Chapter 3: Imaging the Shoulder

Stephen Austin MBBS, FRCR

Consultant Radiologist,

Great Ormond Street Hospital for Sick Children,

London

Introduction

For many years, plain X-rays were the only available means for imaging the shoulder and, as a result, there has been a tendency for radiologists to concentrate on bony aspects of shoulder joint pathology at the expense of the surrounding soft tissues. The newer imaging methods are much more sensitive to soft tissue abnormalities, and an impressive range of these techniques has been applied to shoulder imaging:

Some of these methods (such as MRI) are still under evaluation, and others have found favour with a small number of enthusiasts (digital subtraction arthrography and subacro­mial bursography, for example). This chapter aims to review the available imaging modalities and to discuss the relative pros and cons of each technique.

Plain radiography

The standard anteroposterior (AP) and axial films of the shoulder will be the first imaging investigation for almost all patients presenting with shoulder disease. To get the most informa­tion out of these films, meticulous radiographic technique is essential.

The standard AP projection of the shoulder is taken with the patient slightly oblique, with the arm supinated, and in slight abduction, and should be centred to the coracoid process. The film should be exposed so as to demonstrate both bony and soft tissue detail (Figure 3.1). The acromion, and the greater and lesser tuberosities, should be visible. In the normal patient, the vertical distance between the inferior surface of the acromion, and the super­ior margin of the humeral head (the so-called acromiohumeral interval) should be at least 7 mm. A reduction in this measurement is suggestive of rotator cuff disease, although it is important to point out that this measurement can be artifactually reduced by radiographic factors, such as centring too low, for example. It is often possible to see a thin, dark line of fat in this region (the peribursal fat plane), which represents extrasynovial fat surrounding the subacromial/subdeltoid bursa. The fat line can be seen as a radiolucency lying deep to the deltoid and extending to the greater tuberosity. It is normally 1-2 mm thick, and is best seen in films taken in slight internal rotation. Failure to visualize the fat plane is a sensitive but non­specific sign of periarticular disease, including tears of the rotator cuff.1

Figure 3.1 Standard AP radiograph of the shoulder.

The AP projection is of great value in the detection of soft tissue calcification (Figure 3.2). Many eponymous views have been described to assist in detecting hydroxyapatite deposition in different components of the rota­tor cuff. However, unless non-standard views are performed frequently, they are likely to be poorer in quality, and unfamiliar to their inter­preters, which limits their value.

Figure 3.2 Soft tissue calcification on AP radiograph.

Figure 3.3 Standard axial view.

Figure 3.3 shows the standard axial view, which has been taken with the palm down. In this position, the greater tuberosity lies post­eriorly, and the lesser anteriorly. In the injured patient, the standard axial view may be difficult to obtain. In these circumstances, a modified axial view,  developed by Wallace and colleagues 2 can be helpful. In this technique, the patient is sitting rotated, so that the scapula is parallel to the edge of the horizontal cassette, which is in contact with the arm. The tube is angled at 30 degrees. In this position, the central ray passes just lateral to the coracoid process, through the joint, to the cassette. The effect of this is to produce a magnified, and slightly distorted image of the shoulder (Figure 3.4), but with the advantage that the patient can be X-rayed without removing slings or collars and cuffs.

The search for the hatchet lesion, the hume­ral head defect produced by recurrent anterior dislocation, has generated a large number of eponymous radiographic projections. Whenihe lesion is large, it can be demonstrated by an internally rotated AP projection, or a palm-down axial. More subtle lesions may require special­ized techniques like the Stryker view, arthrogra­phy, or CT to demonstrate them.

A number of indirect plain film signs of rotator cuff disease have been described. In 1964, Cotton and Rideout3 analysed over 100 cases, both radiologically and pathologically, at post­mortem. They found that cyst formation in the upper two-thirds of the anatomical neck was a reliable and constant sign. The acromiohumeral interval was reduced in many cuff tears, but could be normal in the presence of severe disruption. Sclerosis of the greater tuberosity was a very unreliable sign, but sclerosis or remodelling of the inferior acromial surface correlated well with the presence of a cuff tear. The formation of a subacromial spur, an area of new bone formation arising from the inferior surface of the acromion4 has also been descri­bed in some patients with full thickness tears.5

In general, plain films of the shoulder are of greatest value in acute trauma, and in the assessment of calcific periarthritis. Many abnormalities can be seen which suggest the presence of soft tissue disease, but these are usually non-specific, and the X-ray may be completely normal in the presence of a com­plete rotator cuff disruption.

