Introduction

Femoroacetabular impingement (FAI), a common cause of hip pain in adolescents, is characterized by pathological abutment of the femoral head-neck junction with the acetabulum. FAI primarily presents as two types of deformities, cam-type [1, 2] and pincer-type, which can occur alone or co-exist in the same hip [3]. Each causes specific patterns of labral and cartilage injury [3].

Imaging is vital to diagnosis and surgical decision making in FAI. Plain radiographs and computed tomography help define bony anatomy, while magnetic resonance imaging (MRI) helps identify intra-articular cartilage and labral injuries [4]. However, the oblique orientation and curved articular surfaces of the hip joint make imaging via magnetic resonance (MR) challenging and may result in partial volume averaging or the lack of contrast between adjacent tissues [5, 6]. Furthermore, articular cartilage is relatively thin, adding to the difficulty in accurately diagnosing cartilage damage [5].

High-resolution MRI, 3-dimensional (3D) MRI, or a combination of both helps improve hip joint imaging protocols. High-resolution (HR), an ever-changing term, indicates decreased pixel or voxel size to improve an image’s spatial resolution, currently on the order of 0.5 mm per pixel for 2-dimensional (2D) MRI [7]. Especially beneficial for multiplanar reformatted (MPR) images, 3D MRI utilizes isotropic voxels, or voxels with the same dimensions in the x, y, and z planes, allowing resultant images to be reconstructed in any plane [8]. Radial images, reformatted from 3D acquisitions to obtain images at prescribed angular intervals (i.e., 10 to 20°) perpendicular to the femoral neck axis, are used to produce a set of images over the circumference of the femoral head-neck junction and to evaluate the acetabulum and labrum [5, 9,10,11].

MRI sequences identifying pathology in the labrum, transition zone cartilage, and true acetabular cartilage must be accurate and timely. For this study, we developed an FAI Hip MRI protocol that includes the following sequences: (1) 2D axial proton density (PD) imaging with individual axial, sagittal, and coronal plane acquisitions (2D); (2) 3D PD imaging with multiplanar reformats (3D MPR); and (3) radially reformatted (RR) images post-processed from 3D PD acquisitions (Table 1). Our study aims to compare these three MR acquisitions in their ability to evaluate the labrum, chondrolabral junction, and articular cartilage of the hip joint in adolescents with FAI. We hypothesized RR PD imaging would have the highest accuracy in detecting labral tears and 2D PD imaging would have the highest accuracy in detecting cartilage injury.

Table 1 MRI protocol parameters
Full size table

Materials and methods

The study protocol was approved by the Colorado Multiple Institutional Review Board. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Thirty-three consecutive cases of FAI, meeting clinical criteria, and having undergone pre-operative MRI followed by operative management were included in the study. All included patients had undergone pre-operative hip MR with 2D HR PD coronal, sagittal, and axial sequences; 3D PD sequence with multiplanar reformats (MPR) in the coronal, sagittal, and axial planes; and radially reformatted (RR) images constructed from the 3D PD data set and acquired on a 3 Tesla (T) magnet. All studies were completed on a Philips MRI machine. Patients with inadequate surgical notes or hip MRI not including the specific sequences mentioned above were excluded.

For each included hip MRI, the sequences were grouped as follows: (1) 2D HR PD images independently acquired in the coronal, sagittal, and axial planes (named 2D); (2) 3D PD images acquired in the coronal plane with MPR’s in the coronal, sagittal, and axial planes (named 3D MPR); and (3) PD RR images created from the 3D PD data set (named RR) (with the sagittal MPR images used as a reference to help determine the direction of rotation) (Table 1). Each of the three groupings of MR sequences was then evaluated for the following: injury to the (1) labrum, (2) transition zone cartilage, and (3) true acetabular cartilage. The transition zone cartilage was defined as the hyaline cartilage located less than 5 mm away from the labrum, whereas the true acetabular cartilage included all deeper articular cartilage [1, 2].

