation; however, for the purposes of this validation study, we sought to focus solely on the basic functionality of the software using multiple, clinician-trained reviewers to limit the potential impact of bias associated with single reviewers [25]. Future studies examining patient outcomes will incorporate surgical personnel in the analysis. Finally, although the new software has the capability to simultaneously view perpendicular views, that feature was not evaluated in this study. We sought only to validate the fundamental measurement ability of the software, and not to examine the impact of measurement on patient outcomes or the value of this software in complex cases where spinopelvic imbalance is present. Future studies will include clinical follow-up data and a comparison of postoperative outcomes to further evaluate the impact of measurement accuracy on patient outcomes in complex, challenging cases.

Conclusions
The current study validated a new preoperative planning and postoperative assessment software as an accurate and precise tool for acetabular component position and leg length measurements in THA. Anteversion, inclination, and leg length measurements from the new software were not significantly different from both the PRV and the SRV. These findings emphasize the new software's value as an acetabular component position and leg length measurement tool. Further clinical studies are required to evaluate preoperative planning and functional acetabular component measurement tools.

Additional Information
Disclosures