| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Musculoskeletal Imaging |
1 From the Department of Radiology (C.F.B., A.G.B., K.B., B.L.D., C.L.N., R.J.H.) and the School of Medicine (D.K.H.), Stanford University Medical Center, Rm S-056, 300 Pasteur Dr, Stanford, CA 94305; and the GE Corporate Research and Development Center, Schenectady, NY (R.D.D., C.L.D.). Received July 7, 1998; revision requested September 11; final revision received, October 27; accepted March 29, 1999. C.F.B. supported in part by a 1997 RSNA Scholar Award. A.G.B. supported in part by a 1997 Toshiba America/RSNA (1) Seed Grant. Address reprint requests to C.F.B.
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: With an open-configuration, 0.5-T MR imaging system and active image-plane tracking, 10 shoulders were studied in five asymptomatic subjects to establish normal patterns of glenohumeral motion during abduction and adduction and internal and external rotation. Preliminary studies of physical examination during MR imaging, in which a physician examiner applied mechanical force to the humeral head, were also performed.
RESULTS: During abduction and adduction and internal and external rotation maneuvers with active subject muscle contraction, the humeral head remained precisely centered on the glenoid fossa in all asymptomatic subjects, which is in agreement with findings of previous radiographic studies. Application of force to the humeral head by an examiner was associated with as much as 6 mm of anterior translation and 13 mm of posterior translation.
CONCLUSION: Dynamic MR imaging of the glenohumeral joint is possible over a wide range of physiologic motion in vertically open systems. Use of an MR tracking coil enabled accurate tracking of the anatomy of interest. These preliminary measurements of normal glenohumeral motion patterns begin to establish normal ranges of motion and constitute a necessary first step in characterizing pathologic motion in patients with common clinical problems such as instability and impingement.
Index terms: Magnetic resonance (MR), motion studies, 414.121412 • Shoulder, MR, 414.1214
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Subjects were evaluated in an upright, seated position with a flexible transmit-receive, circular large-crown radio-frequency (RF) coil (GE Medical Systems) placed around the shoulder. A fast gradient-echo sequence was used: 19.8/7.2 (repetition time msec/echo time msec); flip angle, 30°–40°; 256 x 128 matrix; field of view, 24–30 cm; section thickness, 7 mm; one signal acquired. Sequential single-plane images were acquired at a rate of approximately 2.5 seconds per image. Total examination time was 10–20 minutes per shoulder. The imaging parameters as listed were chosen to provide the fastest possible acquisition of diagnostic-quality images and were based on findings of pilot studies in asymptomatic volunteers. In this context, diagnostic-quality images were defined as images that depicted the anatomy of interest and demonstrated sufficient contrast to allow reproducible measurements of spatial relationships between relevant osseous structures. Such images could not be substantially degraded by motion or other MR artifacts.
In preliminary imaging of the shoulder without the use of a device for maintaining a consistent image plane, we found it extremely difficult to maintain structures of interest in the image. We therefore applied the method of active image-plane tracking (3) to the shoulder. This practice placed additional constraints on the choice of imaging parameters, as there are limited options for manipulating imaging parameters in the software coupled to the active image-plane device.
Imaging planes were determined by two factors: (a) interactive subject positioning in the magnet by an examiner and (b) active image-plane registration (MR tracking) (3). These factors are closely related because, in its initial implementation, MR tracking was limited to image planes either parallel or perpendicular to the constant magnetic induction field, B0; images could not be prescribed in an oblique plane.
On the basis of the constrained choices among three orthogonal imaging planes, imaging-plane obliqueness to the subject's shoulder was instead achieved by rotating or shifting the subject into an oblique orientation relative to the magnet. Careful positioning was necessary to ensure that the humeral head and glenoid would both be captured on the same image, so spatial relationships could be measured. For abduction, an imaging plane oblique-coronal to the body and parallel to the supraspinatus tendon and scapula was chosen, which is analogous to choosing the plane used for oblique coronal imaging in static MR. For evaluation of internal and external rotation, an axial image plane transverse to the shoulder was used. To obtain this plane, it was necessary to try to maintain both shoulders relatively parallel to the floor, as excessive lateral leaning of the subject created oblique axial sections that frequently did not include both the humeral head and glenoid. Transaxial images were also used for evaluations of applied glenohumeral force.
