A dual engine of deep-learning CNN × radiomics, designed around your existing PACS workflow — it doesn't change your habits, it adds a pair of eyes that never tire.
PANCREASaver® is a medical device software (SaMD): the moment an abdominal CT scan finishes, it automatically reads the images, marks suspicious pancreatic lesions, and sends the alert back into the radiology workflow — so early pancreatic cancer is no longer missed.
Analysis starts automatically when the scan completes — no extra steps for physicians; CNN deep learning and radiomics engines run in parallel.
92.1% sensitivity for lesions <2cm — closing the gap where eyes and conventional tools miss.
5 international journal papers and a nationwide multicenter AUC of 0.95, led by the NTUH team.
Analyzes during the CT acquisition; a full pancreatic read in minutes, catching sub-2cm lesions.
An orange reticle pinpoints lesion location and extent — visible at a glance.
High-risk studies are prioritized automatically; alerts trigger the care team sooner.
Auto-generates structured reads with location, size and risk — less reporting time.
No clicks required; the scan triggers the engine. Radiologists read, AI stands guard.
Built around the PACS workflow — deployed in 4 weeks without changing daily practice.
Patient undergoes abdominal CT; images stream in.
PANCREASaver® analyzes the full pancreas within minutes.
Lesion locked and marked; PACS alert fires.
The radiologist confirms the marks and signs the final report.
From Stone Web Viewer integration to Bridge login, PANCREASaver® is deployed at medical centers, regional hospitals and screening centers — 300+ scans reviewed.
PANCREASaver®