MRI Shoulder – Generic Standard Protocol
Required Protocol at a Glance
Mandatory core sequences for this examination. Detailed rationale, conditional additions and optimisation notes are provided later in the protocol.
1. Executive Summary
MRI is the definitive modality for soft tissue characterisation of the shoulder. It provides the only non-invasive imaging pathway capable of simultaneously evaluating the rotator cuff tendons, the glenoid labrum and capsuloligamentous complex, the long head of the biceps tendon, the subacromial-subdeltoid bursa, the glenohumeral and acromioclavicular joint cartilage, the surrounding musculature, and the osseous structures including marrow signal. No other modality achieves this breadth in a single examination.
The generic non-arthrographic shoulder MRI protocol described in this page is designed for the broad population of patients presenting with non-specific shoulder pain, suspected rotator cuff disease, clinical signs of impingement, early inflammatory arthropathy, post-traumatic pain without specific structural suspicion, or for pre-operative assessment of moderate severity. It deliberately excludes the MR arthrographic component that is essential for high-sensitivity labral tear detection.
1.1 Core Strengths
Rotator cuff assessment is the primary clinical strength of shoulder MRI. Full-thickness tears of the supraspinatus tendon are detected with pooled sensitivity 84–91% and specificity 97% on non-arthrographic MRI; partial-thickness tears with sensitivity 80% and specificity 95% [2]. At 3T, non-arthrographic sensitivity for full-thickness tears approaches 98–100% in experienced centres [4].
Bone marrow and osseous pathology: superior to all other modalities for early bone marrow oedema, stress fractures, Hill-Sachs lesions, bony Bankart lesions, and early avascular necrosis of the humeral head. STIR and fat-suppressed sequences demonstrate marrow changes days before radiographic abnormality. For sequence-level protocol optimisation, vendor terminology and artefact management, see the dedicated MRIninja page STIR Sequence.
Bursal and periarticular pathology: the subacromial-subdeltoid bursa, acromioclavicular joint, and surrounding soft tissues are well demonstrated. Bursitis, fluid collections, and calcific deposits are reliably identified.
Muscle assessment: fatty infiltration and atrophy of the rotator cuff muscles — measured by the Goutallier classification or Dixon-based quantification — are key prognostic factors for surgical outcomes and are visible only on MRI.
Inflammatory arthropathy: bone marrow oedema, synovitis, erosions, and effusion are assessed comprehensively on a single non-invasive study.
1.2 Intrinsic Limitations of the Generic Protocol
A non-arthrographic generic shoulder MRI protocol is a deliberate compromise between breadth, acquisition time, and diagnostic specificity. It is not optimised for every clinical question.
Labral tear detection is the most important limitation. The non-arthrographic protocol has substantially reduced sensitivity for labral tears compared to direct MR arthrography (MRA). For anterior labral tears (Bankart lesions), MRA sensitivity is 91–94% versus approximately 60–70% for non-arthrographic MRI. For SLAP lesions, MRA sensitivity is 86% versus approximately 56% for conventional MRI [6, 7]. When labral pathology is the primary clinical question — history of dislocation, suspected instability, overhead athlete with superior labral pain — the generic protocol is insufficient and direct MRA is required.
Partial-thickness rotator cuff tears are significantly better detected by MR arthrography (sensitivity 67–87%) than non-arthrographic MRI (pooled sensitivity approximately 80% on standard sequences, lower on 1.5T) [2, 6]. When a partial-thickness tear is the operative question, MRA provides more reliable characterisation.
Cartilage assessment is limited at standard protocol resolution. Focal chondral defects below 5 mm and early-grade cartilage softening require dedicated high-resolution sequences and, ideally, cartilage-specific protocols.
Post-operative shoulder presents additional limitations. Metal susceptibility from anchors and hardware degrades image quality around the operative site. Post-operative labral and tendon repair integrity requires dedicated arthrographic protocols in most cases.
When dedicated child protocols are required: MR arthrography (direct) for labral tears and instability; ABER position supplementation for anteroinferior labral and partial undersurface rotator cuff tears; post-operative protocol with metal artefact reduction for revision assessment; dedicated cartilage protocol; and dedicated brachial plexus or neurovascular protocol when neuropathic pain is suspected.
2. Main Clinical Indications
2.1 Standard Indications
Rotator cuff disease is the most common indication for shoulder MRI. This includes the spectrum from tendinopathy to partial and full-thickness tears, calcific tendinopathy, and subacromial impingement syndrome. The ACR Appropriateness Criteria designate MRI without contrast as "usually appropriate" for atraumatic shoulder pain in the context of rotator cuff evaluation when clinical signs suggest significant tendon pathology or when conservative management has failed [1]. The generic protocol is sufficient for this indication in the large majority of cases.
Post-traumatic shoulder pain accounts for a substantial proportion of shoulder MRI requests. After acute trauma, the primary targets are rotator cuff tears (acute full-thickness tears in middle-aged and older patients, avulsion-type injuries in younger patients after dislocation), bony injuries not visible on radiographs, and acromioclavicular ligament injuries. For acute post-dislocation shoulder in athletes where instability repair is planned, direct MR arthrography is preferred over the generic protocol because labral integrity directly determines surgical approach. However, for the first post-traumatic MRI in the non-athlete population, the generic protocol provides adequate initial assessment.
Suspected adhesive capsulitis (frozen shoulder) does not require a modified protocol. The generic shoulder MRI demonstrates the characteristic features — axillary recess capsular thickening and T2 signal reduction, coracohumeral ligament thickening, obliteration of the fat pad in the rotator interval — sufficiently for diagnosis. The generic protocol is appropriate.
Inflammatory arthropathy including rheumatoid arthritis, spondyloarthropathy (reactive, psoriatic, ankylosing spondylitis), calcium pyrophosphate arthropathy, and crystal arthropathy affecting the shoulder joint is well assessed on the generic non-contrast protocol. If synovitis quantification is needed for treatment response assessment, post-contrast sequences are indicated.
Acromioclavicular joint disorders including osteoarthritis, impingement by hypertrophic osteophytes, and distal clavicle osteolysis are visible on the generic protocol, which includes direct coverage of the AC joint on all three planes.
Biceps tendon pathology including tendinopathy, partial and full rupture of the long head, and subluxation from the bicipital groove is well assessed on the generic protocol, particularly on axial sequences.
Pre-operative planning for rotator cuff repair surgery benefits from the full generic protocol including assessment of tear size, retraction, muscle atrophy and fatty infiltration, and bone architecture. These are the primary determinants of surgical feasibility and outcome prediction.
Suspected avascular necrosis of the humeral head is an indication where MRI is definitively superior to all other modalities, demonstrating subchondral changes at a stage when radiographs and CT are normal.
Occult fractures — particularly greater tuberosity fractures after acute trauma — are indicated when radiographs are negative but clinical suspicion is high. STIR sequences identify marrow oedema with high sensitivity in this context.
2.2 Urgent Red Flags Requiring Expedited or Emergency Imaging
The shoulder joint does not typically produce life-threatening emergencies, but the following clinical scenarios require expedited imaging or clinical escalation.
| Red flag scenario | Recommended action |
|---|---|
| Acute massive rotator cuff tear with marked functional loss in trauma | Expedited MRI within days; confirm diagnosis before surgical planning |
| Suspected septic arthritis of the glenohumeral joint | Urgent MRI (same day if possible); MRI provides joint effusion characterisation and marrow oedema assessment prior to joint aspiration |
| Humeral head fracture with neurovascular compromise | CT first for fracture characterisation; MRI adjunctive if brachial plexus injury suspected |
| Acute dislocation in young athlete (first event) | Radiographs first to exclude fracture; MRI or MRA within 1–2 weeks for surgical planning |
| Suspected rapidly progressive bone lesion / primary bone tumour | Urgent MRI for complete staging sequence; do not delay for generic shoulder pain indications |
| Post-operative acute neurological deficit | Urgent MRI with MAIS/MARS sequences if metal hardware is present |
3. Preparation Reference
Universal MRI safety screening, contraindication assessment, and IV access documentation are covered in the general MRI preparation page and are not repeated here. The following items are specific to the shoulder.