The grouped 2D, 3D MPR, and RR MRI images for each MRI were independently reviewed, in random order to avoid bias, by two board-certified musculoskeletal radiologists with 5 and 7 years of experience. Readers were blinded to the results of arthroscopy at the time of review. The Beck classification for labral and transition zone cartilage injury in FAI was utilized with scoring from 0 to 4 (Table 2) [12]. The true acetabular cartilage was graded based on the Outerbridge classification with scoring from 0 to 4 (Table 3) [13]. The location and percentage of hip joint affected by the labral, transition zone, and true acetabular cartilage injury, when present, was also recorded. The location of the injury was assessed from 11 to 3 o’clock sectors based on the clock-face method of visualizing the acetabulum (Fig. 1) [14,15,16]. Percentage of hip joint affected was defined as the total number of consecutive clock levels affected by the injury which was converted into a percentage based on the following equation: [number of consecutive affected levels/12].

Table 2 Beck classification system for labral and transition zone cartilage injury
Full size table
Table 3 Outerbridge classification system for acetabular cartilage injury
Full size table
Fig. 1
figure 1

Orientation of clock face on acetabular rim used for location of labral, transition zone cartilage defects, and acetabular cartilage defects

Full size image

Intra-operative images and operative reports for all patients were then reviewed by a single orthopedic surgeon for the purpose of defining the presence of labral, transition zone, and true acetabular cartilage pathology.

Statistical methods

Using the intra-operative visualization as the gold-standard, the sensitivity, specificity, and accuracy of the three separate methods of MRI sequences in identifying the presence of labral, transition zone, and true acetabular cartilage injuries were assessed from the 11 o’clock to 3 o’clock positions. This region of the hip was selected as it was associated with highest prevalence of intra-articular pathology noted at arthroscopy (Table 4). Bland-Altman methods were used to estimate agreement between percentage of the hip affected by the injury as measured by the MRI methods relative to intra-operative visualization. The bias or the mean difference between the MRI-based method and arthroscopic visualization, as well as the limits of agreement, was calculated for each of the three methods.

Table 4 Prevalence of intra-articular injuries by position, as noted at arthroscopy
Full size table

Results

Thirty three hips (13 female, 20 male) were included in the sample. Twenty-six of the hips were diagnosed with cam-type impingement (1 pincer-type, 6 mixed-type) and the average age at the time of surgery was 16.6 years (range: 12.7–24 years). On average, 17.9 weeks elapsed between pre-operative imaging and operative management (range: 0.714–86.0 weeks). Labral and transition zone injuries were most frequently identified at the 2 o’clock position. True acetabular cartilage injuries were most frequently identified at the 12 o’clock position (Table 4). The classification of the severity of pathology and the proportion of hip affected by the injuries are described in Tables 5 and 6. Figure 2 shows intra-operative findings of a patient included in this study with a labral tear and transition zone cartilage injury (Fig. 2a, b) and the corresponding MRI in radial reformatted image (Fig. 2c) and MPR images (Fig. 2d, e, f). Figure 3 shows the intra-operative (Fig. 3) and MRI findings of another patient included in the study with a transition zone cartilage injury with the corresponding radially reformatted image (Fig. 3b), MPR image (Fig. 3c), and HR image (Fig. 3d). On average, an imaging study took 32 min and 21 s to accomplish.

Table 5 Classification of pathology in 11–3 o’clock region-based on intra-operative visualization
Full size table
Table 6 Size of injury/percentage of hip joint based on intra-operative visualization
Full size table
Fig. 2
figure 2

a Intra-operative image of a 16-year-old male’s left hip with Beck grade 3 (flap) transition zone chondral injury from 1 to 3 o’clock (seen inferior to probe) as part of Beck grade 3 (detached at chondrolabral junction) labral tear (seen superior to probe). b Intra-operative image of a 16-year-old male’s left hip with Beck grade 2 (debonding/carpet or wave sign) transition zone chondral injury from 11 to 1 o’clock (seen inferior to the probe) as part of a Beck grade 3 (detached at chondrolabral junction) labral tear (seen superior to probe). c Corresponding radially reformatted PD image obtained from the 3D acquisition showing a Beck grade 2 (debonding/carpet or wave sign) transition zone chondral injury (red arrow) as part of a Beck grade 3 (detached at chondrolabral junction) labral tear (blue arrow) in a 16-year-old male. d Corresponding sagittal MPR MR imaging of a 16-year-old male with a Beck grade 2 (debonding/carpet or wave sign) transition zone cartilage injury (red arrow). e Corresponding coronal MPR MR imaging of a 16-year-old male with a Beck grade 2 (debonding/carpet or wave sign) transition zone chondral injury (red arrow) as part of a Beck grade 3 (detached at chondrolabral junction) labral tear (blue arrow). f Corresponding sagittal HR MR imaging of a 16-year-old male with a Beck grade 2 (debonding/carpet or wave sign) transition zone cartilage injury (red arrow)

