Normal Shoulder

Imaging Modalities and Technical Aspects

Conventional radiography of the shoulder ( Table 6-1 , Fig. 6-1 ) is recommended for any primary evaluation of suspected pathology, including fractures, dislocations, bone tumors, and infection. It is readily available, inexpensive, and uses minimal ionizing radiation. However, soft tissue evaluation is limited, and patient positioning may be difficult (e.g., due to pain, fracture, ankylosis).

Computed tomography ( Table 6-2 ; see also eTable 6-1 ) enables evaluation of complex fractures, dislocations, degree of healing (callus formation, partial union, nonunion, adequate reduction, joint congruity), the type of matrix calcification (osteoid, cartilage) in some bone tumors, poorly visualized or suspected bone abnormality on radiography, and assessment of soft tissue calcifications.

CT arthrography ( Fig. 6-2 ) is indicated for assessment of suspected labral tears when there is a contraindication to MRI. Intraarticular injection of iodine contrast material allows visualization of the internal capsular anatomy and pathology. A conventional single- or double-contrast arthrogram is performed (as described under MR arthrography), first using diluted iodinated contrast material (3 to 5 : 1) with or without air for a total volume of 13 to 15 mL, and the CT is performed without delay to avoid extravasation and dilution of contrast (and, thus, loss of capsular expansion). The patient should be positioned supine and the arm injected in a neutral position. The contralateral arm is elevated and flexed at the elbow to decrease shoulder girdle girth and allow the injected shoulder to be positioned within the gantry center. To avoid additional radiation exposure, it is not routine to scan in multiple patient positions There are, however, specific indications and/or benefits for doing so: external rotation improves visualization of both anterior and posterior labral tears, and prone positioning improves visualization of the posterior labrum. It should be noted that rotation of the arm alters the morphology of the capsulolabral complex, and the change in position of the glenohumeral ligament may mimic an anterior labral tear.

Conventional magnetic resonance imaging ( Tables 6-3 and 6-4 , and Fig. 6-3 ) allows for visualization and assessment of soft tissue anatomy and pathology of the rotator cuff, capsulolabral and neurovascular structures, and osseous pathology. It is multiplanar and nonionizing, but limited by expense, claustrophobia in closed magnets, long examination time, and patient and respiratory motion artifacts. Currently, suggested parameters utilize a 1.5-tesla magnet with a phase-array surface coil, slice thickness of 3 to 4 mm, 512 × 512 matrix, and a 14- to 16-cm field of view. Sections are obtained in three planes: (1) axial , from the superior aspect of the acromioclavicular joint through the proximal humeral shaft including the insertion of the pectoralis muscle; (2) oblique coronal , from the coracoid process to the infraspinatus muscle oriented along the longitudinal plane of the scapula; and (3) oblique sagittal , from the neck of the glenoid through the deltoid muscle oriented in a plane perpendicular to the oblique coronals (see eTable 6-2 ).

Indirect MR arthrography ( Table 6-5 ; Fig. 6-4 ) is indicated in suspected capsulolabral lesions. It takes advantage of bulk flow and diffusion of contrast material from the vascular supply into synovial tissue lining the bursae, joint capsule, and tendon sheaths to ultimately pool into the joint space (see also Video 6-1 to view a full series of MR arthrographs). Gadopentetate dimeglumine-based contrast agents shorten the T1 relaxation time of tissues, which can be used to produce arthrographic T1-weighted fat-suppressed images, enabling anatomic and physiologic assessment of joint pathology. Intra- and extra-articular soft tissues may be assessed. The synovial membrane is vascular, and injected contrast medium will diffuse into the joint over time. This property is advantageous when diagnosing inflammatory arthropathies such as rheumatoid arthritis that result in synovial hyperplasia. Synovial tissue, which is normally non-enhancing and intermediate in signal intensity, will enhance on post contrast images in such inflammatory conditions. This is a conspicuous finding on the less costly indirect MR arthrographic imaging (versus direct MR arthrography). It does not require fluoroscopic guidance or joint injection and is superior to conventional MRI in delineating the labrum when there is minimal joint fluid. Limitations include inability to control the volume of contrast diffusing into the joint; insignificant joint distention unless there is a preexisting effusion; and the fact that enhancement of subacromial bursa can obscure a rotator cuff tear—this potential pitfall may be avoided by comparing pre- and postcontrast images.