Figure 3.4 Nottingham axial view.

Arthrography

Arthrography is at present the most common special radiological investigation of the shoul­der performed, and its practice is still on the increase, although there are suggestions that it may be supplanted by less invasive methods.[6] There are a number of different ways of per­forming a shoulder arthrogram. The single contrast technique, using iodinated contrast media alone, was first described in the 1930s,[7] but the double contrast technique (using less contrast, but with the addition of air) with erect filming provides better visualization of the soft tissue anatomy, and is now more widely per­formed.

Many different techniques for puncturing the joint are described, but the author uses an
anterior approach,[8] which gives cons stent results. After a standard series of control films, which include internal and external rotat on AP films, and an axial film, the patient is placed supine on the X-ray screening table, w'th the arm elevated slightly, and in a neutral position. A point 2 cm lateral and below the coracoid is marked. The skin is cleaned and infiltrated with local anaesthetic, and a 21-gauge, short bevel­led needle is advanced under screening con­trol, with a slight medial inclination into the joint. A useful way of telling if the needle is in the joint is to fill the hub of the needle with local anaesthetic. When the capsule is punctured, the level of fluid will fall.

Five ml of positive contrast medium is then injected, followed by 15-20 ml of room air. If the needle is in the correct place, contrast wiII flow into the axillary recess and subscapularis bursa. Gentle manipulation of the arm, and particularly adduction, may force contrast through a defect in the rotator cuff into the subacromial bursa, which does not normally fill. After contrast has been injected, a further series of films is taken, including erect films.

Figure 3.5 Normal shoulder arthrogram. Note that long head of biceps can be seen, as can the intra-articular triangle, into which the arthroscope is inserted.

Figure 3.5 from a normal arthrogram series shows the contrast confined to the joint cap­sule, outlining the intra-articular portion of the tendon of the long head of biceps, and the tendon sheath. Contrast here outlines the deep surface of the rotator cuff, and there is no sign of contrast in the region of the subacromial bursa. Although non-filling of the tendon sheath was at one time regarded as abnormal, it is now known that this can be seen in up to 10 per cent of normal shoulders. Leakage of contrast from the biceps tendon sheath can also be seen occasionally, and this is also regarded as normal. In the presence of a full thickness cuff tear, the joint capsule is continuous with the subacromial or subdeltoid bursa, and contrast will enter this space.

Figure 3.6 Arthrogram in internal rotation showing rotator cuff tear, with leak of contrast material into the subacromial bursa. Contrast has also leaked down the biceps tunnel (which is normal) and then escaped.

Figure 3.7 Normal CT arthrogram of the shoulder showing anterior and posterior aspects of the labrum.

On the internal rotation view (Figure 3.6) the superior surface of the cuff is demonstrated but is deficient at one point, which is the site of the defect. On internal rotation, the infraspinatus muscle is brought to lie laterally and under the acromion on the AP view, and if this does not appear intact, then the tear must be large in size. Sometimes, little contrast will enter the subacromial bursa, but the site and size of the tear will be well shown. Occasionally, with a small tear, positive contrast will not enter the subacromial bursa, but air will. This can easily be confused with the fat outlining the sub­deltoid bursa, which is most easily seen on the internal rotation view. If there is any doubt, delayed films and/or CT should be performed.

Axial views with the double contrast method can demonstrate the glenoid labrum well, particularly the anterior portion, and this has been combined with conventional and com­puted tomography in the investigation of shoul­der instability. Conventional arthrography has also been used in other clinical disorders, notably 'adhesive capsulitis', where a small-volume joint, with obliteration of normal synovial recesses, and lymphatic intravasation have been reported. This is one situation where the single contrast technique is probably superior. Treatment of the disorder by repeated capsular distension during arthrography has been described.[9]

The major disadvantages of arthrography for rotator cuff lesions are that partial thickness tears will only be shown if they involve the deep surface of the cuff, and these only occasionally. Partial tears involving the superficial surface or body of the cuff cannot be demonstrated. In addition, filling of the subacromial bursa is not unusual following a successful rotator cuff repair, so that the technique is not as helpful in the assessment of postoperative problems. However, it is quick and simple to perform, requires little or no specialized equipment, and has a very low complication rate.

