Interobserver and Intraobserver Repeatability and Reliability of Three-Dimensional Postoperative Evaluation Software in Total Hip Arthroplasty

Background: We investigated the reliability of measurement of the alignment and position of the prosthetic components after total hip arthroplasty (THA) using three-dimensional computed tomography (CT)-based postoperative evaluation software. Methods: We evaluated the postoperative CT data from 20 hips in 20 patients, using postoperative evaluation software. The alignment and three-dimensional positioning of the cup and stem were assessed by eight orthopedic surgeons for repeatability (intraobserver reliability) and reproducibility (interobserver reliability) of postoperative evaluation software using intra-class correlation coecients (ICC). The radiographic inclination (RI) and radiographic anteversion (RA) of the cup, and anteversion, varus-valgus angle and exion-extension angles of the stem were measured for alignment. The implant positioning was measured along three axes, X-axis (transverse), Y-axis (sagittal), and Z-axis (longitudinal). Results: The intra- and inter-observer ICC of alignment measurement were very good for both cup and stem (0.86–1.00). The intra-observer and inter-observer ICC of cup positioning were very good in X-axis and Y-axis (0.91–0.94) and good in Z-axis (0.68–0.80). The intra-observer and inter-observer ICC of implant positioning were very good for the stem (0.98–0.99) for all axes. Conclusion: CT-based postoperative evaluation software was able to evaluate the position of total hip implants with high reproducibility. at 2-week intervals by the same observer (A) for the 20 patients. The intra- and interobserver differences of alignment measurements and implant position for acetabular and femoral components were calculated.


Background
In primary and revision total hip arthroplasty (THA), appropriate implant size and placement is important to prevent complications such as dislocation [1][2] [3], wear of the polyethylene liner, and loosening of the components [4][5] [6]. Lewinnek et al. proposed a "safe zone" of cup alignment, and Widmer et al.
proposed an optimal combination of cup alignment and stem anteversion to minimize the risk of impingement [7]. To achieve precise implant placement, the use of a navigation system and a uoroscopy-guided technique have been reported [8] [9]. Currently, the gold standard for postoperative assessment of component orientation, inclination, anteversion of the acetabular component, and varusvalgus angle of the femoral component is conventional plain radiography [10][11] [12]. While these measurement methods can evaluate implant alignment accurately, they are less reliable in assessing the components' exact size and position in the acetabulum and femur, [13] because it is di cult to assess three-dimensional (3D) implant alignment using a 2-dimensional (2D) plain radiograph or intraoperative measurement [1] [4]. In the case of revision total hip arthroplasty with unkown implant size because of undergoing primary total hip arthroplasty at other hospitals, case urgency, patient migration, clerical error, or destruction of records, it would be very helpful to predict the implant size and alignment accurately [14].
A computed tomography (CT)-based method using low-dose digital stereo-radiography, commercialized as the EOS imaging system, has been reported to have good reliability [15][16] [17].
Recently, a CT-based 3D preoperative planning and postoperative evaluation software, which can measure the alignment and 3D position of the implant has become available and has shown good reliability for preoperative planning [18] [19]. In total knee arthroplasty good reliability has been reported regarding the femoral and tibial component positions and alignment [20]. The aim of this study was to evaluate the accuracy, intraobserver, and interobserver reliability of the CT-based postoperative evaluation software in THA.

Study design
We performed a retrospective study using CT-based 3D preoperative planning and postoperative analysis software for THA. The study was approved by our hospital institutional review board and all patients gave written informed consent before any study-related procedures were performed.

Subjects
We selected perioperative CT data from 20 hips in 20 patients who underwent primary THA for osteoarthritis (OA) of the hip and osteonecrosis of the femoral head (ONFH). The operations were all carried out by the same senior surgeon (N.J.

Planning and analysis
Both preoperative and postoperative CT scans from the bilateral iliac wing to the tibial plateau were performed with a slice thickness of 1 mm using a helical CT scanner (Aquilion ONE; TOSHIBA Medical Systems Corporation, Tokyo, Japan). The CT data were transferred to ZedHip (Lexi Corporation, Tokyo, Japan). This preoperative planning software enables the surgeon to simulate placing the prosthetic components into their proper positions in the 3D space of the CT data using a computer-aided design model [18]. It is also possible to compare the postoperative component size and position with the position planned preoperatively. It determines the coordinates of the acetabular and femoral sides using skeletal reference points. Each coordinate was also adapted for postoperative implant positioning and alignment evaluation. The cup positioning and alignment were evaluated by a functional pelvic plane coordinate system ( Fig. 1) and the stem was evaluated by the coordinate system recommended by the International Society of Biomechanics (ISB) (Fig. 2) [23] [24].Preoperative and postoperative coordinates were uni ed by an "image matching" system mounted on postoperative evaluation software. The "Image matching" system can automatically superimpose the preoperative and postoperative CT images (Fig. 3).
The following parameters regarding implant alignment and positioning were calculated automatically and component size was manually by postoperative evaluation software [1].

Component size accuracy
We investigated the concordance rates of each component (head, cup, and stem) between 3D CT-based postoperative templating and the actual implant used.

Alignment measurement
Radiographic inclination (RI) and radiographic anteversion (RA) were evaluated for the acetabular component alignment. RI is the angle between the acetabular axis and the Z-axis projected onto the XZ plane and RA is the angle between the acetabular axis and the Y-axis projected onto the XZ plane ( Fig. 4).

