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

2026-04-17 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Medical Devices & Health Technology · Depth: Expert, quick

Summary

MARCH (Multi-Agent Radiology Clinical Hierarchy) is a new multi-agent framework designed to generate 3D radiology reports from CT scans, addressing issues like clinical hallucinations and lack of iterative verification common in existing Vision-Language Models (VLMs). This framework mimics the professional hierarchy of radiology departments, assigning specialized roles to different agents. It includes a Resident Agent for initial drafting using multi-scale CT feature extraction, multiple Fellow Agents for retrieval-augmented revisions, and an Attending Agent that manages an iterative, stance-based consensus process to resolve diagnostic discrepancies. Evaluated on the RadGenome-ChestCT dataset, MARCH significantly surpasses current state-of-the-art baselines in both clinical fidelity and linguistic accuracy, demonstrating the benefits of human-like organizational structures in high-stakes medical AI.

Key takeaway

For NLP Engineers developing medical report generation systems, MARCH offers a blueprint for enhancing reliability and accuracy. You should consider implementing multi-agent architectures that mirror human clinical hierarchies to improve diagnostic consistency and reduce hallucinations. This approach can lead to more trustworthy AI outputs in critical healthcare applications.

Key insights

A multi-agent framework mimicking radiology hierarchies improves CT report generation fidelity and accuracy.

Principles

• Emulate human organizational structures for AI reliability.
• Specialized agent roles enhance complex task performance.

Method

MARCH uses a Resident Agent for initial drafts, Fellow Agents for retrieval-augmented revision, and an Attending Agent for iterative consensus to resolve diagnostic discrepancies in CT report generation.

In practice

• Apply multi-agent systems to mimic clinical workflows.
• Utilize multi-scale feature extraction for medical imaging.

Topics

• MARCH Framework
• Multi-Agent Systems
• Radiology Report Generation
• Clinical Hierarchy Emulation
• CT Feature Extraction

Best for: NLP Engineer, Computer Vision Engineer, AI Scientist, Research Scientist, AI Architect

Related on AIssential

• MARCH: Multi-Agent Radiology Clinical Hierarchy for CT Report Generation — Modeling human clinical hierarchies with multi-agent AI improves radiology report accuracy and reduces hallucinations.
• CAREAgent: Clinical Agent with Structured Reasoning and Tool-Integrated for Order Generation — CAREAgent generates fine-grained clinical orders using a two-stage data construction and hybrid training approach, outperforming prior agent methods.
• Enhancing Clinical Trial Patient Matching through Knowledge Augmentation and Reasoning with Multi-Agent — Multi-agent systems can augment LLMs with domain knowledge to improve clinical trial patient matching accuracy.
• UniRG: Scaling medical imaging report generation with multimodal reinforcement learning — Reinforcement learning with clinically meaningful rewards significantly improves medical vision-language model reliability and generality.
• Exploring the feasibility of conversational diagnostic AI in a real-world clinical study — Supervised conversational AI for pre-visit history taking is feasible, safe, and well-received by patients and clinicians.
• MedRLM: Recursive Multimodal Health Intelligence for Long-Context Clinical Reasoning, Sensor-Guided Screening, Evidence-Grounded Decision Support, and Community-to-Tertiary Referral Optimization — MedRLM recursively processes heterogeneous patient data as an external clinical environment for auditable, workflow-aware decision support.

Open in AIssential →

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