Back to products
In progress

RealActivity Imaging

Multi-cancer imaging AI platform supporting clinician assessment across six pathways, with per-slice explainability and structured FHIR R4 reporting.

Progress
Planning
Building
Testing
Launch
Screenshot
RealActivity Imaging dashboard showing a brain MRI series flagged Priority 1, glioma, Grade IV, 99.9% probability, with a per-slice concern chart showing which slice drove the verdict

Live capture — the app's own demo example (a synthetic Priority 1 glioma case), showing the per-slice concern chart and which slice actually drove the verdict

About this project

A radiologist reviewing a multi-slice MRI or CT series has to decide which image in the stack actually matters, then defend that call. RealActivity Imaging scores every slice in a series independently across six cancer pathways — brain (MRI), lung (CT), breast (mammography), liver (CT), skin (dermoscopy), and spine (MRI) — and reports which slice drove the verdict rather than assuming the geometric middle of the volume is the right one to show.

Every result returns a calibrated confidence score, ICD-10 and SNOMED coding, an occlusion-sensitivity attention map, and a TIxAI trustworthiness score measuring how well that attention overlaps the actual finding — independent of whether the verdict itself is correct. Clinicians sign in with their hospital's existing Microsoft Entra ID credentials under role-based access, every action is recorded to an audit log, and results export as FHIR R4 Diagnostic Reports with a SMART on FHIR launch flow tested against a public sandbox. It is a research platform, not a medical device — every output requires review by a qualified clinician before any diagnostic or treatment decision.

Built with RealActivity, LLC
Tech stack
Medical Imaging
DICOM
ONNX Runtime
FastAPI
FHIR R4
Python

§Contact

Interested in this project?

Reach out to talk about RealActivity Imaging.

tejansree1229@gmail.com