Statistics
The statistical analysis was performed with JMP® 14 (SAS Institute Inc., Cary, NC, USA). To evaluate the component size accuracy, eight observers performed 3D CT-based postoperative templating without knowing the clinical information. The accuracy was measured with concordance rates of postoperative templating and the actual implant size within a range of ± 1size. The repeatability (intraobserver reliability) and reproducibility (interobserver reliability) of postoperative evaluation software were calculated using intraclass correlation coe cients (ICC). The measurements were performed by three independent observers (A, B, and C) and two successive measurements were performed at 2-week intervals by the same observer (A) for the 20 patients. The intra-and interobserver differences of alignment measurements and implant position for acetabular and femoral components were calculated.

Results
Component size accuracy Table 2 shows the concordance rates of each parts. The exact concordance rate of head, cup, and stem was 96.6% (309/320), 94.7% (303/320), and 97.8% (313/320), respectively. The concordance rates of postoperative templating within a range of ± 1size was 100% in all component.  Table 3 shows the intra-and interobserver ICC for alignment measurement. The intra-observer and interobserver ICC were 0.972 and 0.970-0.982 in RA, respectively, and 0.955 and 0.892-0.965 in RI respectively, for the acetabular component. The intra-observer and inter-observer ICC were 0.993 and 0.999, respectively, in anteversion, 0.956 and 0.987-0.995, respectively, in varus-valgus angle, and 0.991 and 0.994-0.997, respectively, in exion-extension angle for the femoral component.

Discussion
This study investigated the component size accuracy, and the intra-and inter-observer reliability of CTbased postoperative evaluation software in THA. The results showed that postoperative evaluation software can evaluate postoperative implant size and orientation with perfect to very good reliability.
The use of two-dimensional plain radiography has been standard for the evaluation of the acetabular component. Radiographic inclination is measured between the long axis of the implant and the tear-drop line. The ratio of the short axis and long axis of the implant is widely used for assessing the anteversion [12]. Some reports have suggested that the reliability of the acetabular component measurement on plain radiographs using a normal PACS system is high [26][27] [28]. The EOS system is a novel imaging method using biplanar low-dose X-rays, which can evaluate implant alignment semiautomatically [16]. Lazennec et al. reported that the reliability of conventional acetabular component measurement on plain radiographs was lower than the performance of the EOS system [13]. However, the measurement of radiographic anteversion is a problem that is yet to be solved. In our study, we demonstrated very good reproducibility for radiographic anteversion. Since the pelvic coordinate axis is set after CT imaging and is available as a reference, it is not affected by posture and limb position. We found a slight decrease in the reliability of inclination. With this system, the reference point for the distal implant edge can affect the inclination and positioning on the Z-axis. The relatively lower reliability of the inclination and Z-axis indicates that the reference point may vary.
Although many reports on the reliability of evaluation methods after THA have concerned acetabular components, there are a few reports on the femoral component [16][17] [27]. Some studies focused on the stability of the component and the bone reaction of the femur [29], however, there have been no studies exploring the reliability of the femoral component varus-valgus and exion-extension angles. The alignment of the femoral component is, however, thought to be important in preventing impingement and dislocation of the femoral head [30]. Lee YK et al. reported that the reliability of identifying stem anteversion by plain radiographs is high; intraobserver reliability was 0.944 and interobserver reliability 0.934 [31]. Another study using the EOS system reported intraobserver reliability of 0.998 and interobserver reliability of 0.997 [17]. The reliability of femoral component anteversion in this study is equal to or better than other methods. In addition, with this method, the proximal bone axis and 3D coordinate system of the femur can be placed automatically from the CT image, so it is not affected by the orientation of the lower limbs.
The 3D position of each implant is thought to affect the biomechanics of the joint [32], the impingement of components, and leg length [33]. Therefore, the 3D position of each implant is also measured and installed by navigation [34]. Nonetheless, there are few reports on its positional evaluation method [35]. Leg length is measured from the transverse line and femoral offset is measured from the hip center on the X-ray. However, the sagittal positional shift (Y-axis, in this study) cannot be evaluated without a 3D approach. This method, therefore, can be useful in evaluating the leg length discrepancy or postoperative biomechanics of the hip joint.
This system allows us to simulate the range of motion and evaluate impingement after surgery. Precise evaluation of the implant position and alignment with residual osteophytes after surgery, identi ed by postoperative CT, can enable the assessment of the risk of dislocation due to impingement. It can also provide information that is useful for patient guidance. Another potential clinical advantage of this system is the long-term evaluation of implant stability since this method can evaluate implant alignment and position three-dimensionally; it would be possible to detect even slight loosening.

Conclusion
The CT-based postoperative evaluation software was able to evaluate the position of the implant with high reproducibility. It can, therefore, be a useful tool for evaluating the accuracy of implant size and placement after primary THA and before revision THA.    direction is from distal to proximal femur. X-axis: The line perpendicular to the Z-axis and passing through the origin on the plane created by the three points of CFH, ME and LE. Positive direction is from left to right of the body. Y-axis: The line perpendicular to the Z and X axes and passing through the origin. Positive direction is from the posterior to the anterior of the body.

Figure 3
Image matching of the pre-and postoperative CT images. Preoperative CT images were fused to the postoperative images automatically as bone surfaces matched. Blueline presents the bone 3D image of the preoperative femur and the white line presents the bone 3D image of the preoperative pelvis.