DIEP Automated Perforator
Clinical AI Workflow
A surgical planning workflow for DIEP flap reconstruction at Tampa General Hospital. The project frames perforator selection as an auditable clinical AI pipeline: ingest imaging evidence, identify candidate vessels, rank perforators, and keep human review central to the final decision.
Role
AI Workflow Designer
Clinical Systems Researcher
Context
Tampa General Hospital DIEP workflow
Duration
Prototype
Tools
Computer Vision
Workflow Design
Clinical AI
Human Review
Planning Objective
DIEP reconstruction depends on selecting reliable perforators while preserving as much abdominal muscle as possible. The workflow is designed to reduce manual review time by surfacing candidate perforators, ranking them with transparent criteria, and keeping the evidence trail visible for surgeons and researchers.
System Workflow
Imaging intake: collect CTA or related scan-derived inputs and align them to the planning case.
Candidate detection: identify perforator regions and vessel paths that should be reviewed.
Ranking layer: score candidates using features such as location, continuity, vessel confidence, and planning relevance.
Review interface: present ranked candidates with supporting evidence so a clinician can confirm, reject, or adjust the suggested plan.
Why It Matters
The goal is not to replace clinical judgment. It is to make the repetitive parts of perforator discovery faster, more consistent, and easier to audit. Each recommendation should be traceable back to the image evidence and ranking signals that produced it.