3.1 Anatomy-Specific Preparation Items
Arm position and internal rotation avoidance: the standard position for non-arthrographic shoulder MRI is the arm at the side in slight external rotation or neutral position. Internal rotation of the arm reduces the cross-sectional area of the supraspinatus visible in the oblique coronal plane and impairs evaluation of the anterior labrum [3]. Patients who present with the arm naturally internally rotated due to pain should be given adequate time to relax; gentle padding can assist maintaining a neutral or slightly externally rotated position.
Clothing: any metallic fasteners, underwire bras, or clothing with metallic prints over the shoulder region must be removed. A gown open at the relevant shoulder should be used.
AC joint and clavicle area metalwork: document any history of prior AC joint surgery, clavicle fixation hardware, or shoulder arthroplasty. Metal hardware at or near the shoulder significantly affects image quality and may require metal artefact reduction sequences. Declare this to the reporting radiologist.
Prior injection history: corticosteroid or hyaluronic acid injections into the subacromial bursa within 4–6 weeks of the MRI may produce bursal fluid signal alteration. Document injection date in the request history. Intra-articular gadolinium injection history in prior MR arthrography produces persistent signal alteration for variable periods.
Post-operative shoulder: patients with prior rotator cuff repair, Latarjet procedure, labral reconstruction, shoulder arthroplasty, or AC joint reconstruction require specific protocol modification and advance notice to the department. Anchor type and metal composition determine the degree of susceptibility artefact and the need for MARS sequences.
Bilateral assessment: when bilateral shoulder pathology is clinically suspected (inflammatory arthropathy, bilateral impingement), both shoulders may be imaged in sequence. However, standard coil coverage is unilateral; repositioning is required between sides.
Contrast preparation: for standard non-contrast shoulder MRI, contrast preparation is not required. For post-contrast examinations (inflammatory assessment, post-operative evaluation, suspected neoplasm), standard IV access and GBCA safety screening apply.
3.2 Patient Positioning on the MRI System
Standard position: supine, feet-first entry. The affected shoulder is positioned as close to isocentre as possible. Standard gantry positioning for shoulder MRI typically requires slight lateral offset toward the ipsilateral shoulder to optimise coil coverage and isocentre placement.
Arm position: arm at the side, thumb pointing upward (slight external rotation, 10–20°). This position:
- opens the supraspinatus outlet, reducing magic angle effect on the most anterior supraspinatus fibres
- places the anterior labrum in the optimal position for assessment
- relaxes the capsuloligamentous complex
Padding under the elbow and forearm prevents minor involuntary motion from discomfort during acquisition.
Coil selection: dedicated surface coil or phased-array shoulder coil provides the highest SNR for shoulder MRI. The coil should be positioned to encompass the entire glenohumeral joint, supraspinatus muscle belly (cranially), the axillary recess (inferiorly), and the AC joint. Do not use a spine coil or body coil — the SNR penalty is unacceptable for diagnostic shoulder imaging.
Centering: isocentre at the centre of the glenohumeral joint. A practical landmark is the midpoint between the AC joint and the axillary crease, approximately at the level of the coracoid process palpated anteriorly. Verify centering on the topogram.
Immobilisation: wrap or secure the coil firmly against the shoulder without compressing the deltoid. Instruct the patient to breathe normally but avoid shoulder movement. Small bolster under the contralateral knee to reduce back discomfort during the examination reduces voluntary motion.
Common positioning errors:
- Insufficient external rotation: the most common error; reduces diagnostic quality for both rotator cuff and anterior labrum
- Excessive lateral patient offset: places the glenohumeral joint outside the optimal coil sensitivity region
- Coil too cranial: excludes the axillary recess from coverage; misses inferior capsular pathology
- Coil too caudal: reduces coverage of the supraspinatus muscle belly and distal clavicle
Practical check before scanning: on the three-plane localiser, confirm (i) the glenohumeral joint is visible in all three planes; (ii) the AC joint is included superiorly; (iii) the axillary recess is included inferiorly; (iv) the humeral head and acromion are within the coil sensitivity region.
4. Standard Protocol Design
The standard shoulder MRI protocol is built around three mandatory multiplanar sequences — the orthogonal triad of oblique coronal, oblique sagittal, and axial — each with specific tissue weighting optimised for the shoulder anatomy and pathology targets. Conditional and optional sequences are added based on the clinical question.
4.1 Mandatory Core Sequences
| # | Sequence | Plane | Status |
|---|---|---|---|
| 1 | PD-weighted TSE with fat suppression (PD-FS) | Oblique coronal | Mandatory |
| 2 | PD-weighted TSE with fat suppression (PD-FS) | Oblique sagittal | Mandatory |
| 3 | PD-weighted TSE with fat suppression (PD-FS) | Axial | Mandatory |
| 4 | T1-weighted TSE without fat suppression | Oblique coronal | Mandatory |
| 5 | STIR or PD-FS | Coronal or axial | Mandatory (bone marrow screen) |
4.2 Conditional Sequences
| Sequence | Indication | Plane |
|---|---|---|
| Post-contrast T1-FS (SPAIR/Dixon) | Suspected synovitis, inflammatory arthropathy, neoplasm, post-operative enhancement | Oblique coronal + axial |
| 3D isotropic PD-FS TSE (SPACE/VISTA/CUBE) | Desired MPR for complex anatomy, suspected subtle labral tear, cartilage assessment | Oblique coronal — reformatted |
| STIR (if not used as core) | Metal near field, B0 inhomogeneity, failed spectral fat suppression | Oblique coronal |
| ABER position PD-FS | Partial undersurface rotator cuff tear, anteroinferior labral tear | Oblique axial in ABER |
| T2*-weighted GRE | Suspected calcific tendinopathy characterisation, chondral fragment, haemorrhage | Oblique coronal or axial |
| Dixon T1 | Rotator cuff muscle fat fraction quantification (Goutallier grading) | Oblique sagittal |
4.3 Rationale Summary Per Sequence
Oblique Coronal PD-FS is the primary diagnostic sequence for the rotator cuff and represents the most clinically valuable acquisition in the shoulder MRI protocol. The oblique coronal plane, prescribed parallel to the supraspinatus tendon, displays the full thickness of the supraspinatus and infraspinatus tendons in longitudinal cross-section — the optimal orientation for detecting and characterising tears. Fat suppression is mandatory: the tendon and fluid signal would overlap without it, and the contrast between tendon, fluid (T2-bright), and the fat surrounding the bursal and articular surfaces is the diagnostic substrate.
PD weighting (TE 20–40 ms, long TR) is preferred over T2 weighting (TE 60–80 ms) for shoulder tendons because the magic angle effect — which is described further under artefacts — is minimised at shorter TE. The supraspinatus tendon fibres run at approximately 55° to B0 in the oblique coronal plane, making them vulnerable to magic angle T1 and PD signal artefacts. T2 weighting partially suppresses this effect but at the cost of reduced SNR and longer acquisition.
What this sequence detects well: full-thickness and partial-thickness supraspinatus and infraspinatus tears; tendinopathy; calcific tendinopathy (signal void); subacromial-subdeltoid bursitis; bone marrow oedema in the humeral head and greater tuberosity; os acromiale and acromial morphology.