MR tracking uses a specialized hardware and software configuration that includes a miniature RF coil for spatial localization of the imaging plane (3,4). In our initial implementation, a 7-mm-diameter loop of copper wire surrounded a 2-mL plastic cuvet containing dilute gadopentetate dimeglumine (Magnevist; Berlex Laboratories, Wayne, NJ) (1:200 ratio with water). The system uses a four-excitation Hadamard pulse sequence to calculate the position of the protons within the sensitive volume of the small RF coil. When run in an interleaved sequence alternating with an image acquisition, MR tracking interactively provides control over the position of the image section.
In the shoulder studies, the same examiner (C.F.B., D.K.H.) that positioned the subject in the imager adjusted the position of the MR tracking coil to capture the anatomy of interest, then taped the coil to the subject. Direct visual feedback was available via image display on LCD monitors inside the imager. The MR tracking coil was placed on the skin overlying the anterolateral aspect of the acromion for abduction and adduction maneuvers and anterior to the coracoid process for the internal and external rotation and stress maneuvers.
We adapted measurement methods initially developed for projection radiographic studies (5,6) for quantifying the relationship of the humeral head to the glenoid fossa. Initially, measurements on five shoulders were made with the imager operating console by using standard geometric constructions and measurement tools. Because this process was tedious and because we wanted an offline system for analysis, subsequent measurements were made with macros written for NIH IMAGE software (National Institutes of Health, Bethesda, Md) operating on Macintosh desktop computers (Apple Computers, Cupertino, Calif). Most of the measurements reported were performed by one author (D.K.H.), with some completed by another (C.F.B.). No systematic studies of intra- or interobserver variability in the measurements have been done.
The humeral head center was determined by prescribing a circle on the humeral articular surface. The center of the circle was then defined as the center of the head. This method has been validated in earlier studies and relies on the uniformity of the radius of curvature or the relative sphericity of the humeral articular surface (6). The margins of the glenoid were used to construct a line along and parallel to the glenoid fossa; the geometric center of this line determined the center of the glenoid. To do this, care was taken to not include the glenoid labrum or portions of the joint capsule in the line.
Humeral head centering on the glenoid was determined by constructing a perpendicular line from the humeral head center to the glenoid line. By measuring the position of intersection of this perpendicular line with the glenoid line, a linear position of the humeral head along the glenoid was established for any shoulder position. During imaging, we used active image-plane tracking to maintain the midglenoid in the image plane. Findings of preliminary studies showed that the humeral head remained centered on the glenoid whether measured in axial or oblique coronal sections to the glenoid. Given this and the symmetry of the glenoid, we did not observe systematic changes in measurements of humeral position when images were not obtained precisely through the glenoid center.
Another parameter needed to characterize the glenohumeral motion pattern was the position of the long axis of the humerus relative to the subject's body. For abduction and adduction, imaging started with the subject in the position of maximum adduction; this point was defined as 0° abduction. During normal abduction, overall humeral elevation occurs as a result of two coupled mechanisms—the glenohumeral and scapulothoracic components of abduction (6). The glenohumeral component is a result of humeral rotation on the glenoid. The scapulothoracic component is a result of scapular rotation on the thorax. Initially, abduction is primarily a result of glenohumeral motion, with scapulothoracic motion composing the latter stages.
In MR imaging in the coronal oblique plane, glenohumeral motion primarily changes the angle of the humeral shaft with respect to an external reference (as long as the subject's body maintains a similar position), and scapulothoracic motion primarily changes the angle of the glenoid face relative to the external reference. We measured each of these components separately; however, for this report, their combined effect, or the net change in humeral angle relative to baseline, was recorded for each increment of motion.
For internal and external rotation maneuvers, imaging began with the patient's shoulder in a position of maximum internal rotation. Degree of humeral rotation was determined by measuring the angle between the center of the humeral head and the bicipital groove. By using a convention established by Davis and colleagues (7), we defined 0° of rotation as the point where a line connecting the humeral head center and the bicipital groove was parallel to the glenoid face.
Measurements of glenohumeral translation with stress applied to the humeral head were analogous to those of internal and external rotation motions performed by the patient. The humerus was kept in a position of neutral (0°) rotation and adduction during these maneuvers.
In an attempt to apply reproducible stress to each of the subjects, the maneuvers were performed by a single examiner (C.F.B.), and subjects were asked to relax their muscles during the test. Despite this practice, it was difficult to ensure that the same amount of stress was applied to each subject; the results should be viewed with this potential lack of reproducibility in mind.