The procedure requires intravenous injection of a 15-mL solution of 0.1 mmol/kg gadopentetate dimeglumine. Greater concentrations of gadopentetate dimeglumine, including 0.2 and 0.4 mmol/kg, have not been shown to derive greater arthrographic benefit.

Oblique coronal fast spin-echo T2-weighted fat saturation is included for the identification of preexisting extra-articular fluid collections (bursitis) and to evaluate possible magic angle artifact in the distal rotator cuff and labrum (which creates the appearance of supraspinatus tendinitis, due to increased signal on sequences with short echo time in images of tissues with well-ordered collagen fibers in one direction, such as tendon or articular hyaline cartilage).

The first sequence recommended is imaging in the ABER (abducted externally rotated) position to ensure patient compliance with the entire exam. The palm of the hand is positioned against the dorsal aspect of the craniocervical junction, which allows for static assessment of stress placed on the anterior glenoid. Coronal scout images of the patient in the ABER position are performed, from which sagittal images are prescribed along the longitudinal axis of the humerus, resulting in oblique sagittal series of sections from the anteroinferior labrum through the posterior superior labrum.

Direct MR arthrography ( Table 6-6 ; Fig. 6-5 ) allows for distention of the joint capsule to demonstrate to best advantage the intracapsular structures (i.e., glenoid labrum, glenohumeral ligaments, articular surface of the rotator cuff). This is a two-phase procedure in which the intraarticular injection of contrast material is performed under fluoroscopic visualization, and then the patient is transferred to the MR scanner for diagnostic imaging. The technique of intraarticular injection must avoid the cartilage, labrum, and capsular attachments to yield diagnostic utility. Although multiple techniques have been described, the anterior approach is most common and has been modified over time.

The anterior approach to glenohumeral joint injection requires the patient to be positioned supine with the shoulder in external rotation, which exposes more of the articular surface of the humeral head anteriorly and increases the intraarticular area available for needle insertion. The first step requires localization of the desired needle position, which is medial to the superior third of the humeral head that is covered by the joint capsule. Prep the area, drape it in sterile fashion, and anesthetize the subcutaneous tissue. Needle size may vary, but commonly, a 20- to 22-gauge 3.5-inch spinal needle is used. The needle tip is then advanced in an anteroposterior direction to the humeral head, avoiding contact with the glenoid labrum.

The posterior approach requires the patient to be positioned prone with the ipsilateral shoulder raised off the table with a pad. The needle is directed toward the inferomedial aspect of the humeral head. After local anesthesia, a 21-gauge spinal needle is advanced vertically under fluoroscopic guidance toward the cartilage of the humeral head.

Chung and colleagues have demonstrated that an anterior approach may result in penetration of the anterior stabilizing structures of the glenohumeral joint, which has a tendency toward anterior instability. The study included six shoulders from fresh cadavers, using an 18-gauge needle with markers in the anterior and posterior approaches. The marker for the anterior approach traversed through the subscapularis muscle or tendon in all cases, the inferior glenohumeral ligament in two cases, and the anterior inferior labrum. On the other hand, the marker for the posterior approach traversed the posterior inferior labrum in a single case without violation of the anterior structures. Thus, it has been recommended to use the posterior approach if the patient presented with anterior instability or anterior syndromes. The posterior approach will decrease the likelihood of injecting into the subscapularis tendon or inferior glenohumeral ligament.

On confirming intraarticular location using as little iodinated contrast agent as possible (2 to 3 mL), approximately 15 mL of a 0.1 mmol/kg solution of gadopentetate dimeglumine is injected and the patient is taken to the MR scanner for sequence acquisition. The volume of the injection ranges between 10 and 20 mL. Injections of less than 15 mL decreases the likelihood of extra-articular leak, which may be mistaken for a full-thickness rotator cuff tear. Exercise after shoulder arthrography has no beneficial or detrimental effect on MRI quality or on the depiction of rotator cuff tear. Ideally, MRI should begin within 30 minutes of joint injection to minimize absorption of contrast.