Full size image
Fig. 3
figure 3

a Intra-operative image of a right hip with Beck grade 3 (flap) transition zone chondral injury from 12 to 2 o’clock (seen inferior to the probe). b Corresponding radially reformatted PD image of a 23-year-old male obtained from the 3D acquisition showing Beck grade 3 (flap) transition zone chondral injury (red arrow). c Corresponding coronal MPR MR imaging of a 23-year-old male with a Beck grade 3 (flap) transition zone cartilage injury (red arrow). d Corresponding coronal HR MR imaging of a 23-year-old male with a Beck grade 3 (flap) transition zone cartilage injury (red)

Full size image

Labral injuries

For labral injuries, the 3D MPR and RR MRI groupings were associated with highest sensitivity and accuracy in assessing the presence of an injury between the 11 and 3 o’clock positions (Table 7). The 3D MPR grouping demonstrated the lowest levels of bias (smallest mean difference between MRI evaluation and surgical evaluation) in assessing the size of the injury (Table 8).

Table 7 Accuracy in identifying intra-articular pathology
Full size table
Table 8 Agreement regarding injury size or percentage of hip joint affected by injury
Full size table

Transition zone injuries

The 2D MRI grouping was the most accurate in identifying the presence of transition zone injuries between the 11 and 3 o’clock positions (Table 7). The RR grouping demonstrated lower levels of bias when assessing the size of the transition zone injury relative to the other groupings (Table 8).

True acetabular cartilage injuries

The 3D MPR grouping was the most accurate MRI sequence in identifying the presence of acetabular cartilage injuries in the 11–3 o’clock positions (Table 7). The RR grouping was associated with a lower level of bias when assessing the size of the acetabular cartilage injury relative to the other groupings (Table 8).

Discussion

FAI, a morphological abnormality of the acetabulum and proximal femur, can cause hip pain and dysfunction while being linked to early osteoarthritis in otherwise healthy hips [3, 12, 17]. FAI often produces the most severe damage to the acetabular labrum and cartilage in the anterosuperior region [3, 6, 8].

Imaging assessment of bony morphology and cartilaginous pathology provides important information for orthopedic surgeons. Increased accuracy of diagnostic imaging improves clinical decision making and focuses surgical planning. MR imaging is the most common modality for evaluation of the labrum, transition zone cartilage, and true acetabular cartilage. The ideal MRI sequence would be both accurate and efficient. Therefore, we sought to determine the best sequences for evaluation of the soft tissue structures in the hip.

3D MPR and RR groupings were superior to the 2D grouping in assessing labral tears. The 3D MPR grouping had the lowest bias in assessing the size of the labral tears. Previously related studies have evaluated the accuracy of 3 T MRI and 3 T MR arthrogram to assess labral tears relative to direct visualization during arthroscopy [18,19,20,21,22]. Sensitivity and specificity estimates for 3 T MRI ranged between 61 and 100% and 50–100%, respectively, with one study reporting 98% accuracy [1, 18,19,20,21,22]. Two studies using 3 T arthrogram techniques have reported 90–95% sensitivity and 0% specificity in one study and 84% specificity in the other. Only one of these studies using 3 T machines included radially reformatted sequences in their image evaluation, while the others utilized only coronal, sagittal, and axial 2D sequences. Therefore, these studies have focused on the utility of 3 T MR and 3 T MR arthrography for the detection of labral tears, but none has included 3D MR imaging, and none has directly compared 3D MR and radially reformat sequences to one another. Our results for 3D MPR and RR sequences performed similarly with 98.4% sensitivity, 0% specificity, and 95.5% accuracy.

The 3D MPR as well as RR groupings was acquired in the same fashion. They may be more sensitive and accurate in assessing the labrum than other structures because the labrum is a low-signal intensity structure with less water content than the adjacent transition zone and acetabular cartilage resulting in an improved contrast resolution. Additionally, the 3D MPR sequence has a higher signal to noise ratio (SNR) than 2D which is attributed to both the slightly larger voxel size as well as the greater number of measurements intrinsic to this type of acquisition, a factor of the number of phase encoding steps in the slice direction.

