Abstract
Objectives:
Cam-type femoroacetabular impingement (FAI) occurs due to the presence of cam morphology that leads to a lack of femoral head-neck offset, causing abnormal contact at the femoral head and acetabulum. FAI clinically presents as symptoms including limited range of motion at the hip, moderate to severe pain, locking or catching during ambulation, and lower back pain. Over the past decade, statistical shape modeling (SSM) has been increasingly used to evaluate anatomical variations of cam morphology in FAI patients. SSM is a population-based technique that generates particle models from 2D or 3D geometries, which can be used to quantify shape variations and differences. Existing literature on SSM for cam-type FAI include a variety of studies that implement 2D or 3D imaging to better understand spatial distribution of the cam deformity, length of cam protrusion, and differences in mean shape between cam FAI patients vs controls. However, a large majority of studies utilize computed tomography (CT) scans, which incurs ionizing radiation to subjects. Recently, magnetic resonance imaging (MRI) has seen increased clinical use for diagnosis of cam morphology and labral tears associated with FAI. Therefore, our group is interested in developing an SSM framework from MRI-derived 3D models of the hip to assess cam morphology shape in our FAI patient cohort. Previously, our group compared the shape of the proximal femur in patients with cam-type FAI before and after hip arthroscopy, to quantitatively describe the resection area. However, our group has not yet assessed preoperative shape variations among our cohort of FAI patients. Therefore, our research objective for this study was to use SSM to determine and describe shape modes of the proximal femur in a cohort of cam-type FAI patients.
Methods:
For this institutional review board approved study, 66 patients with FAI were enrolled from the high-volume clinical practice of a single surgeon (SJN). Inclusion criteria included the follow: age 14-45, α-angle > 50°, and clinical diagnosis of cam-type FAI. Exclusion criteria included: Tönnis Grade > 1, history of developmental hip disorder or dysplasia, and prior surgery to contralateral hip. Pre-operative MRIs from each patient were segmented using Mimics software to generate .STL files. All 1.5T MRI scans were obtained by utilizing a 3D gradient dial-echo MRI sequence with two separate echo times, as previously described. Exported DICOMs of the pre-operative proximal femur MRI data were utilized to conduct segmentation of the bone (Mimics). Segmented .stl data were imported into ShapeWorks. Correspondence points (2,048 particles total) were used with a Procrustes scaling transformation to remove the effect of femur size from the model. The mean particle configuration across all included 3D models was then used to generate the “mean shape” of the preoperative femur. Principal component analysis was performed on the resulting particle models to determine shape modes of variation.
Results:
The patient cohort had the following demographic information, presented as mean ± standard deviation: age (29.83±11.53 years), BMI (23.53±4.19), and male/female (15/51). Nine shape modes were observed to explain 90.8% of the variance in shape across the 66 models, and are depicted in Figure 1. The largest variability in Mode 1 was seen circumferentially around the femoral head-neck junction. Mode 2 was shown to have variability at the head of the femur, particularly in the area encircling the fovea. Mode 3 showed variability in the section of the femoral head that articulates against the acetabulum. The largest variability in Mode 4 is observed in the trochanteric fossa. In Mode 5, some variability was observed at the super section of the greater trochanter, and in Mode 6, some variability was seen along the intertrochanteric crest. The largest variability in Mode 7 was observed within the posterior section of the femoral neck and at the superior section of the greater trochanter. The trochanteric fossa and posterior surface of the femur near the end of the intertrochanteric line demonstrated largest variability within Mode 8. The largest variability in Mode 9 was observed within the middle portion of the intertrochanteric crest.
Conclusions:
This study is among the first to implement MRI scans with a relatively large sample size (compared to that of literature) with SSM to analyze proximal femur shape in a cohort of patients with cam-type FAI. The results of this study demonstrate the feasibility of performing SSM for studying FAI using MRI scans. In future work, we plan to further expand our sample size, with the eventual goal of building a classifier of proximal femur shapes associated with cam-type FAI. In addition, the data gathered from this study will aid in advancing the goal of using MRI data to correlate shape models with PROs and motion parameters, which we also intend to conduct in future studies. By studying the shape variability of the proximal femur within pre-operative cam-type FAI patients, we are better equipped to describe the cam-deformity and determine potentially aberrant contact conditions. Furthermore, the results of this study may help physicians better classify cam-type FAI and provide clinicians better guidelines for patient-specific care, based on their hip anatomy.