What this sequence misses: subscapularis tears (better seen axially); anterior and posterior labral tears (better seen axially); biceps tendon in the groove (better seen axially).
Oblique Sagittal PD-FS provides the second major rotator cuff plane, showing the supraspinatus, infraspinatus, teres minor, and subscapularis in true cross-section at the level of the glenoid. This plane is essential for rotator cuff muscle assessment (atrophy and fatty infiltration), for characterising the acromial morphology and AC joint from a second plane, and for assessing the size and shape of the supraspinatus outlet. Tear retraction, which is a critical determinant of surgical feasibility, is measured on the sagittal plane. The sagittal plane is also the primary plane for assessing the coracohumeral ligament and the rotator interval.
Axial PD-FS is the primary plane for the glenohumeral labrum and the long head of the biceps tendon. The anterior and posterior labrum, the inferior glenohumeral ligament, the subscapularis tendon, and the biceps tendon at its origin and within the bicipital groove are all optimally displayed on axial sequences. A fatigue-suppressed PD on axial provides fluid sensitivity for cartilage assessment of the glenohumeral joint surface — though sensitivity for focal chondral lesions on standard 3 mm axial slices is limited.
Coronal T1 without fat suppression is the essential complement to the fat-suppressed sequences and provides three critical functions that cannot be achieved by fat-suppressed sequences alone: (i) characterisation of signal intensity — distinguishing T1-bright fat (lipoma, fatty infiltration of rotator cuff muscles) from fluid, haemorrhage, and proteinaceous material; (ii) anatomical landmark identification with high spatial resolution and no signal loss from failed fat suppression; (iii) assessment of bone cortex and trabecular structure, which is suppressed on fluid-sensitive sequences. The Goutallier staging of rotator cuff fatty infiltration relies on T1 (or equivalent non-fat-suppressed sequences) to grade the proportion of fat replacement within the muscle.
STIR provides bone marrow oedema sensitivity that is field-strength independent and robust to B0 inhomogeneity — critical in the shoulder region where the proximity of the humeral head, acromion, AC joint, and clavicle creates local field perturbations that degrade spectral fat suppression uniformity at the transition between these structures. STIR is the preferred bone marrow screening sequence in any shoulder with suspected stress reaction, occult fracture, early avascular necrosis, or inflammatory marrow involvement.
4.4 Sequence Matching and Cross-Sequence Consistency
The three primary planes (oblique coronal, oblique sagittal, axial) must be prescribed on the same oblique coronal reference — specifically the plane of the supraspinatus tendon — so that the oblique coronal and oblique sagittal images are genuinely orthogonal to each other. If the coronal and sagittal are not prescribed from the same reference, they will not be orthogonal and anatomical structures will not register between planes, complicating interpretation of tear size and location.
When post-contrast sequences are acquired, the post-contrast T1-FS must use the same slice prescription (same angulation, same FOV, same slice positions) as the pre-contrast T1 to allow meaningful comparison. Subtraction (post minus pre contrast) is not routinely used for shoulder MRI but may be applied when there is strong T1-bright signal (e.g. proteinaceous fluid, calcific deposit) that could obscure enhancement.
For serial follow-up studies, the oblique coronal angulation must be documented in the radiological report or saved as a protocol preset to enable reproducible tear size comparison between examinations.
4.5 Fat Suppression, Contrast-Specific and Region-Specific Technical Modifiers
Fat suppression is mandatory for fluid-sensitive sequences in shoulder MRI. Without fat suppression, the periarticular fat — which is abundant around the shoulder — overwhelms the fluid signal in the bursa, cartilage, and the tendon surface, eliminating the contrast that makes tear detection possible. This is not an optional optimisation; it is a diagnostic requirement.
Technique selection is field-strength dependent:
At 3T, spectral fat saturation (SPAIR or CHESS/ChemSat) provides excellent homogeneous fat suppression throughout the shoulder region and is the preferred choice for PD-FS sequences at 3T. The shorter wavelength at 3T means that local B0 inhomogeneity at the shoulder — particularly at the AC joint and the medial humeral head — produces some non-uniformity of spectral fat saturation, but this is generally acceptable at 3T with dedicated shimming.
At 1.5T, spectral fat saturation performs adequately for the shoulder in most patients, but B0 inhomogeneity at the corners of the FOV (lateral deltoid, medial chest wall) can produce incomplete fat suppression that simulates or obscures periarticular signal. STIR is more robust for bone marrow screening at 1.5T and should always be included when the spectral fat saturation quality is questioned.
Dixon fat suppression (mDixon, IDEAL, IDEAL IQ) provides the most robust, B0-independent fat suppression for shoulder MRI and is the preferred technique at both field strengths where available. It produces simultaneous fat-only and water-only images, which provide both diagnostic fat suppression and the fat-fraction maps needed for Goutallier grading of rotator cuff muscles from a single acquisition. Its implementation for standard PD-FS sequences is vendor-dependent and protocol calibration is required to achieve correct TE pairing.
STIR has a critical role as a backup fat suppression technique and as the primary bone marrow screening sequence when spectral methods are unreliable. However, STIR cannot be used post-gadolinium (see Section 6.3) — this rule applies identically to the shoulder as to the spine.
Post-contrast fat suppression: when gadolinium-enhanced sequences are acquired, fat suppression is mandatory for the post-contrast T1. Without it, the T1-bright periarticular fat and any T1-bright signal (haemorrhage, calcification, proteinaceous fluid) cannot be distinguished from true gadolinium enhancement. Dixon or SPAIR are preferred for post-contrast T1-FS in the shoulder.
4.6 Slice Positioning — Complete Technical Reference
Technical reference — click to expand / collapse
Why Slice Positioning Matters in Shoulder MRI
Correct slice angulation in shoulder MRI is not merely cosmetic. The diagnostic accuracy for rotator cuff tears, labral abnormalities, and tendon dimensions depends critically on the slices being genuinely aligned with the anatomical structures they are intended to assess. A supraspinatus tendon displayed at 30° obliquity instead of true longitudinal alignment appears spuriously thickened or thinned and produces false-positive or false-negative tear interpretation.
Anatomical Landmarks
The supraspinatus tendon runs from the greater tuberosity insertion medially along the floor of the subacromial space to the supraspinatus muscle belly. On the axial or sagittal localiser, it appears as a low-signal band extending lateral to the glenoid at the level of the supraspinatus fossa. This is the primary landmark for all oblique coronal and sagittal prescription.
The glenohumeral joint axis — the line from the centre of the humeral head through the centre of the glenoid — defines the true axial plane of the shoulder, which is rotated approximately 30–45° from the anatomical transverse plane of the body due to the retroversion of the glenoid and the external rotation of the arm.
The glenoid fossa seen on the coronal localiser provides the reference for sagittal oblique prescription.
Planning Sequence
All oblique planes are planned from the three-plane localiser (scout). Best practice is to use a fat-suppressed PD or T2-weighted scout if available, as it shows the supraspinatus tendon directly. A T1 or gradient echo localiser also provides adequate landmarks. For sequence-level protocol optimisation, vendor terminology and artefact management, see the dedicated MRIninja page Gradient Echo (GRE/FLASH) Sequence.
Oblique Coronal Slice Prescription
Reference: axial (transverse) localiser image at the level of the supraspinatus tendon.
Alignment: draw the prescription line parallel to the long axis of the supraspinatus tendon as it runs from the musculotendinous junction medially to the greater tuberosity insertion laterally. In most patients, this requires an angulation of approximately 40–50° from the true coronal plane (anterior to posterior obliquity corresponding to the inclination of the scapular spine). The exact angle varies with patient anatomy and arm position.
