Abstract
Background Abnormal femoral torsion has been linked to osteoarthritis in the knee as well as to patellofemoral pain. Inexpensive, valid, and reliable methods for assessing femoral torsion are needed. Ultrasound (US) is a noninvasive and clinically accessible method that can be used for the assessment of bone morphology, such as femoral torsion.
Objective The objective of this study was to determine the concurrent validity of US for the measurement of femoral torsion with a reference method, magnetic resonance imaging (MRI).
Design Repeated measurements of femoral torsion were obtained with US and MRI in a laboratory setting.
Methods Twenty-eight people (4 men, 24 women; mean age=26.8 years [SD=4.0 years], mean body height=170.3 cm [SD=8.0 cm], mean body weight=64.7 kg [SD=9.8 kg]) participated in this study. T1-weighted axial oblique images of the femoral neck and epicondylar axis were acquired with a 1.5-T magnetic resonance system. Ultrasonographic measurements then were obtained by a tilting technique with a linear transducer that was 4.5 cm long and operated at a frequency of 10 MHz and a depth of 5 cm.
Results The average angles of anteversion measured by US and by MRI were 20.7 degrees (SD=11.0) and 19 degrees (SD=11.3), respectively. The reliability, reported as the intraclass correlation coefficient [ICC (2,1)], of repeated measurements of in vivo femoral torsion by US was .98. The reliability [ICC (2,1)] of magnetic resonance image analysis was .96. The standard error of the measurement for US was 2.2 degrees, and that for MRI was 1.9 degrees. The concurrent validity of US with MRI (R2) was .93 (r=.96).
Limitations Obtaining measurements by US requires appropriate training before data collection.
Conclusions Ultrasound measurement of femoral torsion has high concurrent validity with in vivo MRI and may be used when an assessment of bony morphology is needed but MRI is not available.
Because femoral transverse-plane alignment has been linked to femoral or hip rotation and related lower-extremity pathologies,1,2 its measurement has been of interest to both researchers and clinicians. Femoral torsion is the long-axis rotation of the femoral shaft relative to its neck in the transverse plane. This shaft-to-neck relationship becomes apparent when the femur is viewed in the long axis relative to the posterior condyles (Fig. 1). The angle formed by a line drawn through the femoral neck axis relative to a line drawn tangent to the posterior femoral condyles is the angle of anteversion. This angle has been reported to range from 5 to 40 degrees, but the average in adults is 10 to 20 degrees.3 A torsion angle of greater than 20 degrees is considered excessive femoral anteversion, whereas a torsion angle of less than 10 degrees is considered femoral retroversion.1,2
Femoral torsion as viewed in the long axis. The reference is a line drawn parallel to the posterior femoral condyles. The angle of torsion is found by measuring the angle created by the bisection of the axis of the femoral neck (a line connecting the centroids of the femoral head and shaft) and a line parallel to the tabletop on which the posterior condyles are resting. This method of torsion angle determination requires “overlaying” of the proximal and distal aspects of the femur.
The anteversion angle shows a gradual decrease with age. This change is attributed to the progressive unwinding of femoral torsion from an average of 45 degrees at birth to 8 to 15.5 degrees in adulthood.4–8 This decrease in the anteversion angle is associated with bone growth and is effected by muscle activity as children begin to bear weight, use muscles, and assume normal, adultlike gait patterns.9,10 Excessive femoral torsion is not uncommon and has been associated with certain neurologic and orthopedic conditions. For example, children with cerebral palsy have a high prevalence of excessive femoral anteversion, which contributes to difficulty with ambulation and, in conjunction with asymmetric pull by spastic hip muscles, leads to an increased risk for hip subluxation.11,12 Abnormal femoral torsion has been observed in several orthopedic conditions in adults, such as knee and hip osteoarthritis, patellofemoral pain, and acetabulum labral tears.1,2 Although Tonnis and colleagues1,2 showed that increased retroversion is associated with hip osteoarthritis, whether abnormal torsion contributes to the development of other common orthopedic conditions has not been well established. This factor, among other reasons relating to function, provided the motivation for and interest in studying femoral torsion and developing clinically accessible methods for its assessment.
