An Artifact-based Agent Framework for Adaptive and Reproducible Medical Image Processing

arXiv cs.AI / 4/27/2026

📰 NewsDeveloper Stack & InfrastructureModels & Research

共有:

Key Points

The paper argues that medical image processing in real clinical settings requires dataset-aware workflow configuration plus end-to-end provenance tracking, not just model design.
It proposes an artifact-based agent framework that uses an artifact contract and semantic layer to represent intermediate/final outputs and to support goal-conditioned assembly of workflow configurations from modular rules.
A dedicated workflow executor is used to preserve deterministic computational graph construction and to record enough provenance for transformations and decisions to be re-executed reliably.
The agent runs locally to better satisfy privacy constraints, and evaluations on real CT/MRI cohorts show adaptive configuration synthesis and reproducible results across repeated runs, along with artifact-grounded semantic querying.

Abstract

Medical imaging research is increasingly shifting from controlled benchmark evaluation toward real-world clinical deployment. In such settings, applying analytical methods extends beyond model design to require dataset-aware workflow configuration and provenance tracking. Two requirements therefore become central: \textbf{adaptability}, the ability to configure workflows according to dataset-specific conditions and evolving analytical goals; and \textbf{reproducibility}, the guarantee that all transformations and decisions are explicitly recorded and re-executable. Here, we present an artifact-based agent framework that introduces a semantic layer to augment medical image processing. The framework formalizes intermediate and final outputs through an artifact contract, enabling structured interrogation of workflow state and goal-conditioned assembly of configurations from a modular rule library. Execution is delegated to a workflow executor to preserve deterministic computational graph construction and provenance tracking, while the agent operates locally to comply with most privacy constraints. We evaluate the framework on real-world clinical CT and MRI cohorts, demonstrating adaptive configuration synthesis, deterministic reproducibility across repeated executions, and artifact-grounded semantic querying. These results show that adaptive workflow configuration can be achieved without compromising reproducibility in heterogeneous clinical environments.