Harrison.Rad 1.5 Technical Report: A radiology foundation model that can draft reports from images, priors and clinical context

2026-07-07 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Computer Vision & Pattern Recognition, Health & Medical Research · Depth: Expert, quick

Summary

Harrison.Rad 1.5 (HR1.5) is a radiology-specific multimodal large language model designed to draft reports from images, prior studies, and clinical context, addressing the growing demand for imaging and radiology workforce shortages. It processes interleaved text and visual inputs to generate structured and unstructured text for plain-film radiology, including computed radiography, chest, musculoskeletal, abdominal, spine, pelvic x-rays, and mammography. HR1.5's training involves a three-stage pipeline: domain adaptation on radiology reports, contrastive vision-encoder training with curriculum-based hard negatives on ~6 million image-report instances, and visual-question-answering fine-tuning. Evaluated using a Findings-Diagnosis scoring framework and benchmarked on RadBench, HR1.5 is the only system to meet the simulated FRCR passing standard, achieving the highest accuracy on closed-format clinical questions and internal multi-body-part and mammography reporting. The model also includes explainability features like Grad-CAM heatmaps and attention analysis.

Key takeaway

For radiology department heads evaluating AI solutions to mitigate reporting backlogs, Harrison.Rad 1.5 offers a validated approach. You should consider integrating this multimodal LLM to automate initial report drafting, reducing radiologist workload and improving turnaround times. Its demonstrated ability to meet FRCR passing standards suggests a high level of clinical utility. Explore its explainability features to ensure responsible deployment and build trust in AI-assisted diagnostics.

Key insights

Harrison.Rad 1.5 is a multimodal LLM that drafts radiology reports, meeting FRCR standards through a specialized three-stage training.

Principles

• Domain adaptation improves LLM performance.
• Curriculum-based hard negatives enhance vision-encoder training.
• Multimodal models can exceed human exam standards.

Method

HR1.5's three-stage pipeline includes domain adaptation of a base LLM on radiology reports, contrastive vision-encoder training on ~6 million image-report instances, and VQA fine-tuning.

In practice

• Integrate multimodal LLMs for report drafting.
• Use Findings-Diagnosis framework for evaluation.
• Apply Grad-CAM for model explainability.

Topics

• Radiology AI
• Multimodal LLMs
• Medical Imaging
• Clinical Report Generation
• FRCR Examination
• Explainable AI

Best for: AI Scientist, Machine Learning Engineer, Research Scientist

Related on AIssential

• Scalable Training of Spatially Grounded 2D Vision-Language Models for Radiology — Spatially grounded radiology VLMs can be trained at scale without manual annotations, maintaining language quality.
• Beyond Scalar Scores: Exploring LLM-based Metrics for Clinical Significance Evaluation in Radiology Reports — LLM evaluators for radiology reports suffer "discrimination bias," but targeted data synthesis and fine-tuning can sharpen clinical significance boundaries.
• MARCH: Multi-Agent Radiology Clinical Hierarchy for CT Report Generation — A multi-agent framework mimicking radiology hierarchies improves CT report generation fidelity and accuracy.
• Multimodal foundation models exploit text to make medical image predictions — Multimodal medical AI models heavily rely on text, which can both enhance and degrade image-based diagnostic accuracy.
• The Slop Paradox: How Synthetic Standardization Erodes Clinical Uncertainty and Cross-Modal Alignment in AI-Rewritten Radiology Reports — AI-rewriting of clinical text for standardization can paradoxically degrade cross-modal alignment despite appearing cleaner.
• Intelligent Healthcare Imaging Platform: A VLM-Based Framework for Automated Medical Image Analysis and Clinical Report Generation — A VLM-based framework automates medical image analysis and clinical report generation across multiple imaging modalities.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.