Governance-Ready Small Language Models for Medical Imaging: Prompting, Abstention, and PACS Integration

TL;DR

This paper introduces a comprehensive framework for deploying small language models in medical imaging workflows, emphasizing governance, calibration, and integration with clinical standards to enable safe and effective utility.

Contribution

It presents a prompt-first deployment framework with calibration, oversight, and integration strategies tailored for medical imaging applications using small language models.

Findings

Reflection-oriented prompts improve lighter models' performance

Operational calibration reduces expected errors and oversight burden

Outputs are mapped to DICOM, HL7, and FHIR standards for integration

Abstract

Small Language Models (SLMs) are a practical option for narrow, workflow-relevant medical imaging utilities where privacy, latency, and cost dominate. We present a governance-ready recipe that combines prompt scaffolds, calibrated abstention, and standards-compliant integration into Picture Archiving and Communication Systems (PACS). Our focus is the assistive task of AP/PA view tagging for chest radiographs. Using four deployable SLMs (Qwen2.5-VL, MiniCPM-V, Gemma 7B, LLaVA 7B) on NIH Chest X-ray, we provide illustrative evidence: reflection-oriented prompts benefit lighter models, whereas stronger baselines are less sensitive. Beyond accuracy, we operationalize abstention, expected calibration error, and oversight burden, and we map outputs to DICOM tags, HL7 v2 messages, and FHIR ImagingStudy. The contribution is a prompt-first deployment framework, an operations playbook for calibration, logging, and change management, and a clear pathway from pilot utilities to reader studies without over-claiming clinical validation. We additionally specify a human-factors RACI, stratified calibration for dataset shift, and an auditable evidence pack to support local governance reviews.