Several imaging techniques for the measurement of femoral anteversion have been described. Biplanar radiography was first used to quantify femoral torsion; however, exposure to radiation makes it impractical for clinical use.13,14 Computerized tomography (CT) replaced biplanar radiography in the 1970s and became the gold standard for the accurate imaging and measurement of femoral torsion in comparison with an anatomical reference.15 Although CT enables the direct measurement of bony structures, it does so through a series of 2-dimensional x-ray images taken around a single axis of rotation, rendering it capable of primarily axial views. It also exposes patients to ionizing radiation, which has been shown to increase cancer risk. Recently, magnetic resonance imaging (MRI) was shown to have many advantages over CT. Although MRI is more expensive to perform, radiation exposure is not a concern with this method; in addition, MRI allows the use of different imaging planes, thereby improving visualization of the femoral neck axis.16 For these reasons, MRI has become the gold standard for musculoskeletal and anatomical imaging, specifically, the assessment of femoral neck torsion.17,18
Ultrasound is a noninvasive, inexpensive, and clinically accessible method that has been shown to be reliable for the assessment of bone morphology, such as femoral torsion.19 In recent years, several published studies of the use of US for anteversion measurement have deemed it to be a reliable method for femoral torsion measurement, but few studies have addressed its validity.19–23 Whereas MRI and radiography provide for visualization of the detailed 2-dimensional shape of bone, US uses surface contours on the femur as landmarks to measure the torsional angle directly without projecting 2 planes (proximal femur and distal femur) onto 1 combined reference plane, as is the case with MRI. Terjesen et al21 found a high correlation (r=.71) between the angle formed by a line drawn tangent to the head and greater trochanter, as measured by US, and the anteversion angle, as measured by biplanar radiography. Aamodt et al18 demonstrated that the head-trochanter tangent line measured by US highly corresponds (r=.95) to that measured by CT because the femoral head and greater trochanter landmarks are the same. In light of technological advances and the ability to view images in various planes, it now is necessary to establish the concurrent validity of US with the method considered by most researchers to be the current gold standard for the measurement of femoral anteversion (ie, MRI). Therefore, the specific aim of this study was to determine the concurrent validity of US for the measurement of femoral torsion with MRI as the reference method.
Method
Participants
Twenty-eight people (4 men, 24 women; mean age=26.8 years [SD=4.0 years], mean body height=170.3 cm [SD=8.0 cm], mean body weight=64.7 kg [SD=9.8 kg]) participated in this study. Before participation, all participants were informed about the nature of the study, and informed consent was obtained as approved by the Institutional Review Board of the University of Southern California. People were excluded from participation in the study if they had undergone any bony surgical realignment of the lower extremity or if they failed to meet any of the MRI safety requirements (ie, the presence of metal implants or pacemakers).
Instrumentation
B-mode ultrasonographic imaging was performed with a linear multidimensional transducer (Sonoline Antares*) that was 4.5 cm long and operated at a center frequency of 10 MHz; the axial and lateral resolutions of the US transducer at a 5-cm depth were 0.25 and 0.31 mm, respectively. Real-time B-mode US produces an image derived from a transducer that transmits and receives a high-frequency sound (here, 10-MHz) signal. The US echoes are converted into radiofrequency signals, which are converted in the image memory into an anatomically correct image displayed on the computer screen. An industrial inclinometer (Magnetic Polycast Protractor†) attached to the transducer recorded the angle of femoral torsion.
Magnetic resonance images were acquired with a 1.5-T MRI system.‡ In MRI, a powerful magnetic field is used to align the nuclear magnetization of atoms in the body. Radiofrequency fields are introduced to alter the alignment of this magnetization, causing the nuclei to rotate and produce a magnetic field signal that is detectable by the scanner. This signal can be manipulated with additional magnetic fields, and a visual image can be constructed. T1-weighted images of the proximal femur and distal femur were acquired with the following pulse sequence: repetition time=450 milliseconds, echo time=8.1 milliseconds, field of view=24 × 24 cm, matrix=256 × 256, and slice thickness=5 mm.
Procedure
Validation and reliability of a US transducer tilting technique for the assessment of femoral torsion.
Before participant testing, we validated the technique with 5 dry cadaveric femora placed on a stable flat surface and measured in 6 different positions. These measurements were obtained 3 times and repeated 1 week apart. The static femoral positions were documented with a photograph taken perpendicular to the plane of the femoral condyles (Fig. 1), as well as with a US image obtained with the transducer tilting technique described by Hudson et al.20 In brief, the angle of femoral torsion was the angle formed by a line drawn through the center of the femoral neck and head and a line representing the horizontal plane of the posterior femoral condyles. Using ultrasonography, the transducer was tilted until the image of the femoral head and neck was horizontal on the computer screen. By tilting the transducer to achieve the horizontal image of the femoral neck on the screen, we were, in essence, creating a new horizontal reference line. The relative angle of tilt (measured with the inclinometer) needed to achieve this horizontal image then was recorded as the angle of torsion.