Coverage: the slice package must extend from anterior to posterior to include the entire supraspinatus tendon, the subscapularis tendon anteriorly, the infraspinatus and teres minor posteriorly, the AC joint superiorly, and the axillary recess and inferior glenohumeral ligament inferiorly. This typically requires 20–28 slices at 3 mm thickness.
Phase encoding direction: for oblique coronal, set phase encoding anterior-posterior (A-P). This displaces any motion artefacts (from breathing, swallowing, vascular pulsation) in the A-P direction, away from the rotator cuff tendon which lies in the superior-inferior direction. Setting phase encoding S-I in this plane would propagate artefacts through the supraspinatus tendon.
Common errors: (i) angulation based on the body coronal rather than the tendon axis — produces oblique sections through the supraspinatus, reducing tendon visibility; (ii) insufficient posterior coverage — the infraspinatus and teres minor are missed; (iii) insufficient inferior coverage — the axillary recess and inferior capsule are not included.
Oblique Sagittal Slice Prescription
Reference: oblique coronal scout (the newly planned oblique coronal) or the axial scout.
Alignment: draw the prescription line perpendicular to the oblique coronal plane just prescribed, so that the sagittal slices are truly orthogonal to the coronal. The resulting plane runs approximately parallel to the glenoid face.
Coverage: must include the full width of the glenohumeral joint from beyond the medial acromion edge to the lateral cortex of the humeral head; includes the AC joint; extends inferiorly to the axillary recess.
Phase encoding direction: for oblique sagittal, set phase encoding superior-inferior (S-I). This displaces motion artefacts cranio-caudally, away from the supraspinatus muscle cross-section and glenoid face.
Key structures visible: rotator cuff cross-sections in true cross-section at the glenoid level; the supraspinatus outlet (space between the acromion and the superior rotator cuff); muscle belly atrophy and fatty replacement; coracohumeral ligament; long head of the biceps in sagittal profile.
Axial Slice Prescription
Reference: oblique coronal or sagittal scout.
Alignment: the axial plane for shoulder MRI is not the true transverse plane of the body. The axial slices should be perpendicular to the long axis of the glenohumeral joint — that is, perpendicular to the glenoid face as seen on the coronal scout. This places the axial slices in a plane that displays the anterior and posterior labrum, the biceps tendon, and the subscapularis tendon simultaneously without obliquity.
Coverage: must include from the AC joint superiorly to the inferior glenoid rim inferiorly, covering the full height of the labrum. In most patients, 22–30 slices at 3 mm.
Phase encoding direction: for axial shoulder, R-L (right-left) phase encoding is preferred. The dominant motion sources in the shoulder region (breathing, vascular) are anterior-posterior; R-L phase encoding displaces their ghosts medially and laterally rather than through the joint space. Some centres use A-P phase encoding in the axial plane; both are acceptable provided the phase direction is noted in the protocol.
Anterior presaturation band: a presaturation band over the anterior chest and subcutaneous tissues suppresses pulsation artefacts from the subclavian artery on axial sequences; this is recommended when vascular motion artefact is noted.
Verification Before Scanning
On the three-plane localiser, confirm:
- Oblique coronal slices appear parallel to the supraspinatus tendon on the axial scout
- Oblique sagittal slices are perpendicular to the oblique coronal on the axial scout
- The AC joint is included in all three planes
- The axillary recess is included in the inferior extent of both coronal and axial slice packages
- The long head of the biceps tendon in the bicipital groove is covered on axial slices
[Slice Positioning References]
Stoller DW, Tirman PFJ, Bredella MA. Diagnostic Imaging: Orthopaedics. Amirsys/Elsevier. 2004. (Technical / Foundational) — Foundational positioning reference for oblique coronal prescription parallel to supraspinatus tendon; standard reference for shoulder MRI slice prescription technique.
ACR–SPR–SSR Practice Parameter for the Performance and Interpretation of Magnetic Resonance Imaging (MRI) of the Shoulder. American College of Radiology. Revised 2023. Available at gravitas.acr.org. (High — Society guideline) — Defines the three standard planes (transverse, oblique sagittal, oblique coronal), prescribing from glenoid fossa or supraspinatus tendon as reference landmarks; arm position guidance.
Chhabra A, Soldatos T, Subhawong TK, et al. The application of three-dimensional isotropic resolution SPACE MR imaging in musculoskeletal practice. J Magn Reson Imaging. 2012;35(2):299–309. PMID: 21994108. DOI: 10.1002/jmri.22825. (Moderate — Technical paper) — Documents MPR planning from 3D isotropic acquisitions; directly applicable to 3D shoulder MRI slice positioning.
5. Optimisation Strategy
5.1 Artifact Reduction by Source
Magic angle artefact is the most important and most misunderstood artefact in shoulder MRI. It occurs when collagen fibres (tendons) run at approximately 55° to the main magnetic field (B0). At this angle, the dipole-dipole relaxation mechanism that normally causes T1 and T2 shortening in fibrous tissue is partially cancelled, producing apparent T1 and PD signal increase in the tendon that can simulate a partial-thickness or intrasubstance tear. It is particularly problematic for the supraspinatus tendon at its critical zone (1 cm proximal to the insertion), which is naturally oriented at approximately 55° to B0 in the standard oblique coronal plane.
Magic angle effect is TE-dependent: it is most prominent at short TE (T1, PD, sequences with TE < 40 ms) and disappears at longer TE (T2-weighted sequences, TE > 60 ms). The practical consequence: apparent signal in the supraspinatus tendon on PD-FS must always be correlated with the T1 coronal and, if necessary, a T2-weighted sequence. Signal increase present on both PD and T1 but absent on T2-weighted images is likely magic angle rather than tear. A true tear shows signal on all fluid-sensitive sequences and on corresponding planes.
Chemical shift artefact at the bone-fat interfaces (greater tuberosity, humeral head, acromion) produces apparent high signal at the fat side of the interface and a signal void on the opposite side. This is most visible on coronal PD-FS and can simulate subchondral changes or periosteal pathology. Wider receiver bandwidth reduces the degree of displacement; at 1.5T, 130–200 Hz/px; at 3T, 200–400 Hz/px.
Susceptibility artefact from metallic hardware is common in post-operative shoulders. Suture anchors, synthetic ligament grafts, and arthroplasty components produce signal voids and geometric distortion. The extent depends on material composition. Titanium anchors produce less artefact than stainless steel. MARS sequences (wider bandwidth, STIR rather than spectral FS, reduced TE) reduce but do not eliminate susceptibility effects. Prior radiographs or CT to characterise the hardware are essential before MRI planning.
Motion artefact is the primary cause of non-diagnostic shoulder MRI in clinical practice. The shoulder joint is mobile; patients who are in pain or anxious frequently make small involuntary movements during acquisition. Motion degrades coronal PD-FS — the longest acquisition and the most diagnostically critical. Practical measures: (i) adequate patient explanation; (ii) padding to minimise discomfort; (iii) using the most critical sequence (coronal PD-FS) as the first or second sequence after the localiser, when the patient is freshest; (iv) considering short acquisitions for claustrophobic patients.
B0 inhomogeneity and failed fat suppression at the shoulder are more problematic than in most body regions because the combination of air-tissue interfaces (axilla), bone-air interfaces (AC joint region), and the curved morphology of the shoulder creates local field perturbations. The result is characteristically patchy fat suppression — particularly laterally (deltoid-humeral interface) and at the AC joint. Recognition is important: areas of incomplete fat suppression appear relatively bright and can simulate oedema or pathology. Comparison with non-fat-suppressed sequences (T1) resolves most cases. STIR or Dixon overcomes B0-dependent failures.