Sensitivity and specificity for detection of superior labral anteroposterior lesions is better with additional maneuvers such as arm traction using 1- to 3-kg weights applied to the wrist combined with external rotation than if no arm traction is used.

An alternative to fluoroscopy-guided is ultrasound-guided injection of the glenohumeral joint, which has been found to be significantly less time consuming, to be more successful on the first attempt, to cause less patient discomfort, and to obviate the need for radiation and iodine contrast. It may also be done with open MRI guidance and even blindly, resulting in considerable time savings and scheduling coordination.

Direct MR arthrography allows detection of capsulolabral pathology and partial-thickness and vertical rotator cuff tendon tears. Better distention of the joint capsule, particularly the labral-ligamentous complex, allows easier depiction of irregular tears versus more smoothly delineated anatomic variants such as sublabral sulci and foramina.

Its limitations include patients' adversity to the procedure, safety, cost, extra time and labor per procedure, scheduling, nursing, ionizing radiation, coordinating use of both fluoroscopy suite and MRI, and the direct involvement of the radiologist. Postprocedure assessment of pain and discomfort shows that direct arthrography is better tolerated than the MRI itself. The potential for contrast reaction is considerably greater for the iodinated agents. Although the overall risk of complications with minimally invasive intraarticular needle arthrography is very low, the potential complications are considerable. An infection introduced to the joint can result in septic arthritis, osteomyelitis, or fasciitis. Arthrography may cause direct damage to the nerves, capsule, or ligaments. Leak of contrast through the capsular puncture site can cause spread of contrast along the fascial planes into the subdeltoid space, causing a “bursogram,” which can be misinterpreted as a full-thickness rotator cuff tear. Accidental injection of gas can lead to an incorrect diagnosis of loose bodies from the magnetic susceptibility artifact. However, attributing the exact cause of a susceptibility artifact should be based on its location (i.e., joint bodies are typically located in the more dependent portions, and gas bubbles rise to the nondependent portions of the joint).

Ultrasonography is rapid, low cost, and preferred by patients. Middleton and colleagues surveyed 118 patients who underwent both ultrasonography and MRI of the shoulder for suspected rotator cuff disease and found that 79% preferred ultrasonography. Even if MRI remains the primary modality for suspected rotator cuff disease, ultrasonography, rather than conventional arthrography, should be considered the alternative modality for patients with contraindications to MRI because it is noninvasive, is more rapidly performed than arthrography, and can demonstrate tendinosis, bursal surface tears, and subdeltoid bursitis, none of which can be diagnosed arthrographically. It offers greater resolution than MRI and allows “dynamic” scanning of tendons in motion. A subluxing biceps tendon may be demonstrated during internal and external rotation of the humerus, and entrapment of the supraspinatus tendon and subdeltoid bursa between the greater tuberosity and the acromion may be demonstrated during abduction of the arm. Communication during the exam is also advantageous.

It is more operator dependent than MRI and has a long learning curve for both performing and interpreting the examination. This is because the images are not anatomically intuitive as they are with MRI or CT and because the diagnostic images are dependent on proper patient positioning and transducer placement. Ultrasonography cannot assess the joint space as well as MRI and cannot demonstrate the deep surface of the acromioclavicular joint for capsular hypertrophy or spur formation, either of which may narrow the supraspinatus outlet and cause clinical impingement. A torn supraspinatus tendon that has retracted more than 3 cm cannot be demonstrated, but this is a relative limitation because nonvisualization of the tendon edge indicates retraction at least beyond the top of the humeral head (i.e., beyond the 12 o'clock position). A qualitative grading system of mild, moderate, or severe atrophy has yet to be developed for fatty atrophy of the supraspinatus muscle, which can be demonstrated as echogenic replacement of muscle.

The structures of interest are superficial and highly ordered, linear structures, so a high-frequency (5 to 12 MHz) linear transducer is used. Tissue harmonic imaging, if used, may increase the conspicuity of tears but does not increase the diagnostic accuracy. The ultrasound beam must be perpendicular to the tendon, because angulation can create artifactual hypoechogenicity, simulating a tear—an artifact known as anisotropy, a sonographic phenomenon created when highly organized parallel tendon fibers are not 90 degrees to the insonating beam ( Fig. 6-6 ).

