No critical blood smear
should wait for review.
To help pathologists prioritize blood smear slides through safe, non-diagnostic AI triage — so expertise is applied where it matters most.
How PathoIntern Works
A four-step pipeline that assists — never replaces — the pathologist's judgment.
Upload Patch
Lab technologist uploads a digitized blood smear patch image (PNG/JPG).
Generate Embedding
CTransPath encodes the patch into a 768-dimensional morphological feature vector.
Score & Prioritize
k-NN anomaly detection assigns a 0–100 criticality score and Urgent/High/Elevated/Routine tier.
Pathologist Reviews
Pathologist reviews the prioritized worklist, views similar cases, and submits a verdict.
Four-Tier Triage Classification
Scores represent pattern similarity to reviewed cases — not diagnostic probability.
Final interpretation always by licensed pathologist
Scientific Foundation
Four peer-reviewed publications support the problem statement and technical approach.
Khatab et al. (2024)
Critical Reviews in Clinical Laboratory Sciences
Pathologist workloads rising 5–10% annually. Burnout causes diagnostic errors. No AI triage tools currently exist for blood smear review.
Baltruschat et al. (2021)
European Radiology
AI-driven worklist reordering cut critical-case turnaround by 56% — without the AI making any diagnoses. Triage, not diagnosis.
Dippel et al. (2024)
NEJM AI
CTransPath + anomaly detection achieved 95% AUROC for histopathological triage — the exact approach PathoIntern implements.
Chen et al. (2024)
Nature Medicine
UNI foundation model embeddings generalize across 34 pathology tasks, validating embedding-based similarity search for Phase II.
Built for Pathologists. Controlled by Pathologists.
PathoIntern acts as a digital intern — it performs first-pass analysis and flags anomalies. Every decision, every override, every verdict is fully audited. The pathologist always has the final word.