MARCH: Multi-Agent Radiology Clinical Hierarchy for CT Report Generation

TL;DR

MARCH introduces a multi-agent framework that mimics radiology department hierarchy to improve the accuracy and reliability of CT report generation, addressing hallucinations and verification issues.

Contribution

This work presents a novel multi-agent system that models clinical workflows, significantly enhancing report quality over existing single-model approaches.

Findings

Outperforms state-of-the-art baselines in clinical fidelity.

Achieves higher linguistic accuracy in generated reports.

Demonstrates the effectiveness of organizational modeling in AI reliability.

Abstract

Automated 3D radiology report generation often suffers from clinical hallucinations and a lack of the iterative verification found in human practice. While recent Vision-Language Models (VLMs) have advanced the field, they typically operate as monolithic "black-box" systems without the collaborative oversight characteristic of clinical workflows. To address these challenges, we propose MARCH (Multi-Agent Radiology Clinical Hierarchy), a multi-agent framework that emulates the professional hierarchy of radiology departments and assigns specialized roles to distinct agents. MARCH utilizes a Resident Agent for initial drafting with multi-scale CT feature extraction, multiple Fellow Agents for retrieval-augmented revision, and an Attending Agent that orchestrates an iterative, stance-based consensus discourse to resolve diagnostic discrepancies. On the RadGenome-ChestCT dataset, MARCH significantly outperforms state-of-the-art baselines in both clinical fidelity and linguistic accuracy. Our work demonstrates that modeling human-like organizational structures enhances the reliability of AI in high-stakes medical domains.