The patient is seated, and the examiner may stand or sit facing, at the side of, or behind the patient. Unlike sonographic evaluation of other joints in the body, which focuses on the specific tendon or ligament of clinical suspicion, the examination of the shoulder should encompass all four muscles and tendons of the rotator cuff, the tendon of the long head of the biceps, the subdeltoid bursa, and the acromioclavicular joint, because all of these structures may be involved in rotator cuff dysfunction. The posterior joint capsule, posterior labrum, and spinoglenoid notch can also be visualized and should be evaluated.

To evaluate the biceps tendon, the forearm is supinated and placed on the thigh, bringing the bicipital groove anteriorly. By placing the transducer transversely across the humeral head, the long tendon of the biceps will be seen in cross section in the bicipital groove ( Fig. 6-7 ; see also Video 6-2 ). By turning the transducer longitudinally, the biceps tendon will appear as an echogenic fibrillar structure between the deltoid and humerus ( Fig. 6-8 ). The tendon can be followed distally from its musculotendinous junction to its proximal aspect around the humeral head. The patient's arm is then externally rotated, making sure to keep the elbow as close to the body as possible, thus bringing the subscapularis tendon into view in cross section, analogous to an oblique sagittal MR image ( Fig. 6-9 ). By turning the transducer 90 degrees so that it is transverse to the arm, the longitudinal extent of the subscapularis tendon is seen as it inserts on the lesser tuberosity ( Fig. 6-10 ). Some fibers extend across the bicipital groove, forming the transverse humeral ligament.

To visualize the supraspinatus tendon, the patient's hand is placed behind the back (the “Crass” position ) or on the buttock with the elbow pointed back (the “modified Crass” position). These maneuvers bring into view the distal aspect of the supraspinatus tendon, which would otherwise be obscured by the acromion. Either of these positions is accurate for diagnosis of full-thickness tear of the supraspinatus tendon, but the modified Crass position overestimates the size of the tear in the transverse plane. The transducer should be oriented 45 degrees or approximately midway between the transverse and longitudinal planes to visualize the longitudinal course of the supraspinatus, analogous to the oblique coronal plane on MRI ( Fig. 6-11 ). The transducer is then rotated 90 degrees to visualize the tendon in the transverse plane, analogous to an oblique sagittal MR image. In this plane, the deltoid muscle, which is hypoechoic with hyperechoic fascial planes, is just deep to the subcutaneous fat. Underneath the deltoid is the normally thin subacromial-subdeltoid bursa surrounded by thin hyper­echoic peribursal fat. The supraspinatus tendon appears echogenic and sits directly on the humerus. A thin anechoic rim of cartilage covers the hyperechoic cortex ( Fig. 6-12 ). As the transducer is moved anteriorly around the curvature of the humeral head in this oblique transverse plane, the biceps tendon will be seen in cross section. The biceps tendon is used as a reference point for measuring the location of tears or other abnormalities. The 2 cm of cuff tissue immediately posterior to the biceps tendon is the supraspinatus. Posterior to that is the infraspinatus.

Finally, the posterior aspect of the shoulder is evaluated. The arm is brought forward and placed on the thigh, or the patient can be asked to reach across his or her chest to the contralateral arm. The transducer is positioned just inferior to and parallel to the spine of the scapula. The infraspinatus muscle is followed laterally as it crosses the joint and becomes the tendon, to its insertion on the greater tuberosity ( Fig. 6-13 ). Sliding the transducer medially allows the posterior aspects of the humeral head and glenohumeral joint to be seen. The posterior glenoid labrum appears as a homogeneous hyperechoic triangle ( Fig. 6-14 ). Sliding the transducer even more medially will bring the spinoglenoid notch into view ( Fig. 6-15 ).

Normal Anatomy

Labrum

The glenoid labrum is a fibrous structure surrounding the edge of the osseous glenoid, which increases the depth of the glenoid fossa and hence the stability of the glenohumeral joint. More importantly, it serves as the anchoring structure for the glenohumeral ligaments and the long head of the biceps tendon superiorly ( Fig. 6-16 ). The normal labrum is firmly attached to the glenoid margin of the scapula and the scapular periosteum.