Magic angle at the biceps tendon in the bicipital groove: the long head biceps tendon also runs at a variable angle relative to B0 as it courses through the groove, creating magic angle susceptibility at the level of the superior groove and intratubercular segment. This should not be interpreted as tendinopathy in the absence of tendon thickening, surrounding fluid, and corroborating axial findings.
5.2 Protocol Efficiency and Throughput
A full diagnostic shoulder MRI protocol at 3T can be completed in 20–30 minutes total table time. At 1.5T, the longer TR/TE requirements of fat-suppressed sequences extend acquisition time to 30–45 minutes.
Short protocol for claustrophobic patients, limited cooperation, or high-throughput screening:
A minimum viable protocol consists of oblique coronal PD-FS + T1 coronal + axial PD-FS. This three-sequence protocol covers the primary rotator cuff and labral pathology targets in approximately 12–15 minutes at 3T. Sagittal coverage can be extracted from a 3D isotropic acquisition if time allows.
3D isotropic sequences: the 3D PD-FS TSE (SPACE/VISTA/CUBE) acquired in the oblique coronal plane provides full MPR capability — axial, sagittal, and any oblique plane from a single acquisition. A single 3D acquisition of 5–8 minutes replaces two or three separate 2D acquisitions, reducing total protocol time. The compromise is that 3D TSE rotator cuff assessment shows comparable sensitivity to 2D for supraspinatus and infraspinatus full-thickness tears, but reduced specificity for subscapularis tendon tears [8]. For most clinical applications, a hybrid approach (3D coronal + 2D axial) is optimal.
2D sequences remain more robust for slice-specific pathology requiring guaranteed spatial resolution in a specific plane. For equivocal findings on 3D, targeted high-resolution 2D in the optimal plane provides a diagnostic backup.
5.3 Field Strength Considerations
3T advantages: higher SNR translates directly into improved in-plane resolution (smaller voxels achievable) — the critical requirement for small tears, partial-thickness characterisation, and labral pathology. Non-arthrographic sensitivity for full-thickness supraspinatus tears approaches 98–100% at 3T in experienced centres, substantially above 1.5T performance [4]. Shorter acquisition times are achievable because higher SNR can be traded for faster imaging.
3T disadvantages: magic angle artefact is more prominent at 3T because the longer TE required to eliminate it conflicts with the standard PD weighting preference. Chemical shift artefact (frequency-encoding direction) is doubled at 3T relative to 1.5T for the same bandwidth, requiring wider bandwidth. SAR limitations at 3T require reduced flip angle or extended TR compared to 1.5T for equivalent TSE sequences. B0 inhomogeneity effects are amplified, making spectral fat suppression less uniform.
1.5T: remains a clinically adequate platform for the large majority of shoulder MRI indications. Full-thickness rotator cuff tear assessment, bursal pathology, bone marrow screening, and most labral pathology (with appropriate protocols) achieve diagnostically acceptable sensitivity and specificity. Where 1.5T falls short is in partial-thickness tear characterisation and subtle labral changes where the higher SNR of 3T provides a meaningful advantage.
Practical department choice: for dedicated MSK units performing high-volume shoulder MRI, 3T is preferred. For general radiology departments where shoulder MRI is one of many indications, 1.5T with a well-optimised protocol produces the large majority of clinically required information.
6. Contrast Use Principles Specific to Shoulder MRI
Universal GBCA safety content applies — see general contrast administration page. The following addresses shoulder-specific contrast indications.
6.1 Non-Contrast Standard Protocol — Sufficient For
Non-contrast shoulder MRI is sufficient for:
- Full-thickness rotator cuff tear assessment and staging (atrophy, retraction, fatty infiltration)
- Subacromial-subdeltoid bursitis
- Calcific tendinopathy
- Tendinopathy without tear suspicion
- Osteoarthritis and AC joint assessment
- Adhesive capsulitis
- Acromial morphology and supraspinatus outlet assessment
- Bone marrow screening (occult fracture, stress reaction, avascular necrosis)
- Routine pre-operative rotator cuff repair assessment
6.2 Gadolinium Indicated — Region-Specific Contexts
Direct MR arthrography (intra-articular gadolinium) is a separate protocol, not an addition to the standard protocol. It is the reference standard for labral tear and instability assessment. The ACR criteria and current systematic review data support direct MRA as the most accurate technique for Bankart lesions (sens 94%), SLAP lesions (sens 86%), and full-thickness rotator cuff tears (sens 97%) [6]. This is a dedicated child protocol and is not developed further here.
Intravenous GBCA (indirect arthrography) may be used as a compromise between direct MRA and non-contrast MRI in patients who require some joint distension assessment but cannot or will not undergo intra-articular injection. Evidence for indirect MRA in the shoulder is limited compared to direct MRA; its primary use is in specialist settings for labral assessment when direct MRA is declined.
Post-contrast T1-FS with intravenous gadolinium is specifically indicated for:
- Suspected inflammatory synovitis (rheumatoid arthritis, spondyloarthropathy) — enhancement pattern characterises synovitis activity
- Suspected rotator cuff or periarticular infection
- Post-operative assessment for rotator cuff repair integrity when clinical question requires enhancement characterisation
- Suspected soft tissue or bone tumour involving the shoulder girdle
- Suspected parsonage-turner syndrome / neuralgic amyotrophy — enhancement of denervated muscle is an early diagnostic feature
6.3 Post-Contrast Acquisition Timing
When post-contrast T1-FS is acquired for inflammatory assessment, imaging should begin 3–5 minutes after injection to allow adequate synovial enhancement while the joint fluid signal remains relatively lower than the enhancing synovium. Earlier imaging (< 2 minutes) may show incomplete synovial enhancement; later imaging (> 10 minutes) allows further gadolinium diffusion into the joint fluid, reducing the synovium-fluid contrast.
STIR is absolutely contraindicated post-gadolinium — the same rule applies to shoulder MRI as to all other anatomical regions. STIR acquired after gadolinium injection will suppress gadolinium-enhanced tissues at the null point, producing false-negative images for synovitis and enhancement. STIR must always be acquired before gadolinium injection.
7. Reporting Essentials
7.1 Interpretation Framework
The shoulder MRI report should systematically assess five anatomical compartments: the rotator cuff, the glenoid labrum and capsuloligamentous complex, the articular surfaces and cartilage, the periarticular soft tissues and bursae, and the osseous structures. Isolated compartment interpretation — reading only the rotator cuff without reviewing the labrum and bursa — is a common cause of missed diagnoses.
The broad diagnostic axes for shoulder MRI interpretation are:
Degenerative / chronic: the dominant pathophysiology in patients over 40 years, where tendinopathy, cuff thinning, osteophytosis, and cartilage loss coexist and require integrated assessment. The cardinal question is whether observed pathology is mechanically significant and correlates with the clinical presentation.
Traumatic / acute: in younger patients and after defined injury, the focus shifts to structural disruption — acute full-thickness tears, labral injuries, bone marrow contusions, and ligamentous injuries.
Inflammatory: synovitis, erosions, marrow oedema, and enhancing pannus require systematic assessment of all joint surfaces; contrast is indicated when this is the primary clinical question.
Neuropathic: muscle denervation oedema (T2 bright, swollen muscle belly without tendon tear) as in quadrilateral space syndrome, suprascapular nerve entrapment, or Parsonage-Turner syndrome has a characteristic pattern — affected muscle distribution corresponds to the innervation territory rather than the tendon anatomy.
