The shoulder is the body's most mobile joint, which is exactly why it is also the hardest to interpret from a single image. Drop a shoulder MRI, X-ray, or CT series into our viewer and a four-model AI panel walks through the rotator cuff, labrum, and bones together, translating what it sees into language you can bring to an appointment.
Browse Shoulder FindingsMost shoulder workups begin with plain films before anything cross-sectional is ordered: a true AP view, a scapular Y view, and an axillary view together show alignment, joint space, and whether the humeral head sits centered in the glenoid. When a rotator cuff or labral problem is suspected, MRI or an MR arthrogram — where contrast is injected into the joint to outline the labrum — takes over, since fluid and cartilage detail simply do not show up on X-ray. Before a surgeon plans shoulder stabilization surgery for recurrent instability, a CT scan is often added specifically to measure glenoid bone loss, because the amount of bone missing from the socket changes which procedure is recommended.
Once a study is rendered in your browser, the consortium steps through the cuff tendons — supraspinatus, infraspinatus, subscapularis, and teres minor — checking for tendinosis, partial tearing, or a full-thickness gap with muscle retraction. It also traces the labrum around the glenoid rim, screens the acromioclavicular joint and subacromial space for impingement or bursitis, and flags joint-space narrowing or calcific deposits that point toward arthritis or calcific tendinitis. The output is a plain-language shoulder report: which tendon or structure looked abnormal, roughly how severe the finding appears, and which imaging planes support that read, written so a non-radiologist can follow it before a specialist appointment.
These are the questions people typing a shoulder MRI or X-ray report into a search bar usually want answered first, before they dig into a specific condition page.
No. Many labral tears, especially degenerative SLAP lesions in older shoulders, are managed with physical therapy first. Surgery is more often discussed after a traumatic dislocation, when instability keeps recurring, or when a tear involves the biceps anchor.
Bone alignment can look completely normal on X-ray while the rotator cuff, labrum, or capsule is inflamed or torn, since none of those soft tissues are visible on plain film. That gap is exactly why MRI gets ordered next when pain persists.
Adhesive capsulitis restricts motion in every direction, including when someone else moves the arm for you, because the joint capsule itself has thickened and tightened. A cuff tear more often limits active lifting while passive range stays close to normal.
No. The panel can describe what a scan shows, but the decision to operate weighs age, activity goals, tear size, and how a shoulder has responded to conservative care — judgments that belong to an orthopedic surgeon, not a model.
Bone and joint problems around the shoulder girdle each read differently depending on the modality that caught them. Pick the guide that matches the wording already sitting in your report.
Once a shoulder has dislocated, the labrum, the front of the socket, and the back of the humeral head all become suspects. These pages walk through the tear pattern and bone-loss findings that typically follow a dislocation.
Shoulder complaints rarely stand alone; they tend to sort into a handful of patterns once imaging is added to the story. Rotator cuff degeneration and tearing shows up as tendinosis that progresses to partial-thickness fraying and eventually a full-thickness gap, usually building slowly with age and overhead work rather than a single moment of injury. Impingement and subacromial bursitis sit next door, where a tight space under the acromion pinches the cuff and bursa during overhead reaching, producing pain that mimics a tear without one being present yet. Instability and labral injury follow a different path — a dislocation or repeated subluxation stretches or tears the labrum and capsule, leaving the joint loose even after the bone goes back into place. Stiffness and adhesive capsulitis form their own cluster, where the capsule itself thickens and every direction of motion tightens, often after a period of immobilization or in people managing diabetes. Glenohumeral and AC joint arthritis shows up as cartilage thinning, osteophytes, and joint-space narrowing that build over years. Acute fractures — clavicle, proximal humerus, scapula — sit apart from the rest because they follow a single traumatic event and are read on X-ray and CT rather than MRI. Knowing which cluster a report's wording belongs to makes the rest of the page far easier to navigate.
If a report already names a structure — supraspinatus, labrum, AC joint, glenoid — jump straight to the matching condition page above rather than reading every guide on this list. If the report is still pending and only symptoms are known, start with the viewer that matches the study you were given, upload it, and let the plain-language summary point you toward which cluster to read next.
Nothing on this page is a diagnosis, and it does not replace a shoulder surgeon or musculoskeletal radiologist reading your actual study alongside your exam and history. Imaging findings and symptoms do not always line up — plenty of shoulders with cuff fraying on MRI are painless, and plenty of painful shoulders scan clean.
Seek urgent, in-person care rather than reading further if the shoulder looks visibly deformed or is locked in one position after an injury, if you suddenly cannot lift or rotate the arm with any strength, if numbness or tingling spreads down the arm or into the hand, or if the joint is hot, swollen, and accompanied by fever — that combination can signal a joint infection and needs same-day evaluation.
Upload a shoulder MRI, X-ray, or CT and get a private, plain-language read.
Start AnalysisAcross 11 tracked shoulder conditions, the consortium looks for rotator cuff tendinosis and tearing, labral tears including SLAP and Bankart patterns, subacromial impingement, adhesive capsulitis, arthritis, and fracture. Four models cross-check each other's reads across the available imaging planes before a summary is produced, and that summary stays informational rather than diagnostic.
Reliability tracks scan quality, coil placement, and how classic the finding is — small partial-thickness cuff tears are harder to characterize than a full-thickness gap. Four independent models are synthesized rather than relying on one, which cuts down single-model blind spots, but an orthopedic surgeon or radiologist should still confirm anything that guides treatment.
No account is needed. Choose the DICOM files on your device, the browser parses them locally, and only the rendered frames needed for analysis are sent onward. The original DICOM data never leaves the machine you uploaded it from.
No. Treat the summary as preparation for that visit, not a substitute for it. Bringing a plain-language explanation of your MRI or X-ray to the appointment can make the conversation with your surgeon or physical therapist more focused, but the treatment decision stays theirs.
A dedicated shoulder protocol with oblique coronal, oblique sagittal, and axial proton-density fat-saturated sequences on a 1.5T or 3T magnet gives the panel the most to work with, and an MR arthrogram adds contrast detail for a suspected labral tear. Motion artifact or a worn-out coil noticeably lowers confidence.
Not reliably. The models were trained mainly on skeletally mature shoulders, and an open growth plate at the proximal humerus can look unfamiliar to them. Pediatric shoulder imaging needs review from someone experienced in growing bone.
Yes, though anchors, sutures, and post-surgical scarring make the read harder and lower the confidence score the panel assigns. Always route post-operative imaging back to the surgeon who performed the repair rather than relying on the summary alone.
Your raw files stay on your device the entire time. Only the rendered frames the AI models need travel over an encrypted connection, and those frames are discarded once the analysis finishes rather than kept on our servers.