Reading Room XR

A hands-first radiology reading room on Meta Quest: compare scans, window, zoom, and measure with your bare hands.

Comment

Inspiration

I run the PACS and teleradiology infrastructure for a statewide government hospital network in Bihar, India. Radiologists who read remotely often work from a single laptop screen, while a hospital reading room has several large diagnostic monitors. A headset can give anyone that multi-monitor reading room anywhere. Hand tracking makes it practical: radiologists already pinch, drag, and point all day, and they do not want to hold controllers.

What it does

  • Opens a seated reading room with three large monitors, a worklist, and a measurements panel, all within arm's reach.
  • Load studies from the worklist with a pinch, then scroll slices, adjust window/level, zoom, pan, probe Hounsfield units, and measure findings in millimetres.
  • PRIOR and CURRENT scans scroll in sync, so you can measure how a lung nodule has grown between visits (8.6 mm to 13.6 mm in the demo case).
  • Two-hand pinch zooms an image; pinching a panel's title bar moves the monitor anywhere around you; one tap brings the panels back.
  • Every control works with a pinch-ray or by simply tapping it with a fingertip. No controller is needed at any point.

How we built it

  • WebXR with three.js, so it runs in the Meta Quest Browser with no install and opens in seconds.
  • A unified input layer turns WebXR hand pinch-rays, fingertip pokes, and a desktop mouse into the same press/drag events.
  • Each viewport renders CT slices on a canvas with a real window/level transform from Hounsfield units, plus zoom, pan, overlays, and measurement lines.
  • To avoid any patient-data risk, all CT studies are synthetic phantoms generated in code: a chest CT at two time points with a growing nodule, and a head CT with an acute subdural pattern.
  • Seated-first design: local reference space, controls on a tilted desk toolbar.
  • The demo video was recorded in Meta's Immersive Web Emulator (IWER) as a Meta Quest 3 with hand tracking, with scripted hand movements; the narration is an AI voice.

Challenges we ran into

  • Precise image work by hand: mapping small hand movements to window/level and slice scrolling without jitter. Dragging is measured on the image plane, so you can move outside the image while adjusting.
  • Making buttons easy to hit both by pinch-ray and by direct touch, without double presses.
  • Producing realistic-looking CT data without using any real patient images.

Accomplishments that we're proud of

  • A full read, compare and measure workflow done entirely with hands, in under a minute from launch.
  • The PRIOR vs CURRENT comparison: the nodule growth is visible and measurable side by side.

What we learned

Hands are a natural fit for image review: pinch-and-drag for contrast feels closer to a workstation mouse than controllers do, and a spatial layout removes the need to tile windows.

What's next for Reading Room XR

  • Loading real DICOM files (local or from a PACS via DICOMweb) with proper de-identification.
  • Voice dictation of findings into a structured report.
  • Multi-user review, so a radiologist and a referring doctor can look at the same study together.
  • MPR (coronal and sagittal) and 3D volume views.

Built With

Share this project:

Updates

Submission history