ABSTRACT
Acetabular dysplasia is a proposed cause of osteoarthritis of the hip and techniques to identify acetabular dysplasia are important in early diagnosis and prevention of osteoarthritis of the hip. The aim of this study is to develop a method to measure acetabular coverage and volume using computed tomography (CT) images. A software program was used to create three-dimensional models of the acetabulum and femoral head and measure the acetabular volume of each hip from CT images. The three-dimensional models of the acetabulum and femoral head were imported into a post-processing software program to determine the surface areas for the femoral head covered by the acetabulum and the acetabular area covering the femoral head. Using CT scans gathered from 35 individuals, the results of this study demonstrated that the two surface area measurements were linearly correlated with the acetabulum volume. Positive linear relationships with R-squared values of 0.53 and 0.64 were found when comparing the acetabulum volume with the surface area of the femoral head covered by the acetabulum and the surface area of the acetabular area covering the femoral head. The method presented in this study for measuring acetabular coverage and volume using CT images could be used for the development of a more reliable method of diagnosing acetabular dysplasia.
INTRODUCTION
Osteoarthritis of the hip is characterized by joint pain and loss of mobility. According to a 1999 survey, approximately 10 million American adults have been diagnosed with osteoarthritis of the hip [1]. Acetabular dysplasia is a well known precursor of hip osteoarthritis and estimates suggest that it causes secondary osteoarthritis of the hip in 25-50% of patients by 50 years of age [2]. In one study, patients with acetabular dysplasia were 4.3 times as likely to develop incident radiographic osteoarthritis than individuals without acetabular dysplasia [3]. Acetabular dysplasia is a condition associated with inadequate development of the acetabulum and is characterized by a shallower hip socket. Due to the shallowness of the hip socket, patients with acetabular dysplasia generally have incomplete acetabular coverage of the superior and anterior femoral head causing the forces in the hip joint to be concentrated across a smaller contact surface area of the acetabulum. Early diagnosis of acetabular dysplasia is important because preventative measures such as pelvic osteotomies can improve patient symptoms and delay the onset of osteoarthritis.
Acetabular dysplasia is typically assessed from an anterioposterior (AP) radiograph of the hip using Wiberg’s center-edge angle [4]. Sharp’s acetabular angle is another method used to assess acetabular dysplasia from an AP pelvic radiograph [5]. However, pelvic rotation and tilt are difficult to control and have been shown to significantly affect the image projection on a plain radiograph, resulting in discrepancies in the measurement of Wiberg’s or Sharp’s angles [6]. While plain radiographs do provide a means of diagnosing acetabular dysplasia, the lack of three-dimensional (3D) information may not allow the degree or location of dysplasia to be properly assessed. In prior research, several methods have been developed to determine hip joint morphology from plain radiographs, but these methods are complex and require several assumptions [7-10]. These methods assume a spherical shape for the femoral head and acetabulum, thus dysplastic hips with an obvious oval-shape cannot be accurately assessed using these methods.
Ashley A. Weaver1,2, Thomas D. Gilmartin1,2, Adam W. Anz1, Allston J. Stubbs1, Joel D. Stitzel1,2
1Wake Forest University School of Medicine, Medical Center Blvd, Winston-Salem, NC 27157
2Virginia Tech – Wake Forest University Center for Injury Biomechanics, Medical Center Blvd,
Winston-Salem, NC 27157
METHODS
Thirty-five normal individuals with complete pelvis computed tomography (CT) scans were selected from the Emergency Department and Department of Orthopaedic Surgery records at Wake Forest University Baptist Medical Center. Institutional Review Board (IRB) approval was obtained for this study for an academic medical center retrospective chart review. Fifteen female and twenty male patients were selected that ranged in age from 18 to 40 years. These patients were skeletally mature adults with a Risser score of 5. Patients with Tonnis scores of 0 or 1 were selected that had no evidence of osteoarthritis or joint space narrowing. None of the patients selected had any evidence or history of hip trauma or surgery.Pelvis CT images were imported into the software program, Mimics (Materialise, Leuven, Belgium) for segmentation. Segmentation tools in Mimics were used to isolate the femoral head and acetabulum of the left and right hips of each patient resulting in a total of 70 hips in the dataset. The method used to isolate the femoral head and acetabulum of each hip did require some user interaction because manual editing was often necessary to separate the femoral head from the acetabulum. The two-dimensional (2D) masks were then used to create 3D models of the femoral head and acetabulum of each hip in Mimics. A smoothing operation using 15 iterations with a smoothing factor of 0.4 was used to smooth the 3D models in order to reduce artificial bumps from pixilation. The 3D models were then exported to STL files for further processing.STL files of the femoral head and acetabulum of each hip were imported into the program, LS-PrePost (Livermore Software Technology Corporation, Livermore, California). The