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
|
|
|
For the 10 shoulders studied, the mean number of images per shoulder analyzed during abduction was 10.9 ± 3.1 (SD), with a range of seven to 17 images; during adduction, the mean was 9.1 images ± 2.2, with a range of six to 11 images.
By plotting humeral head position on the glenoid as a function of the total abduction angle (including both glenohumeral and scapulothoracic components), one obtains graphs such as those in Figure 3a for abduction and Figure 3b for adduction. Notice that, over the full range of motion, the humeral head remained centered on the glenoid, deviating from the glenoid center point, on average, less than 0.3 cm over the entire motion. Minor fluctuations in humeral head centering were within the range of the SD of the measurements, as shown in the graphs. For the 10 shoulders, the mean amount of maximal abduction was 119° ± 19, with a range of 95°–155°.
|
|
|
|
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
We hypothesize that MR imaging of the shoulder during active joint motion, in which normal structures are engaged in their intended functions to restrain glenohumeral translation, will provide relatively direct insight into glenohumeral biomechanics. While ultimately the goal is to better understand the relationship between specific anatomic lesions and pathologic motion patterns, normal baselines with which to compare abnormal findings must be established.
The results of this study show that a vertically open MR system was capable of allowing dynamic joint evaluation over a wide range of physiologic motion. We have established imaging protocols and methods for quantitative assessment of glenohumeral relationships. In addition, we have begun to practice a form of joint assessment—physical examination during MR imaging.
In studies of abduction and adduction (Figs 1–3), the humeral head remained precisely centered in the glenoid fossa over a wide range of motion. These results concur with those of radiographic studies, although little directly comparable information is available (5,6). Investigators in reported series utilized incremental motion and static radiographs, which may not be appropriate comparisons to the slow but continuous motion we employed.
An additional facet of shoulder motion that can eventually be derived from dynamic MR studies is the relative contributions of the glenohumeral and scapulothoracic components of abduction. This report combines these two components into a net abduction. Separating the components, particularly with regard to differences between subjects and those with altered shoulder mechanics such as large rotator cuff tears (10), will be of interest in the future.
In studies of internal and external rotation of the humerus, the humeral head remained precisely centered in the glenoid over the full range of physiologic motion (Figs 4, 5). This result is similar to that of a static radiographic study by Howell and colleagues (5) of glenohumeral motion in the axial plane.
Use of conventional narrow-bore MR systems in so-called kinematic MR imaging of the shoulder has been reported for asymptomatic subjects. The kinematic technique refers to simulation of dynamic motion by playing multiple static images in a cine loop. These studies may not be directly comparable with ours, because in kinematic studies the muscle groups are typically relaxed. Sans and colleagues (11) performed kinematic MR imaging on 39 shoulders during internal and external rotation and characterized the shape and signal intensity of the labrum and labral position. Imaging time was 32 seconds per position. No quantitative measures of humeral head position on the glenoid were reported. Similar qualitative features were reported by Cardinal and colleagues (12) and by Bonutti and colleagues (13), but again, no quantitation of glenohumeral position was reported.
Whereas previous studies have emphasized morphologic assessments, our work focuses on quantitation of motion patterns during slow but continuous motion initiated and maintained by the subject without the use of a positioning device or restraint. Ideally, further studies will allow integration of morphologic, signal intensity, and quantitative observations together and lead to a comprehensive assessment. Along these lines, it is worth pointing out that a full understanding of dynamic joint motion may require image display and review in a dynamic fashion to appreciate subtle motion features.
We report on dynamic MR imaging of the glenohumeral joint during actual physical manipulation by a physician examiner. This is an exciting aspect of open MR imaging that may have broad potential applications. In principle, the addition of objective anatomic and dynamic information to physical examination maneuvers can provide insights into clinical signs and maneuvers whose exact anatomic basis has only been inferred.
Initial attempts at application of anteroposterior or posteroanterior force to the humeral head to evaluate translation were encouraging, although the maneuvers were initially challenging to perform. As our experience has increased, physical examination and manipulation during MR imaging appears to be routinely feasible. Unquestionably, a learning process is involved, much like learning physical examination or patient-interactive imaging such as ultrasonography.
It is important to point out that it is unknown whether application of force to a relatively relaxed shoulder provides clinically relevant information, as individuals may exhibit joint laxity but not suffer from the clinical problem of instability (14). Further studies on the reproducibility of stress testing in asymptomatic and symptomatic subjects, as well as between examining physicians, will be necessary, as the large SDs in Figure 8 most likely reflect inherent variability in the maneuvers.