A US gel pad§ placed over the greater trochanter of the femur served as a conduction medium. The greater trochanter, confirmed by palpation, and femoral neck were identified by imaging first (Fig. 2). The transducer then was oriented along their axis before being moved slightly medial to the center of the intertrochanteric line. A line drawn tangent to this line was parallel to the head-trochanter line described by Terjesen et al.21 In this position, the transducer was tilted until the image became horizontal on the screen; the corresponding tilt angle of the transducer was recorded from the attached inclinometer, rendering the angle of inclination.
Validation against photographs of a dry cadaveric femur. The following steps were taken during ultrasonographic imaging: the greater trochanter and femoral neck were identified, the transducer was oriented along their axis before being moved slightly medial to the center of the intertrochanteric line, the transducer was tilted until the image became horizontal on the computer screen, and the tilt angle of the transducer was recorded from the inclinometer.
To establish the reliability of the above-described method for participants, 1 examiner imaged 6 participants and averaged 3 measurements taken during 3 sessions, 1 week apart. Participants were positioned supine with the knees bent to 90 degrees off the edge of the plinth, a neutral hip position (0° of rotation, 0° of flexion, and 0° of abduction), and the feet flat on a bench at the end of the plinth (Fig. 3). The US transducer was first placed on the lateral hip to identify the greater trochanter. Next, the transducer was moved medially to identify the femoral neck, oriented along the long axis of the neck, and moved further medially to visualize the intertrochanteric line. Finally, the transducer was tilted until the image became horizontal on the screen, and the tilt angle of the transducer was recorded from the inclinometer (Fig. 3). The angle of tilt was taken to represent the angle of femoral torsion.
Ultrasonographic assessment of anteversion in vivo. Participants were positioned supine with the knees bent to 90 degrees and the lower legs and feet off the edge of the plinth, a neutral hip position, and the feet flat on a bench at the end of the plinth (below the level of the plinth). The US transducer was placed on the lateral hip to identify the greater trochanter and then moved medially to identify the femoral neck. The transducer was oriented along the long axis of the neck, and the intertrochanteric line was visualized. The transducer then was tilted until the image became horizontal on the computer screen, and the angle of tilt was recorded from the inclinometer.
The reliability of the US-guided measurement of anteversion for participants and the measurement of angles by MRI, rendering the anteversion angle, was established using intraclass correlation coefficients [ICC (2,1)]. The test-retest reliability [ICC (2,1)] of magnetic resonance image analysis was .96 (95% confidence interval=.91–.98), and that of US data capture was .98 (95% confidence interval=.96–.99).
Acquisition of data from participants.
Participants attended 2 separate testing sessions. One session involved US imaging to guide the inclinometer measurement of the femoral torsion angle. The other session involved MRI of the femoral head and neck and the distal femur. The examiner was unaware of the results of the MRI because those were taken, recorded, and kept by another investigator.
Ultrasonographic acquisition of images and concurrent measurement of femoral anteversion with a US-guided inclinometer.
Participants were positioned supine with the knees bent to 90 degrees off the edge of the plinth, a neutral hip position, and the feet flat on a bench at the end of the plinth. As the US transducer was placed on the hip to image the greater trochanter, the femoral neck and intertrochanteric line were viewed on the computer screen. The transducer then was tilted until the image became horizontal on the screen; the tilt angle of the transducer was recorded from the inclinometer, signifying the anteversion angle. Essentially, as each image was acquired, a concurrent measurement was taken. The total acquisition time was approximately 10 minutes.
Magnetic resonance acquisition of images.
The MRI technician positioned participants supine with the hip joint supported by pillows in a neutral position (0° of rotation, 0° of abduction, and 0° of flexion). Two image series were obtained. First, an axial oblique series of images were acquired parallel to the femoral neck, bisecting its superior and inferior borders. Second, an axial oblique series of images were acquired through the epicondylar axis. The total imaging time was approximately 10 minutes.
Measurement of femoral anteversion on magnetic resonance images.