7.2 Mandatory Reporting Checklist
Rotator cuff tendons (each individually):
- Supraspinatus: intact / tendinopathy / partial-thickness tear (articular, bursal, or interstitial) / full-thickness tear; if tear: dimensions (anteroposterior, mediolateral width), retraction, shape
- Infraspinatus: intact / tendinopathy / tear
- Subscapularis: intact / tendinopathy / partial superior fibre tear / full-thickness tear
- Teres minor: intact / pathology
Rotator cuff muscles (Goutallier grading on T1 sagittal):
- Supraspinatus: Grade 0–4
- Infraspinatus: Grade 0–4
- Subscapularis: Grade 0–4
- Atrophy index if relevant
Glenoid labrum: anterior / posterior / superior / inferior: intact / signal change / tear / avulsion; Hill-Sachs lesion (size, location); bony Bankart lesion
Biceps tendon: long head at origin, within the groove: intact / tendinopathy / partial tear / full rupture / subluxation from groove
Glenohumeral cartilage: integrity, focal defects with size and depth if present
AC joint: arthrosis grade, osteophyte direction, distal clavicle signal, type 2 vs type 3 acromion
Subacromial-subdeltoid bursa: normal (minimal fluid) / bursitis / communication with cuff tear (indicating full-thickness tear)
Bone marrow: humeral head, greater and lesser tuberosity, glenoid: normal / oedema (marrow oedema pattern) / cystic change / erosion / AVN signs
Coracoclavicular and coracoacromial ligaments: when relevant
Report limitation clause: state whether contrast was used, whether positioning was optimal, and any technical limitations (metal artefact, motion, incomplete fat suppression) that affect diagnostic confidence.
7.3 Structured Reporting
Reports should include: Indication (clinical question stated explicitly); Technique (field strength, sequences acquired, planes, contrast use); Comparison (prior imaging studies and dates); Findings (systematic compartment-by-compartment); Impression (integrated summary of clinically relevant findings referenced to the clinical question); Limitations (technical constraints); Critical communication if required (unsuspected neoplasm, acute infection).
7.4 Incidental Findings — Clinical Decision Framework
Usually benign, no follow-up required: small joint effusion without synovitis (physiological); minor tendinopathy in asymptomatic shoulder; mild AC joint arthrosis; subcortical cysts at the greater tuberosity (degenerative); small acromiohumeral bursitis in elderly patients; acromial type variations.
May require clinical correlation or follow-up: os acromiale (clinical significance depends on symptoms and size — direct communication to the orthopaedic surgeon is appropriate if repair is being considered); significant Goutallier Grade 3–4 fatty infiltration (implies poor prognosis for rotator cuff repair even if the clinical question was not surgical planning — should be stated explicitly); Hill-Sachs lesions (dimensions should be reported to assess engagement risk if instability is clinically suspected).
Require urgent communication: unsuspected soft tissue mass suggestive of sarcoma; bone lesion with aggressive features; septic arthritis (joint fluid with synovial thickening, marrow oedema); avascular necrosis beyond Stage I; subclavian or axillary vessel pathology.
8. MRI Technologist Pearls
8.1 Sequence Order Logic
The recommended sequence order for a standard non-contrast shoulder MRI is:
- Three-plane localiser (scout)
- Oblique coronal PD-FS ← most diagnostically critical; acquire first while patient is freshest
- Oblique coronal T1 ← acquired from same prescription; quick to set up
- Axial PD-FS
- Oblique sagittal PD-FS
- STIR (coronal oblique or axial)
Rationale: the coronal PD-FS is the primary diagnostic sequence. Motion from patient discomfort or anxiety is least in the early part of the examination. Acquiring it first ensures the most important sequence is obtained at optimal quality. The T1 coronal uses the same prescription and is fast to acquire immediately after. The axial sequences are acquired mid-examination when the patient is settled. STIR is last because it has the longest acquisition time and is the most motion-sensitive due to its long TR.
If contrast is to be used: STIR and T1 pre-contrast are acquired before injection; post-contrast T1-FS sequences are acquired 3–5 minutes after injection.
8.2 Positioning Tricks
- For patients with severe pain who cannot fully extend the elbow beside the body: place the arm slightly in front of the body, using a foam wedge between the arm and the lateral torso. This is preferable to slight external rotation loss.
- For very large patients where the shoulder cannot reach isocentre: a 10–15 cm lateral patient offset toward the affected shoulder is acceptable with a surface coil. Verify on localiser that the glenohumeral joint is within the coil sensitivity region.
- For bilateral shoulder assessment: image the more symptomatic side first. Patient repositioning between sides requires resetting the coil and replanning; allow 3–5 minutes between acquisitions.
- For patients with a large deltoid muscle mass (athletes): the phase FOV in the axial plane may need to be extended or the phase encoding direction switched to avoid wrap from the deltoid.
8.3 Fast Salvage Protocol
| Priority | Sequence | Approx. time (3T) | What it covers |
|---|---|---|---|
| 1 | Oblique coronal PD-FS | 4–5 min | Rotator cuff tears, bursitis, bone marrow |
| 2 | Axial PD-FS | 3–4 min | Labrum, biceps, subscapularis |
| 3 | Oblique coronal T1 | 2–3 min | Fatty infiltration, anatomy, T1 characterisation |
Three sequences in approximately 10 minutes provide the minimum clinically interpretable shoulder MRI. A radiologist can reliably assess full-thickness cuff tears and major labral pathology from this abbreviated set.
8.4 Common Avoidable Errors
| Error | Consequence | Prevention |
|---|---|---|
| Oblique coronal not parallel to supraspinatus | Tendon appears oblique; spurious signal; inaccurate tear characterisation | Plan from axial localiser; verify tendon parallel on axial scout |
| Internal rotation of arm | Supraspinatus overlap; anterior labrum suboptimally displayed | Check thumb position at start; pad to maintain neutral/slight ER |
| Insufficient inferior coverage (axillary recess excluded) | Inferior capsular pathology and axillary recess missed | Extend slice package 1–2 cm below inferior glenoid on coronal plane |
| STIR acquired after gadolinium injection | Enhancement suppressed; false-negative for synovitis | Always complete STIR before injection; check sequence order |
| Incorrect FOV with phase wrap | Contralateral shoulder or body wall aliased through joint | Use phase oversampling or R-L phase encoding in axial |
| Coronal and sagittal not orthogonal | Cross-referencing between planes unreliable | Always plan sagittal perpendicular to the prescribed coronal |
| Magic angle misread as partial tear on PD only | False-positive partial tear report | Always correlate PD signal with T2-weighted sequence; T2 signal disappears in magic angle |
9. Quality Control Checklist
Before ending the examination, verify each item:
- [ ] Glenohumeral joint visible in all three planes on localiser
- [ ] Oblique coronal parallel to supraspinatus tendon (check axial scout)
- [ ] Oblique sagittal perpendicular to oblique coronal (check axial or coronal scout)
- [ ] AC joint included in coronal and sagittal
- [ ] Axillary recess included in inferior extent of coronal and axial
- [ ] Bicipital groove fully covered on axial slices (superior to inferior extent)
- [ ] Fat suppression uniform on all PD-FS sequences — no large regions of suppression failure
- [ ] STIR acquired before gadolinium (if contrast used)
- [ ] Post-contrast sequences acquired ≥ 3 minutes after injection
- [ ] No significant motion artefact on coronal PD-FS
- [ ] Metal artefact assessment (if hardware present): MARS sequences completed if indicated
- [ ] Correct laterality documented in sequence labels
- [ ] T1 coronal acquired without fat suppression
- [ ] All six sequences (or minimum salvage three) completed
- [ ] Patient and coil have not shifted between acquisitions (compare isocentre landmarks)
10. Advanced Technical Parameters
Technical supplement — click to expand / collapse
This section is intended for MRI technologists, protocol optimisation specialists, and advanced technical review.