Key advantages of open MR imaging with rapid imaging time include the ability for patients to undergo a wide range of active joint motion and physical examination during MR imaging. With fast imaging, one pays a price of diminished signal-to-noise ratio and spatial resolution relative to longer-duration images in higher field strength magnets. In addition, the image contrast is currently limited, which produces images with T1 or spin-density weighting in the very short repetition time and echo time regime.
Advances in pulse programming, such as the introduction of preparation pulses, may help overcome current imaging contrast limitations. Another notable technical development that enabled the performance of the current study is the use of active image-plane MR tracking (3,4). This permits placement of an MR tracking coil on the patient's skin, with spatial localization of the coil by RF pulses used to define the position of an axial, sagittal, or coronal plane.
At the time the current study was performed, the software coupled to this tracking device was limited in that only imaging planes parallel or perpendicular to the constant magnetic induction field, B0, of the magnet could be obtained. More recent versions of the tracking software enable use of oblique imaging and multiple tracking devices, which will aid in patient positioning and selection of imaging axes (Dumoulin CL, personal communication, 1999). The interactive MR examination is currently a demanding undertaking in which an examiner actively positions the patient to obtain a desired image plane relative to the body and manipulates the MR tracking device to select a specific level of interest. Also, transferring the images to a separate computer and performing the measurements is tedious and time-consuming. With further experience, it is hoped that a comprehensive examination can be performed in a reasonable amount of time.
Our study had a number of limitations. First, the results reported were for a small number of subjects in a young age group and included both female and male volunteers. Age, sex, and dominance of the extremity may influence glenohumeral motion patterns and need to be studied further.
Second, the motions studied were confined to single-plane motions that do not entirely mimic everyday shoulder motion. Measurements were also confined to the planes of motion, such that for abduction and adduction we measured the superoinferior centering of the humerus on the glenoid but did not test for possible shifts in the anteroposterior plane.
Third, while we were able to obtain images during continuous movement, the motion was slow relative to the dynamic and ballistic movements that may give rise to pain in the shoulder in patients, particularly in athletes (15). As always, faster imaging is desirable, as long as crucial anatomic and signal intensity information is not sacrificed. Along these lines, preliminary results with "fluoroscopic" MR sequences are encouraging (16).
Finally, the measurements reported have not been evaluated for intra- or interobserver variability. We plan to continue to characterize normal motion patterns and validate the measurement methods described in this preliminary work.
In conclusion, our preliminary experience with dynamic, physiologic MR imaging during glenohumeral motion is very encouraging. Further experience with asymptomatic and symptomatic subjects will be necessary to prove that the techniques are clinically relevant. Technologic advances in the form of more flexible MR tracking devices and optimized MR pulse sequences will be helpful. We are optimistic that open MR imaging can provide useful insights into not only normal joint function but also altered mechanics in patients with instability. As we begin to examine symptomatic individuals, having a clear idea of motion patterns in asymptomatic subjects will be a requisite.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Author contributions: Guarantor of integrity of entire study, C.F.B.; study concepts, C.F.B., A.G.B., D.K.H., R.J.H.; study design, C.F.B., A.G.B., D.K.H.; definition of intellectual content, C.F.B., A.G.B., D.K.H.; literature research, C.F.B.; clinical studies, C.F.B., D.K.H., B.L.D., K.B., C.L.N.; data acquisition, C.F.B., D.K.H., B.L.D., K.B., C.L.N.; data analysis, C.F.B., D.K.H.; manuscript preparation, C.F.B., D.K.H.; manuscript editing, A.G.B., K.B., B.L.D., C.L.N., R.D.D., C.L.D.; manuscript review, all authors.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  G. E. Gold, G. P. Pappas, S. S. Blemker, S. T. Whalen, G. Campbell, T. A. McAdams, and C. F. Beaulieu Abduction and External Rotation in Shoulder Impingement: An Open MR Study on Healthy Volunteers Initial Experience Radiology, September 1, 2007; 244(3): 815 - 822. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. J. Bureau, M. Beauchamp, E. Cardinal, and P. Brassard Dynamic sonography evaluation of shoulder impingement syndrome. Am. J. Roentgenol., July 1, 2006; 187(1): 216 - 220. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. C. Schiffern, R. Rozencwaig, J. Antoniou, M. L. Richardson, and F. A. Matsen III Anteroposterior Centering of the Humeral Head on the Glenoid In Vivo Am. J. Sports Med., May 1, 2002; 30(3): 382 - 387. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