offset tool in LS-PrePost was used to further isolate the regions of overlap on the acetabulum and femoral head for the purpose of obtaining surface areas for the femoral head covered by the acetabulum and the acetabular area covering the femoral head. Nodes of interest on the acetabulum were first isolated by selecting an element in the middle of the acetabulum and then applying an adaptive propagation at an angle of 7 degrees. A similar method was used to isolate the nodes of interest on the femoral head using an adaptive propagation at an angle of 4 degrees. For the isolated acetabulum and the femoral head, care was taken to ensure that all nodes on the overlapping regions of the two models were selected, and a tool was used to select any islands of nodes that were not selected using the adaptive propagation.The newly created shells of the isolated acetabulum and femoral head were then displayed together, so that the surface areas of the regions of overlap on the acetabulum and the femoral head could be obtained. All nodes on the isolated femoral head were first selected, and then nodes that did not overlap with the isolated acetabulum were removed (Figure 1A). This was completed by aligning the edges of the acetabulum with the horizontal axis and removing nodes above the overlap. These nodes were used to create a new shell that represented the surface area of the femoral head covered by the acetabulum (Figure 1B). To find the surface area of the acetabulum covered by the femoral head, a similar operation was performed. The isolated acetabulum and femoral head were displayed together and all nodes on the isolated acetabulum were selected. Nodes that did not overlap with the isolated femoral head were removed (Figure 1C) and the remaining nodes were used to create a new shell that represented the surface area of the acetabulum covering the femoral head (Figure 1D). The measurement tool in LSPrePost was then used to calculate the surface areas of the femoral head covered by the acetabulum and the acetabular area covering the femoral head from the two newly created shells.
Figure 1. Selection of surface area shells. A. Isolated femoral head and acetabulum with overlapping femoral
head region nodes highlighted. B. Femoral head area covered by acetabulum. C. Isolated femoral head and
acetabulum with overlapping acetabulum region nodes highlighted. D. Acetabular area covering femoral head.
To obtain the volume of each acetabulum, a copy of the acetabulum mask in Mimics was created. On each of the sagittal slices on this mask, a line was drawn across the acetabulum to enclose the volume of the acetabulum (Figure 2A). A tool was used to fill in the volume inside the enclosed area. A Boolean operation was then used to subtract this mask from the original acetabulum mask to create a new mask of the acetabulum volume (Figure 2B). The volume of the acetabulum was then calculated from this mask in Mimics.
Results
The results of the study are summarized in Table 1 and graphically represented in Figures 3A and 3B. Data from both the right and left hips is presented in all tables and graphical representations. Averages and standard deviations of the measured surface areas and acetabulum volumes for all 70 hips studied are summarized in Table 1. Scatter plots of the acetabulum volumes and the overlapping surface areas on the acetabulum and femoral head are depicted in Figures 3A and 3B. A positive linear correlation with an R-squared value of 0.64 was found between the acetabular volume and the acetabular area covering the femoral head (Figure 3A). Similarly, a positive linear relationship with an R-squared value of 0.53 was found when plotting the femoral area covered by the acetabulum versus the acetabulum volume (Figure 3B).
DISCUSSION
Conventional methods to diagnose acetabular dysplasia such as Wiberg’s center-edge angle or Sharp’s acetabular angle rely on plain radiographs to measure the acetabular coverage of the femoral head. However, these methods are limited in that they do not provide 3D information and can be affected by patient positioning and pelvic orientation.
| Femoral head covered by acetabulum, cm2 | Acetabular coverage of femoral head, cm2 | Acetabulum volume, cm3 |
|---|---|---|
| 25.5 ± 4.67 | 35.1 ± 5.54 | 29.9 ± 6.09 |
Figure 3. Scatter plots of overlapping surface area and acetabular volume measurements. A. Acetabular area covering femoral head versus acetabular volume. B. Femoral area covered by acetabulum versus acetabular volume.
The ability to accommodate for differences in pelvic orientation is particularly important for dysplastic hips because the patient may exhibit compensating mechanisms such as pelvic tilt to make up for insufficient acetabular coverage [7]. Although methods have been proposed to improve the assessment of acetabular coverage and accommodate for pelvic tilt and rotation on plain radiographs, these methods rely on assumptions that are not always feasible for patients with acetabular dysplasia [7-10].
The methodology presented in this study enables surface area and volume measurements of the hip joint to be calculated directly from CT images of a patient.
This method is not limited by patient orientation and incorporates 3D information instead of relying on a 2D image projection to assess acetabular coverage of the femoral head. Contrary to previous studies that utilize plain radiographs to estimate acetabular coverage, no simplifying assumptions on the patient’s hip joint morphology are necessary to employ the method outlined in this study.
