Computed tomography-based automated 3D measurement of femoral version: Validation against standard 2D measurements in symptomatic patients

Abstract

To validate 3D methods for femoral version measurement, we asked: (1) Can a fully automated segmentation of the entire femur and 3D measurement of femoral version using a neck based method and a head-shaft based method be performed? (2) How do automatic 3D-based computed tomography (CT) measurements of femoral version compare to the most commonly used 2D-based measurements utilizing four different landmarks? Retrospective study (May 2017 to June 2018) evaluating 45 symptomatic patients (57 hips, mean age 18.7 ± 5.1 years) undergoing pelvic and femoral CT. Femoral version was assessed using four previously described methods (Lee, Reikeras, Tomczak, and Murphy). Fully-automated segmentation yielded 3D femur models used to measure femoral version via femoral neck- and head-shaft approaches. Mean femoral version with 95% confidence intervals, and intraclass correlation coefficients were calculated, and Bland-Altman analysis was performed. Automatic 3D segmentation was highly accurate, with mean dice coefficients of 0.98 ± 0.03 and 0.97 ± 0.02 for femur/pelvis, respectively. Mean difference between 3D head-shaft- (27.4 ± 16.6°) and 3D neck methods (12.9 ± 13.7°) was 14.5 ± 10.7° (p < 0.001). The 3D neck method was closer to the proximal Lee (-2.4 ± 5.9°, -4.4 to 0.5°, p = 0.009) and Reikeras (2 ± 5.6°, 95% CI: 0.2 to 3.8°, p = 0.03) methods. The 3D head-shaft method was closer to the distal Tomczak (-1.3 ± 7.5°, 95% CI: -3.8 to 1.1°, p = 0.57) and Murphy (1.5 ± 5.4°, -0.3 to 3.3°, p = 0.12) methods. Automatic 3D neck-based-/head-shaft methods yielded femoral version angles comparable to the proximal/distal 2D-based methods, when applying fully-automated segmentations.

Keywords: deep learning; femoral osteotomy; femoral version; hip arthroscopy.

FIGURE 1

Automated 3D measurement of femoral version (neck method and head‐shaft method) and the four different 2D measurements of femoral version are shown. For the femoral neck 3D method, the femoral neck axis was defined as the best fitting line between the medial and lateral femoral neck surfaces. For the head‐shaft 3D method the angle between the line connecting the femoral head center with the center of the femoral shaft at the level of the lesser trochanter was chosen. Using the femoral head center as the first proximal reference the 2D methods differed regarding the selection of the second proximal reference while the femoral condyle axis served as reference distal reference for all measurement methods alike. (A) For the Lee et al method, the femoral head center is connected on the same slice as the proximal femoral neck axis at the level of the greater trochanter. (B) For the Reikeras et al. method the proximal reference at the level of the femoral neck is selected where the anterior and posterior cortices appear parallel. (C) For the Tomczak et al.17 method the center of the greater trochanter as the base of the femoral neck is chosen. (D) For the Murphy et al. method the proximal femoral axis was defined at the base of the femoral neck just above the lesser trochanter. (A–D) Used with permission from Schmaranzer F, Lerch TD, Siebenrock KA, et al. 2019. Differences in Femoral Torsion Among Various Measurement Methods Increase in Hips With Excessive Femoral Torsion. Clin Orthop Relat Res 477(5), 1073–1083. https://doi.org/10.1097/CORR.0000000000000610.

FIGURE 2

(A) Reconstructed 3D models and detected anatomical landmarks. (B) The normal vector of the plane (gray triangle) passes through epicondyle reference points (red circles) and femoral head center (blue square) was used to define the X axis. Magnified view of the epicondyle model, the fitted cylinder, and its reference points. Magnified view of the femoral head, the fitted sphere, and center of the fitted sphere. (C) The line connecting epicondyles center (red square) and femoral head center (blue square) defines the Y axis.

FIGURE 3

Determining the shaft axis and the intertrochanteric region. (A) The zone on the femur that does not belong to the femoral head or neck is initially considered as the femoral shaft. Planes (cyan) with normal vector of the initial shaft axis (obtained from principal axes) with 5 mm intervals starting from the bottom of the region are defined along the axis. (B) The intersections of these planes with femur was used to establish best‐fit circles (blue). (C) The red circle (control point) indicates the center of the femoral shaft right below the lesser trochanter. The solid black line indicates the final axis of the femoral shaft.

FIGURE 4

Determining the posterior condyle axis. The condyle is locally devided into medial and lateral zones. For each zone, the most posterior point (the point with minimum X) is selected. The line connecting these two points is the posterior condyle axis.

FIGURE 5

Determining the femoral neck (FN) axis. The initial neck axis was established based on the center points of the medial and lateral FN surfaces and was then refined based on the center of the best‐fit circles along the neck. Only one of the best‐fit circles are shown.

FIGURE 6

Assessment of mean femoral version across the four 2D methods and two 3D‐based methods.

FIGURE 7

(A–D) Bland Altman plots to evaluate the systematic bias between 2D and 3D proximal and distal measurement methods. The solid black line represents the mean difference between the respective 2D and 3D methods. The dark, dashed black line represents the 95% limits of agreement. The lighter dashed line represents 1 standard deviation of the corresponding intra‐rater reliability of CT measurements displayed in Table 3. (A) Between the Lee‐ and 3D neck methods, 35 of 57 (61.4%) measured differences were within ± 1 SD of 4.3°. (B) Between the Reikeras‐ and 3D neck‐methods, 39 of 57 (68.4%) differences were within ± 1 SD of 3.6°. (C) Between the Tomczak‐ and head‐shaft methods, 26 of 57 (45.6%) differences were within ± 1 SD of 3.2°. (D) Between the Murphy‐ and head‐shaft methods, 29 of 57 (50.9%) differences were within ± 1 SD of 3.0°. LOA, limits of agreement.

FIGURE 8

A 32‐year‐old woman undergoing CT for assessment of femoral version using 2D and 3D measurement methods. (A) Section through the femoral rotation center and femoral condyles which serve as the most proximal landmark and the distal reference axis of 11°. (B–E) Femoral version angles increase from the proximal to the distal femoral landmarks: (B) 1° for the Lee method, (C) 6° for the Reikeras method, (D) 26° for the Tomczak method, (E) 29° for the Murphy method. (F) Segmentation of the entire femur for automated 3D measurement of femoral version is shown. The proximal 2D methods of Lee and Reikeras compared well to the 3D femoral neck version of 2° while the distal 2D methods compared well to the 3D head shaft method of 27°.