10.1 Oblique Coronal PD-Weighted TSE with Fat Suppression
Tissue Contrast Logic
Proton density weighting is achieved by long TR (≥ 2500–4000 ms, allowing full T1 recovery, eliminating T1 contrast) and short TE (20–40 ms, minimising T2 contrast). The resulting image shows tissue contrast driven predominantly by the proton density (water content) of each tissue type. Tendons (highly ordered collagen, low free water) appear low signal. Joint fluid and bursal fluid (high free water) appear moderately bright. Fat appears moderately bright on non-fat-suppressed PD. With fat suppression, the fat signal is eliminated, dramatically increasing the contrast between fluid-filled structures and the surrounding soft tissues.
The rationale for PD rather than T2 weighting for rotator cuff sequences: at TE 20–40 ms, the magic angle effect in collagen fibres oriented at 55° to B0 is partially present but at a lower magnitude than at T1 TE (< 15 ms). At T2 TE (60–80 ms), magic angle is largely suppressed but SNR is significantly reduced and ETL must be extended, increasing blurring. PD represents the optimal compromise between magic angle suppression and SNR for tendon imaging.
Key Parameters
| Parameter | 1.5T | 3T | Rationale |
|---|---|---|---|
| Sequence type | 2D TSE-PD | 2D TSE-PD | Standard |
| TR | 2500–4000 ms | 2500–3500 ms | Long TR for PD weighting |
| TE | 25–40 ms | 20–35 ms | Short TE minimises T2 contrast; shorter at 3T due to tissue T2 |
| ETL | 7–12 | 6–10 | Moderate ETL; balance between speed and blurring |
| Slice thickness | 3–4 mm | 3 mm | No gap preferred; 0 mm interslice gap |
| Gap | 0 mm | 0 mm | |
| FOV | 140–180 mm | 130–160 mm | Small FOV for high in-plane resolution |
| Target in-plane resolution | ≤ 0.5 × 0.5 mm | ≤ 0.4 × 0.4 mm | Critical for rotator cuff tear characterisation |
| Fat suppression | SPAIR or Dixon | SPAIR or Dixon | B0-robust preferred; CHESS acceptable at 1.5T uniform B0 |
| Phase encoding | A-P | A-P | Displaces motion artefacts away from cuff |
Diagnostic Advantages
Full-thickness supraspinatus tears: sensitivity 84–91%, specificity 97% on non-arthrographic 1.5T MRI, improving to near 98–100% at 3T [4, 5]. Bursal fluid, tendon retraction, and cuff atrophy are well demonstrated on this sequence.
Limitations
Magic angle artefact at the critical zone of the supraspinatus remains the primary limitation. Partial-thickness tears may be overdiagnosed (magic angle) or underdiagnosed (isointense to tendon) at standard resolution. 3 mm slice thickness limits detection of very small partial tears confined to a single articular surface fibre layer.
Common Artefacts
Magic angle (see 5.1); B0 inhomogeneity — patches of incomplete fat suppression; chemical shift at bone-fat interfaces; motion from patient discomfort.
Fat Suppression
Mandatory. SPAIR (spectral adiabatic inversion recovery) preferred at both field strengths for its improved B0 robustness over CHESS. Dixon available on most modern platforms provides both water-only and fat-only images with the best B0 independence.
Black-Blood Pulse, MTC
Not applied in standard shoulder sequences. No clinical role established for these modifiers in rotator cuff assessment.
10.2 Oblique Coronal T1-Weighted TSE (Non-Fat-Suppressed)
Tissue Contrast Logic
Short TR (400–700 ms) produces T1 weighting: tissues with shorter T1 (fat, subacute haemorrhage, proteinaceous fluid, marrow fat) appear bright; muscle and tendon are intermediate; fluid appears dark. This sequence is the essential complement to fat-suppressed PD: any structure that appears T1-bright despite fat suppression on PD cannot be fat — it represents proteinaceous fluid, haemorrhage, or gadolinium enhancement.
Key Parameters
| Parameter | 1.5T | 3T | Rationale |
|---|---|---|---|
| Sequence type | 2D TSE-T1 | 2D TSE-T1 | |
| TR | 500–700 ms | 550–800 ms | T1 weighting; longer at 3T |
| TE | 10–18 ms | 8–15 ms | Minimum TE |
| ETL | 2–5 | 2–4 | Short ETL critical to preserve T1 contrast |
| Slice thickness | 3–4 mm | 3 mm | Same geometry as coronal PD-FS |
| Gap | 0 mm | 0 mm | |
| FOV | Same as coronal PD-FS | Same | Copy geometry |
| Target in-plane resolution | ≤ 0.5 × 0.5 mm | ≤ 0.4 × 0.4 mm | Match coronal PD-FS |
| Fat suppression | None | None | Critical: absence of FS is the diagnostic purpose |
Diagnostic Advantages
Goutallier fatty infiltration grading (Grade 0: no fat; 1: mild streaks; 2: fat < muscle; 3: equal fat and muscle; 4: fat > muscle) is the primary clinical application. Cortical bone definition, periosteal reaction, and calcium deposits (signal void) are well shown. T1-bright lesions (lipoma, cholesterol crystals, fat-containing lesion) are characterised.
10.3 Axial PD-Weighted TSE with Fat Suppression
Tissue Contrast Logic
Same PD-FS physics as Section 10.1. The axial plane places the anterior and posterior labrum, the biceps tendon, and the subscapularis tendon in cross-section — the optimal geometry for these structures.
Key Parameters
| Parameter | 1.5T | 3T | Rationale |
|---|---|---|---|
| TR | 2500–4000 ms | 2500–3500 ms | PD weighting |
| TE | 25–40 ms | 20–35 ms | Short TE |
| ETL | 7–12 | 6–10 | |
| Slice thickness | 3–4 mm | 3 mm | |
| Gap | 0 mm | 0 mm | |
| FOV | 140–160 mm | 130–150 mm | Tight FOV for labral resolution |
| Target in-plane resolution | ≤ 0.4 × 0.4 mm | ≤ 0.3 × 0.4 mm | Labral detail requires highest in-plane resolution in protocol |
| Phase encoding | R-L | R-L | Displaces motion artefacts mediolaterally |
Diagnostic Advantages
Anterior labral tears: non-arthrographic sensitivity approximately 60–70% — significantly lower than MRA (91%) but adequate for full-thickness anterior labral disruptions. Posterior labral tears, SLAP lesions, subscapularis tears, and biceps tendon subluxation are assessed on axial.
10.4 STIR (Short Tau Inversion Recovery)
Tissue Contrast Logic
STIR nulls the fat signal at its T1 null point (TI ≈ 150–175 ms at 1.5T; TI ≈ 200–230 ms at 3T) through an inversion recovery preparation pulse. The suppression is T1-based and therefore independent of B0 homogeneity — the fundamental advantage over spectral fat saturation methods. Fluid, oedema, and inflammatory tissue appear bright against a suppressed fat background.
Key Parameters
| Parameter | 1.5T | 3T | Rationale |
|---|---|---|---|
| TR | ≥ 3000–5000 ms | ≥ 3000–5000 ms | Must be long (≥ 5× TI) for complete inversion |
| TE | 50–80 ms | 40–70 ms | T2 weighting |
| TI | 150–175 ms | 200–230 ms | Fat null point; must be recalibrated at different field strengths |
| Slice thickness | 3–4 mm | 3 mm | |
| Target in-plane resolution | ≤ 0.6 × 0.6 mm | ≤ 0.5 × 0.5 mm | Lower SNR than PD-FS; slight resolution reduction acceptable |
STIR contraindicated post-gadolinium — identical rule as in all other protocols: gadolinium shortens the T1 of enhancing tissues toward the fat null point, suppressing rather than highlighting pathology.