Images were analyzed with Image J software (version 1.36b).‖ An image representing the transection of the femoral neck was first analyzed to determine the femoral neck angle with respect to the horizontal border of the image field of view. The femoral head and shaft were outlined by ellipses, as in Figure 1, and their centroids were determined. A line connecting these centroids was used to define the femoral neck axis in the transverse plane (Fig. 4A). Next, the angle between the femoral neck axis and a horizontal line on the image field of view was measured (Fig. 4A). The angle was considered positive if the femoral head was anterior to the femoral shaft and negative if it was posterior to the femoral shaft.16
Magnetic resonance imaging assessment of femoral anteversion in vivo. (A) The femoral head and shaft were outlined, and their centroids were determined. A line connecting the centroids defined the femoral neck axis in the transverse plane relative to a horizontal line on the image field of view. (B) An axial oblique image through the femoral condyles was used to determine the femoral epicondylar axis. A line connecting the posterior femoral condyles was drawn relative to a horizontal line on the image field of view. For determination of the angle of femoral anteversion, the femoral neck axis angle (A) was added to the femoral epicondylar axis angle (B).
An axial oblique image through the femoral condyles was used to determine the femoral epicondylar axis. The most posterior aspect of each femoral condyle was defined, and a line connecting these 2 areas was drawn (Fig. 4B). This line defined the femoral epicondylar axis in the transverse plane relative to a horizontal line on the image field of view (Fig. 4B). The femoral epicondylar axis angle was considered positive if the lateral condyle was anterior to the medial condyle (indicating a medially [internally] rotated position) and negative if the lateral condyle was posterior to the medial condyle (indicating a laterally [externally] rotated position).16 For determination of the angle of femoral anteversion, the femoral neck axis angle was added to the femoral epicondylar axis angle.16
Data Analysis
For statistical analyses, an average of 3 measurements was used. The validity of the US tilting method relative to the photographs was established with correlational analyses. For the determination of concurrent validity, angles of anteversion measured ultrasonographically by 1 examiner and those determined by MRI were plotted against each other, and r and R2 values were calculated.
Results
The demographic and anthropometric characteristics of the participants are shown in the Table. The average angles of anteversion measured by US and by MRI were 20.7 degrees (SD=11.0) and 19 degrees (SD=11.3), respectively.
The correlation of the ultrasonographic tilting method with the photographs was .93. The standard error of the measurement (SEM) for US was 2.2 degrees, and that for MRI was 1.9 degrees. The angles of anteversion measured ultrasonographically by 1 examiner and those determined by MRI were plotted against each other (Fig. 5). The r and R2 values were .97 (95% confidence interval=.92–.98) and .93, respectively.
Measurement of femoral anteversion with ultrasound (US) (abscissa) and magnetic resonance imaging (MRI) (ordinate). Comparison of data rendered values of r=.97 and R2=.93, showing excellent concurrent validity.
Discussion
In the present study, we established the concurrent validity of ultrasonographic measurements of femoral torsion with the method considered to be the current gold standard for musculoskeletal imaging (ie, MRI). We took care to develop a reliable method that accurately measures the amount of torsion in the femur by using validation with dry femora and masking the examiner to the results of MRI. Our ultrasonographic procedures were based on those developed by Hudson et al,20 with minor modifications made in participant positioning to improve the visualization of the femur. In the study by Hudson et al,20 participants were seated and semirecumbent, but the examiner in the present study found it difficult to clearly visualize the appropriate anatomical landmarks and read the tilt angle from the inclinometer when that positioning was used. Supine positioning of participants allowed better visualization and ease of measurement. This method, however, requires systematic training with cadaveric femora, as well as with study participants, to produce confidence in image acquisition, viewing, and rendering of reliable and valid measurements.
The results of the present study of femoral torsion with the greater trochanter and intertrochanteric line as landmarks for visualization were consistent with previously reported findings.19,21 Trust in methods aimed at the quantification of morphometric dimensions requires computation of the SEM, which is an estimation of the difference between the measured or estimated values and the true values. In the present study, the SEM for US was 2.2 degrees, and that for MRI was 1.9 degrees. Given these values, we could be 95% confident that the true angles of femoral torsion would fall within 4.4 degrees when measured with US and 3.8 degrees when measured with MRI. Terjesen and colleagues3,21 found that the accuracy of ultrasonography was within 5 degrees of that of CT, in agreement with our findings. One difference between the 2 studies was that Terjesen and colleagues3,21 suggested CT had a tendency to overestimate the anteversion angle when compared with US.
