CT-PrepAgent: Bounded Policy and Controlled Execution for Adaptive CT Data Preparation

2026-08-02

AI Summary

This study addresses the challenge that heterogeneous CT acquisition protocols and diverse downstream tasks hinder the generalizability of fixed preprocessing pipelines, as existing approaches rely on handcrafted rules and lack adaptability. To overcome this limitation, the authors introduce, for the first time, a large language model (LLM) agent that constructs structured data-task profiles and employs a bounded decision policy to select optimal DICOM sequences or preprocessing configurations. A controlled deterministic execution mechanism is further designed to enable decision verification, fault-tolerant recovery, and secure isolation. The proposed method achieves state-of-the-art macro-averaged Dice scores across three public CT segmentation benchmarks and improves the output rate from 61.7% to 70.0% on two private raw DICOM cohorts while maintaining stable registration metrics.

Abstract

Heterogeneous computed tomography (CT) acquisitions and diverse downstream task requirements limit the transferability of fixed data preparation workflows across data sources and tasks. Existing approaches typically rely on manually designed or dataset-specific rules, making it difficult to accommodate changes in acquisition conditions and analytical objectives without manual intervention. Large language model (LLM)-based agents have shown promise for automating medical workflows, yet their potential for adaptive CT data preparation remains largely unexplored. To bridge this gap, we propose CT-PrepAgent, which enables adaptive CT data preparation through a bounded policy and controlled deterministic execution. Deterministic inspection constructs structured data--task profiles, from which a policy decides an eligible DICOM series or predefined preprocessing profile, while the controlled execution flow guards, resolves, executes, and verifies the decision with bounded recovery when enabled and safe quarantine otherwise. Across three public CT segmentation tasks, CT-PrepAgent derived data-task adaptive preprocessing decisions and achieved the highest macro-average Dice. On two private raw-DICOM cohorts, CT-PrepAgent increased verified output yield from 61.7\% to 70.0\% and yielded similar registration metrics on common verified outputs. Controlled fault and replay tests validate bounded recovery, safe quarantine, and policy-free replay under tested fault and drift settings.

Problem

• CT data preparation
• heterogeneous acquisitions
• downstream task requirements
• transferability
• adaptive preprocessing

Innovation

• adaptive CT preprocessing
• bounded policy
• controlled execution
• LLM-based agent
• DICOM data preparation