Although care was taken to eliminate sources of error in this study, the method presented requires some user interaction to calculate surface areas and volumes from the CT images. Introduction of error due to user interaction is possible during the CT image segmentation and the isolation of the overlapping regions of the acetabulum and femoral head. It would be useful to conduct future studies to analyze the accuracy and efficacy of the developed method and compare with traditional methods to predict acetabular dysplasia.
Accurate measurements of the acetabulum volume and surface areas of the overlapping regions of the acetabulum and femoral head could aid physicians in the diagnosis of acetabular dysplasia. Although, Wiberg’s center-edge angle and Sharp’s acetabular angle are widely used by physicians in the diagnosis of acetabular dysplasia, the method presented here could provide additional information for borderline patients. This method could also to be beneficial to surgeons in planning and preparing for corrective hip joint surgeries.
CONCLUSIONS
A method for measuring acetabulum volume and surface areas of the overlapping regions of the femoral head and acetabulum is presented. This method utilizes pelvic CT images to create 3D models of the acetabulum and femoral head. Acetabulum volume is measured using CT images to create a model of the enclosed volume of the acetabulum. From these models, the overlapping regions on the acetabulum and femoral head are isolated using a post-processing software program and the surface areas of the femoral head covered by the acetabulum and the acetabular area covering the femoral head can be measured. This method was used to collect surface area and volume measurements for 70 hips of 35 normal individuals. Results showed positive linear correlations between acetabular volume and the two surface area measurements taken from the overlapping regions on the femoral head and acetabulum.
ACKNOWLEDGMENTS
We would like to acknowledge Dr. John Frino of the Department of Orthopaedic Surgery for his collaboration on this study. Special thanks to Lindsey Taylor and Scott Gayzik for their assistance in collecting data for this study.
REFERENCES
[1] AAOS, “Osteoarthritis of the Hip: A Compendium of Evidence-based Information and Resources,” 2003.[2] D. S. Garbuz, B. A. Masri, F. Haddad, and C. P. Duncan, “Clinical and radiographic assessment of the young adult with symptomatic hip dysplasia,” Clin Orthop Relat Res, pp. 18-22, Jan 2004.
[3] M. Reijman, J. M. Hazes, H. A. Pols, B. W. Koes, and S. M. Bierma-Zeinstra, “Acetabular dysplasia predicts incident osteoarthritis of the hip: the Rotterdam study,” Arthritis Rheum, vol. 52, pp. 787-93, Mar 2005.
[4] G. Wiberg, “The anatomy and roentgenographic appearance of a normal hip joint,” Acta Chir Scand, vol. 83, pp. 7- 38, 1939.
[5] I. K. Sharp, “Acetabular dysplasia. The acetabular angle.,” J Bone Joint Surg Br, vol. 43, pp. 268-272, 1961.
[6] S. Jacobsen, S. Sonne-Holm, B. Lund, K. Soballe, T. Kiaer, H. Rovsing, and H. Monrad, “Pelvic orientation and assessment of hip dysplasia in adults,” Acta Orthop Scand, vol. 75, pp. 721-9, Dec 2004.
[7] A. Kojima, T. Nakagawa, and A. Tohkura, “Simulation of acetabular coverage of femoral head using anteroposterior pelvic radiographs,” Arch Orthop Trauma Surg, vol. 117, pp. 330-6, 1998.
[8] N. Konishi and T. Mieno, “Determination of acetabular coverage of the femoral head with use of a single anteroposterior radiograph. A new computerized technique,” J Bone Joint Surg Am, vol. 75, pp. 1318-33, Sep 1993.
[9] G. Zheng, M. Tannast, C. Anderegg, K. A. Siebenrock, and F. Langlotz, “Hip2Norm: an object-oriented crossplatform program for 3D analysis of hip joint morphology using 2D pelvic radiographs,” Comput Methods Programs Biomed, vol. 87, pp. 36-45, Jul 2007.
[10] M. Tannast, G. Zheng, C. Anderegg, K. Burckhardt, F. Langlotz, R. Ganz, and K. A. Siebenrock, “Tilt and rotation correction of acetabular version on pelvic radiographs,” Clin Orthop Relat Res, vol. 438, pp. 182-90, Sep 2005.
Presented at Rocky Mountain Bioengineering Symposium &
International ISA Biomedical Sciences Instrumentation Symposium
17-19 April 2009, Milwaukee, Wisconsin, www.isa.org
For reprints and permission queries, please visit SAGE’s Web site at http://www.sagepub.com/journalsPermissions.nav.
For additional information on ACL knee injuries, or to learn more about what is involved during ACL reconstruction surgery, please contact the office of orthopedic knee surgeon, Dr. Adam Anz, serving the greater Pensacola, Gulf Breeze, and Gulf Coast communities.
