RadPhrases

Enables AI coding agents to perform end-to-end medical image analysis locally, from inspecting CT/MRI/X-ray studies and running containerized segmentation models like MedSAM2, TotalSegmentator, and HD-BET to refining masks, measuring volumes, and writing reports. It pairs preset or custom guideline protocols with a code-customizable viewer that renders images back to agents as annotated PNGs and to humans as interactive HTML.

Enables AI assistants to query, read, download, and move medical imaging data on DICOM servers (PACS, VNA) including patient searches, study retrieval, PDF report extraction, and image transfer to AI endpoints for analysis.

Analyzes mammogram CC/MLO views and sorts each breast into a BI-RADS group with a priority level (high_priority, review, or no_suspicious_findings), so physicians can see which women look most worrying and reach a radiologist first. It masks personal data before analysis, never stores images, and runs on a CPU in 1-3 seconds per image.

An MCP server that exposes a DICOMweb-compliant DICOM archive to AI assistants. It lets any MCP-capable client search studies, series and instances, inspect metadata, read Structured and Encapsulated PDF Reports, and render image frames — all through natural language.