Despite excellent concurrent validity between the 2 methods, discrepancies in the measurements at higher angles of femoral anteversion were noted (Fig. 5). Terjesen and colleques3,21 suggested that, in hips with an increased angle of femoral anteversion (>35°–40°), the image of the femoral neck and the greater trochanter becomes indistinct because of the relative posterior position of the trochanter and because the transducer tends to lose contact with the skin of the trochanter area. We agree with those observations. Measurements acquired by US had a tendency to overestimate the amount of femoral torsion at excessive ranges of anteversion (eg, >30°) in 2 participants in our cohort. It is less likely that MRI underestimated the anteversion angle because the angle was calculated by subtracting the angles measured in single distant slices, taking into account the entire femoral morphology. Therefore, we recommend that in studies of populations anticipated to have high levels of anteversion (≥35°), such as newborns and children, the US method should be modified to account for this possible problem.
The limitations of the present study include a moderate sample size and narrow ranges of ages and body masses, which may make it difficult to generalize the success of the method studied to other populations. Older patients may not be able to assume the test position used in the present study; therefore, it may need to be modified for comfort. Excess soft tissue around the area of the greater trochanter makes it difficult to palpate, but its location can be confirmed by sonographic imaging. It is easier to visualize bony structures on US images because bone is more echoic than soft tissue. Thus, any error of manual palpation will be corrected by repositioning of the transducer. Should the issue of excessive body mass continue to impair visualization of the femur because of the limited depth that an US transducer with a center frequency of 10 MHz can achieve, the use of a US transducer with a center frequency of 5 Hz will allow for better visualization at greater depth. Further studies are needed to establish reliability with a larger sample that is anthropometrically diverse to permit better generalization of the findings.
Another limitation of the present study was the type of inclinometer used. We used an inclinometer with a weighted pendulum that depended on position in the sagittal plane to swing effectively. If the inclinometer was tilted too far in either direction in the sagittal plane, the pendulum would stick and yield inaccurate readings. In contrast, Hudson et al20 used a digital inclinometer that did not depend on position to accurately measure the angle of femoral anteversion. This limitation was identified during the initial validation of the method by use of dry femora. Care was taken to ensure that this problem did not occur during data acquisition by repeating the measurement 3 times and averaging the values.
Ultrasound has been found to be safe, inexpensive, clinically accessible, and reliable; these properties support the clinical usefulness of this method for measuring femoral torsion. The clinical motivation for assessing bone morphology stems from observations that excessive anteversion has been found in people with lower-extremity musculoskeletal pathologies. A simple static analysis of femoral morphology and a shortened abductor lever arm illustrates the functional shortcomings associated with excessive anteversion, which may contribute to reduced hip muscle performance.24 Although this problem cannot be corrected through nonsurgical means, knowledge of the problem may allow for a more realistic determination of clinical prognosis if a painful condition is present. Alternatively, it may lead to a thoughtful exploration of movement strategies in which the excessive anteversion is less deleterious. Two clinical tests are used to assess the degree of long bone torsion at the femur: the categorical Ryder method and the numerically reported Craig test. A recent study by Souza and Powers23 suggested that although clinicians measured femoral anteversion reliably, their measurements had moderate agreement with those obtained by MRI (95% confidence interval for a difference of 11.8°).
An accurate and accessible method for the measurement of femoral anteversion will assist clinicians in determining whether femoral morphology is a component of reduced hip muscle performance. With this information, prognosis and length of an intervention can be determined with greater ease, ultimately improving patient care. The present study demonstrated high concurrent validity of US with the method considered to be the current gold standard for musculoskeletal imaging (ie, MRI). Therefore, we confidently recommend US for the determination of femoral torsion in an adult population. Studies examining the use of this method in clinical settings are needed to further explore the possible impact of femoral torsion on clinical decision making.
Footnotes
-
Dr Kulig, Dr Harper-Hanigan, and Dr Souza provided concept/idea/research design, writing, and data analysis. Dr Harper-Hanigan provided data collection. Dr Kulig provided project management and facilities/equipment. Dr Souza provided participants. All authors provided consultation (including review of manuscript before submission).
-
This study was approved by the Institutional Review Board of the University of Southern California.
-
↵* Siemens Medical Systems Inc, Ultrasound Group, PO Box 7002, Issaquah, WA 98027.
-
↵† Empire Level Manufacturing Corp, Corporate Headquarters, 929 Empire Dr, PO Box 800, Mukwongo, WI 53149.
-
↵‡ General Electric Medical Systems, 800 Centennial Ave, PO Box 1327, Piscataway, NJ 08855.
-
↵§ Parker Laboratories, 286 Eldridge Rd, Fairfield, NJ 07004.
-
↵‖ National Institutes of Health, 9000 Rockville Pike, Bethesda, MD 20892.
- Received November 24, 2009.
- Accepted June 6, 2010.
- © 2010 American Physical Therapy Association
References
Issue highlights