10.5 3D Isotropic PD-FS TSE (SPACE/VISTA/CUBE) — Conditional
Tissue Contrast Logic and Acquisition Design
3D TSE with variable flip angle readout (SPACE: Siemens; VISTA: Philips; CUBE: GE; Multi-shot TSE with CS: Canon) acquires an isotropic voxel volume (typically 0.5–0.8 mm isotropic) in a single oblique coronal acquisition that can be reformatted in any plane post-acquisition. A single 3D acquisition of 5–8 minutes replaces separate coronal, sagittal, and potentially axial 2D acquisitions.
Diagnostic Accuracy vs. 2D
3D isotropic non-arthrographic sequences show comparable sensitivity to 2D for supraspinatus and infraspinatus full-thickness tears (3D sensitivity 95%, 2D sensitivity 99% in the most cited comparative study [8]). However, subscapularis tear assessment is substantially less reliable on 3D isotropic (sensitivity 87% on 3D vs 88% on 2D, but specificity 49% on 3D vs 66% on 2D) [8]. For standard clinical use, a hybrid approach is recommended: 3D oblique coronal for supraspinatus and infraspinatus + dedicated 2D axial for subscapularis and labrum.
Key Parameters
| Parameter | 1.5T | 3T |
|---|---|---|
| Target voxel size | 0.7–0.9 mm isotropic | 0.5–0.7 mm isotropic |
| TE effective | 30–45 ms | 25–40 ms |
| TR | 1500–2500 ms | 1200–2000 ms |
| ETL | 50–100 (variable FA) | 50–100 |
| Fat suppression | Dixon preferred; SPAIR acceptable | Dixon preferred |
[1] American College of Radiology. ACR Appropriateness Criteria® Shoulder Pain — Atraumatic. J Am Coll Radiol. 2018;15(11S):S388–S402. PMID: 30392607. (High — Society guideline) Designates MRI as the most comprehensive modality for atraumatic shoulder pain; defines when MRI without contrast vs MR arthrography vs ultrasound is appropriate.
[2] Lenza M, Buchbinder R, Takagaki Yhon P, Belloti JC, Faloppa F. Magnetic resonance imaging, magnetic resonance arthrography and ultrasonography for assessing rotator cuff disease of the shoulder: a systematic review. Musculoskeletal Care. 2013;11(4):182–196. PMID: 23413143. DOI: 10.1002/msc.1044. (High — Systematic review) Pooled sensitivity/specificity for full-thickness (91%/97%) and partial-thickness (80%/95%) rotator cuff tears on MRI; foundational reference for diagnostic performance data.
[4] Magee T, Williams D. 3.0-T MRI of the supraspinatus tendon. AJR Am J Roentgenol. 2006;187(4):881–886. PMID: 16985127. DOI: 10.2214/AJR.05.0553. (Moderate — Original study) Documents sensitivity 98% for full-thickness and 92% for partial-thickness supraspinatus tears at non-arthrographic 3T; key evidence for 3T superiority.
[5] Samim M, Bhate S, Gyftopoulos S. MRI of the Rotator Cuff. Radiol Clin North Am. 2023;61(2):219–234. DOI: 10.1016/j.rcl.2022.10.005. (Moderate — Review) Current clinical review of rotator cuff MRI including modern fat suppression strategies, 3T protocols, and diagnostic criteria; directly relevant to parameter design.
[8] Hong WS, Jee WH, Lee SY, et al. Diagnosis of Rotator Cuff Tears with Non-Arthrographic MR Imaging: 3D Fat-Suppressed Isotropic Intermediate-Weighted Turbo Spin-Echo Sequence versus Conventional 2D Sequences at 3T. Investig Magn Reson Imaging. 2018;22(4):229–239. DOI: 10.13104/imri.2018.22.4.229. (Moderate — Original comparative study) Documents 3D SPACE vs 2D accuracy for SST/IST and SCT tears at 3T; key evidence for subscapularis limitation of 3D isotropic.
[9] Subhas N, Sakamoto FA, Mariscalco MW, et al. Accuracy of MRI in the diagnosis of shoulder tears with arthroscopy as reference standard. AJR Am J Roentgenol. 2012;198(4):W360–W369. PMID: 22451575. DOI: 10.2214/AJR.11.7097. (Moderate — Prospective study) Arthroscopy-confirmed accuracy of non-arthrographic MRI for labral and rotator cuff tears; documents sensitivity limitations for partial tears and labral lesions.
[10] Razek AAKA, Rashed YM, Kamal S. Articular cartilage: An updated imaging review. J Comput Assist Tomogr. 2021;45(2):275–281. PMID: 33002944. DOI: 10.1097/RCT.0000000000001123. (Moderate — Review) Covers MRI cartilage assessment; relevant to shoulder cartilage sequence optimisation.
[11] Kijowski R, Blankenbaker DG, Stanton PT, et al. Radiologic Association of Musculoskeletal Imaging 3-Tesla MR Imaging of the Shoulder. Radiology. 2007;244(2):504–515. PMID: 17577994. DOI: 10.1148/radiol.2442060530. (Moderate — Original study) Technical parameter documentation for 3T shoulder MRI; documents SNR gains and artefact behaviour at 3T vs 1.5T.
11. Evidence Gaps and Ongoing Debate
Non-arthrographic vs MR arthrography at 3T: the question of whether 3T non-arthrographic MRI has sufficiently closed the diagnostic gap with direct MRA for labral tear detection remains unresolved. Current meta-analytic evidence consistently shows MRA superiority for anterior labral tears and SLAP lesions, but several prospective 3T studies suggest that for specific populations (first-time dislocators, non-athletes), non-arthrographic 3T MRI may provide sufficient information for initial management decisions. Definitive prospective comparative trials are lacking.
2D vs 3D isotropic protocols: while 3D isotropic sequences offer workflow efficiency and MPR capability, their specific limitations for subscapularis and labral assessment at non-arthrographic conditions have not been resolved in sufficiently powered prospective trials. Hybrid protocols (3D coronal + 2D axial) represent a pragmatic solution but standardisation is lacking.
AI-accelerated sequences: deep learning reconstruction applied to undersampled 3D MRI data at the shoulder has shown promising early results for both arthrographic and non-arthrographic examinations, with 2- to 4-fold acceleration maintaining diagnostic accuracy for full-thickness rotator cuff tears [see search results 2025/2026 data]. Validation across vendors, field strengths, and diverse patient populations is ongoing.
Abbreviated protocols for high-throughput screening: the minimum protocol configuration for reliable rotator cuff tear exclusion has not been formally validated in multicentre prospective trials.
Optimal fat suppression technique: while Dixon-based fat suppression provides the most robust and B0-independent fat suppression for shoulder MRI, its availability, optimal TE pairing for shoulder geometry, and comparative diagnostic accuracy to SPAIR in clinical practice have not been systematically studied in large prospective series.
Quantitative cartilage assessment: T2 mapping, T1-rho, and quantitative Dixon fat fraction for shoulder cartilage and rotator cuff muscle assessment are emerging techniques without established clinical thresholds or reproducibility data sufficient for routine use.
12. Evidence-Based References
A. Guidelines / Consensus / Society Recommendations
B. Systematic Reviews / Meta-analyses
C. Important Prospective / Original Studies
D. Technical MRI Papers
E. Landmark Historical References
End of document — MRI Shoulder Generic Standard Protocol — MRIninja v1.0 — May 2026
Child Protocols
Clinical pages derived from this master protocol. These pages document what changes for specific indications.
No child protocols have been published yet.
Related Protocols
Recent PubMed search for this protocol