CT arthrography

The arthrography technique is slightly modified if CT is used, with typically only 1-2 ml of positive contrast and 10-15 ml of air being injected. Thin slices (3-5 mm) are obtained through the shoulder joint, with the arm in s ght internal rotation. Selected slices may be repeat­ed with the arm in external rotation to demons­trate the posterior labrum more effectively.[10] This normal example (Figure 3.7) shows both the anterior and posterior aspects of the labrum.

The main use of the technique is in the investigation of recurrent shoulder instability. Hatchet deformities of the humeral head are well demonstrated by this method. All les'ons were detected by CT arthrography, and there were no false positives in those patients in whom surgical confirmation was available in the above quoted series.

Bony glenoid rim (Bankart) lesions are also clearly seen, but one of the major advantages of the technique is the ability to demonstrate abnormalities of the glenoid labrum. Compared with surgery or arthroscopy, the sensitivity for detection of labral detachment or tears using the technique is close to 100 per cent, particu­larly for the anterior portion. Humeral head lesions are exceptionally well demonstrated by this technique, as are abnormalities of the biceps tendon.

The subacromial bursa during abduction and external rota­tion is noted.[5] The technique is rather more difficult than shoulder arthrography, and the normal bursa can be problematic to find with a needle. It is not widely practised, and many reports of its use come from Japan, where it has also been used in the diagnosis of adhesive capsulitis.

Digital subtraction arthrography

This method uses digital subtraction techni­ques taken from the field of angiography, and requires expensive specialized equipment. It involves injection of contrast into the joint, while keeping the arm still, and taking images at 1 second intervals. The images are then sub­tracted by a computer from a mask image without contrast, in order to obtain bone-free images. Enthusiasts claim that it is more sensi­tive in detecting full thickness and deep surface tears, and can show the site and size of a tear more accurately than the conventional techni­que. It can also be used to evaluate shoulder replacements for loosening. It is technically more difficult than conventional arthrography, and has not found wide application.

Subacromial bursography

The subacromial bursa may be entered inadvertently during conventional arthrography, but direct injection has been used to demons­trate cuff tears, and particularly those partial tears of the superficial surface of the cuff which cannot be shown by conventional arthrogra­phy. It has also been used in the investigation of impingement syndrome, when pooling of contrast in the subdeltoid part of the subacroacromial space. When pooling of contrast in the subdeltoid part of the subacromial bursa during abduction and external rotation is noted [5].  The technique is rather more difficult then shoulder arthrography, and the normal bursa can be problematic to find with a needle.  It is not widely practised, and many reports of its use come from Japan, where it has also been used in the diagnosis of adhesive capsulitis.

Ultrasound

Ultrasound is not a technique commonly asso­ciated with orthopaedics, although interest in the technique is growing. Ultrasound of the rotator cuff was first described in 1984.[11] However, the advance of technology has been such that the development of high resolution scanners and high frequency probes, typically 7.5-10 MHz, has improved the accuracy of the technique. The advantages of the technique are that it is quick and non-invasive, and it is easy to examine both shoulders at one sitting. Images can be obtained in several different planes.

Figure 3.8 Normal rotator cuff on coronal ultrasound: right shoulder seen from in front. D = Deltoid, C = Rotator cuff, H = Humeral head.

The technique has the potential to demons­trate partial thickness tears. Figure 3.8 is an example of a normal cuff in the coronal plane, with the greater tuberosity, and the deltoid muscle lying superficially. Images are also obtained in the sagittal plane, and transversely. It is also important to observe the cuff during rotation of the shoulder. The criteria used to diagnose a cuff tear are controversial. There is general agreement that visualization of a defect, or non-visualization of the cuff as in Figure 3.9 where only the deltoid is visible, and no cuff can be seen, are reliable criteria for the diagnosis of a cuff tear.