For the transition zone cartilage, the 2D grouping was the best in assessing injury while the RR grouping showed the lowest agreement bias; however, our results showed that all three groupings performed relatively poorly. Only one other study has evaluated 3 T MR for transition zone cartilage injury; however, this study included labral and cartilage tears at the transition zone, making this difficult to compare with our report of chondral lesions at the transition zone only [18]. In this study, they found 56% agreement between 3 T MR and arthroscopy, similar to our results with 66.7% sensitivity, 85.2% specificity, and 81.8% accuracy with 2D imaging, which was the most accurate of the three sequences in our study. There were no studies utilizing 3 T arthrogram to diagnose transition zone cartilage injuries to compare in the literature.

The 3D MPR grouping demonstrated slight superiority to the other two groupings for identifying true acetabular injuries; however, all three groups performed relatively poorly. Three previous studies have utilized standard 2D sequences with 3 T MR for evaluation of the acetabular cartilage. Results ranged from 59 to 94% sensitivity and 67–100% specificity and one reporting 56% agreement [18,19,20]. Although not statistically evaluated, one of these studies found increased sensitivity in imaging after including a radial cut along with sagittal, coronal, and axial formats in their evaluation, which they attributed to the ability of a radially reformatted sequence to obtain more in-plane images of the curved acetabular surface [19]. One study has also compared 3 T arthrography with arthroscopy for evaluation of the acetabular cartilage, with 71–81% sensitivity and 82–91% specificity between two readers [20]. Our results were somewhat less accurate compared with these previous reports using both 3 T MR and 3 T MR arthrography (42.6% sensitivity, 66.7% specificity, 47% accuracy). The decreased level of accuracy in our study is likely attributable to differences in our assessment method. We considered both the injury as well as the location of the injury. Thus, in our study, a high level of accuracy depends on both the orientation of the image orientation (clock face is oriented correctly) as well as the identification of the injury. We believe this approach is more clinically relevant as surgical planning depends on an accurate assessment of both the location and intra-articular injuries.

We anticipated more favorable results with the 2D sequences when compared with the 3D MPR and RR groupings in the evaluation of articular cartilage damage given the higher in-plane resolution with 2D acquisitions than with the 3D acquisitions. In addition, the heavier T1 weighting associated with the 2D sequence, due to a significantly shorter repetition time (TR), in general allows for better evaluation of high-water content structures such as articular cartilage. However, as our results indicate, this hypothesis was not completely validated. One possible explanation for this is that the cartilage injuries in this relatively young cohort were few and not very severe (of the 6 cartilage injuries, only 2 were > 1.5 cm). The sensitivity of MRI in detecting these minor abnormalities was not high in our current sequencing parameters. Therefore, in this population compared with an older population, this low sensitivity may be more of a concern.

Our study included several additional limitations. Our sample was not random as each patient underwent a history and physical exam prior to imaging. Selection bias is possible as each patient met preliminary diagnostic criteria for FAI and thus was more likely to have cartilage or labral damage. There was also intrinsic selection bias as the orthopedic surgeons were not blinded to pre-operative imaging before choosing to proceed with operative intervention; their physical examination as well as evaluation of imaging most likely dictated their decision to continue with surgery. Additionally, our study sample did not include a control group without labral or cartilage injury, as operative assessment would have been unethical. We also did not control for the time from MRI to arthroscopy, which ranged from less than a week to over a year. The abnormal labrum and/or cartilage may have undergone some change during this time, limiting the accuracy of MRI.

While the 3D PD sequence with multiplanar reformats as well as RR images performed well in the evaluation of injury to the labrum, all sequences were relatively poor at detecting injury to both transition zone and true acetabular cartilage. Future research and possible options for improvement include: (1) improve the spatial resolution of the 2D sequence by decreasing both the slice thickness and slice gap while keeping pixel size relatively constant; and 2) decrease the voxel size of 3D PD sequence to improve spatial resolution. The second option could obviate the need for a separate 2D acquisition altogether, replacing these sequences with multiplanar reformats as well as radially reformatted images constructed from the 3D PD, allowing for an overall shorter protocol.

Conclusions

Given that labral injury is a primary source of pain and dysfunction for patients with FAI, we sought to determine an efficient and accurate MR imaging sequence for evaluating the morphology of the labrum, transition zone cartilage, and true acetabular cartilage. Including both 3D MPR and RR MRI groupings is favorable for accurate visualization of the hip joint and may help guide diagnosis and treatment planning.