About Analyze My Knee
Who runs this site
Analyze My Knee is built and operated by the small independent team behind JointLens, the umbrella product for our joint-imaging work. No hospital, insurer, clinic chain or scanner vendor stands behind it. The same people write the code, pick the models, pay the bills and read the messages you send. We state that openly because a website willing to comment on your knee scan owes you the name of whoever is doing the commenting.
What happens when you open a knee study here
You drop in a knee MRI series, a weight-bearing radiograph, or the entire folder from a hospital CD. The viewer decodes the pixel data in your browser and lets you scroll the stack, change window and level, and move between sagittal, coronal and axial planes. Ask for an explanation and several general-purpose AI models look at the rendered slices you selected, together with the short note you typed, and answer in ordinary language: whether the cruciate fibers appear continuous, whether a meniscal signal line reaches an articular surface, how much cartilage covers the femoral condyles, whether there is an effusion or marrow edema, and what a popliteal cyst would look like if one were there.
Where it falls short
It does not diagnose a torn ACL. It cannot say whether a meniscus should be repaired or trimmed, and it will not stage your arthritis for a surgical decision. Plain films carry almost no direct cartilage information, so any cartilage remark on a radiograph is an inference drawn from joint-space width rather than a measurement. Partial ligament tears, healed grafts and post-operative knees are precisely the situations where these models are least dependable. Nothing in the answer knows your Lachman test, your pain pattern, your prior films or your surgical history, and no radiologist checks the text before it reaches you.
The safety rails we keep in place
The analyzer stays locked behind a consent screen that states, in one page, that this is not a medical device and not a diagnosis. Every answer repeats the warning, and our Medical Disclaimer names hallucination, missed findings and wrong grading as expected behaviour rather than rare accidents. A knee that locks, gives way, or will not take your weight is a reason to see a clinician today — not a reason to read another AI paragraph.
Your scan stays on your machine
DICOM files, ZIP archives and PDF reports are parsed by code running inside your own browser tab. They are never uploaded, queued or written to our servers, and there is no copy of them once you close the tab. When you request an explanation, what leaves the browser is a rendered picture of the slices you chose plus the words you typed — not the original study and not the patient header. We keep no medical file store, which also means we cannot fetch your scan back for you later. That trade-off is deliberate; the details are in our Privacy Policy.
Talking to us
We publish no support inbox. Corrections, complaints, bug reports and feature requests all arrive through the form below, and they land with the same people who wrote this page.