Figure 3.9 Full thickness rotator cuff tear on sonography (sagittal scan). Top shows line diagram of the normal normal appearance. Middle shows line diagram of rotator cuff tear. Bottom shows sonogram of rotator cuff tear (dark arrows top of cuff and tear, light arrows bottom of cuff). D = Deltoid, C = Cuff, H = Humeral head.

The significance of areas of altered echo­genicity (Figure 3.10) is less certain. Some may represent full thickness and some partial thick­ness tears, but the changes are not consistent, and alterations in echogenicity alone should not be used as the sole criterion for diagnosis.[12] Areas of fibrosis or calcification can also pro­duce abnormal echogenic areas.[13] In one prospective series of 51 shoulders, ultrasound had a sensitivity of 100 per cent for the detection of a cuff tear, but the specificity was only 75 per cent, and the overall accuracy 92 per cent. The major disadvantages of the technique are the time it takes to learn the method, and the fact that arthrography may still be required before surgery. It is sufficiently sensitive to exclude serious cuff tears with confidence in good hands.

Figure 3.10 Coronal scan of left shoulder showing a highly echoic insertion of the supraspinatus tendon.

There have been reports of the use of ultrasound in the evaluation of the postopera­tive shoulder. This is a situation in which arthrography is less helpful, as a successfully repaired cuff may still leak contrast. Unfortun­ately, postoperative changes in cuff echogenic­ity are common, and acromioplasty removes a normal ultrasonic landmark, making it difficult to separate cuff from overlying deltoid. Never­theless, if the criteria of non-visualization, or demonstration of a defect are used, the techni­que is highly accurate in the diagnosis of recurrent tears.[14]

Magnetic resonance imaging (MRI)

With the development of high field NMR scan­ners and improved surface coil technology, it has become feasible over the last few years to undertake high resolution shoulder MRI. The peripheral location, and unique anatomy of the shoulder joint does pose some technical prob­lems for MRI, however. Surface coils are essential to provide high spatial resolution, and novel coils have been designed specifically for shoulder imaging.[15] Software modifications may also be required (for example, to alter the centre frequency) to allow off-centre zooming.[16] T2-weighted images are essential, but either T-i-weighted or proton density images are also important to assess morphology, and the subdeltoid fat plane. Axial images are obtained, as well as coronal oblique images, in the plane of the supraspinatus tendon, and sagittal oblique images, perpendicular to this plane (Figure 3.11).

Figure 3.11 Coronal Tyweighted spin-echo image of a normal shoulder showing the long head of biceps, and the rotator cuff lying between the joint and the subdeltoid bursa.

Like ultrasound, there is no clear consensus on the criteria for diagnosis of a tear. T2-weighted spin echo images frequently show areas of high signal within the cuff, but some of these patients will prove to have intact cuffs at surgery. In general, a normal intensity cuff with a normal subdeltoid fat plane is likely to be normal. An area of increased signal in the region of the cuff on T2-weighted images, with normal morphology, and a normal subdeltoid fat plane probably indicates cuff degeneration without a tear. Most patients with a cuff tear show increased cuff signal, or discontinuity. Loss of the subdeltoid fat plane or fluid in the subacromial bursa on T2-weighted images are also very suggestive findings.[17]

In most of the published series, MRI com­pares favourably with arthrography, although there are always a few tears missed by MRI, but shown by arthrography.[18] Some partial thick­ness tears are only demonstrated by MRI, however.   In   impingement   syndrome,   MRI proved very sensitive to cuff degeneration in one series.[19] However, subacromial injection of steroids can produce MRI changes which mimic cuff pathology. This is a potential draw­back, as many patients will have had local injections before being referred for MRI.

Results in recurrent shoulder dislocation are less encouraging, as the normal glenoid labrum is a low intensity structure on most sequences, and is consequently harder to assess. CT arthrography may be more sensitive in the detection of labral pathology, although it can be demonstrated by MRI. Similar reservations also apply to the diagnosis of capsular stripping on MRI. This is readily demonstrated by CT arthrography, as the joint is distended during the procedure. MRI can only detect capsular separation in the presence of a joint effusion.[20]

The accuracy of MRI in the detection of humeral head defects is probably equal to that of CT arthrography.