New Insights into Automatic Treatment Planning for Cancer Radiotherapy Using Explainable Artificial Intelligence

2025-08-19

AI Summary

AI agents in prostate cancer radiotherapy inverse planning suffer from decision opacity, hindering clinical trust and adoption. Method: We propose the first systematic explainable AI (XAI) framework to analyze the decision-evolution mechanism of Actor-Critic with Experience Replay (ACER) agents across training stages. Using attribution analysis, we quantify how dose-volume histogram (DVH) inputs influence adjustments to treatment planning parameters (TPPs). Contribution/Results: High-performing agents emulate expert clinicians by identifying critical dose violations and adopting global parameter-tuning strategies. Experiments show tuning steps reduced from 22 to 12–13, parameter-space entropy decreased to 0.3, dose violations significantly diminished, and attribution similarity reached 0.25–0.5—demonstrating both strategy convergence and clinical plausibility. This work pioneers an interpretable pathway into the “black-box” decisions of AI-driven radiotherapy planning, providing theoretical foundations and methodological paradigms for trustworthy AI deployment in clinical practice.

Technology Category

• Humans and AI: Explainable AI (XAI) for Human Understanding
• Philosophy and Ethics of AI: Accountability, Interpretability & Explainability
• Multiagent Systems: Adversarial Agents

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Search and Retrieval-Augmented AI: Agentic search
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Objective: This study aims to uncover the opaque decision-making process of an artificial intelligence (AI) agent for automatic treatment planning. Approach: We examined a previously developed AI agent based on the Actor-Critic with Experience Replay (ACER) network, which automatically tunes treatment planning parameters (TPPs) for inverse planning in prostate cancer intensity modulated radiotherapy. We selected multiple checkpoint ACER agents from different stages of training and applied an explainable AI (EXAI) method to analyze the attribution from dose-volume histogram (DVH) inputs to TPP-tuning decisions. We then assessed each agent's planning efficacy and efficiency and evaluated their policy and final TPP tuning spaces. Combining these analyses, we systematically examined how ACER agents generated high-quality treatment plans in response to different DVH inputs. Results: Attribution analysis revealed that ACER agents progressively learned to identify dose-violation regions from DVH inputs and promote appropriate TPP-tuning actions to mitigate them. Organ-wise similarities between DVH attributions and dose-violation reductions ranged from 0.25 to 0.5 across tested agents. Agents with stronger attribution-violation similarity required fewer tuning steps (~12-13 vs. 22), exhibited a more concentrated TPP-tuning space with lower entropy (~0.3 vs. 0.6), converged on adjusting only a few TPPs, and showed smaller discrepancies between practical and theoretical tuning steps. Putting together, these findings indicate that high-performing ACER agents can effectively identify dose violations from DVH inputs and employ a global tuning strategy to achieve high-quality treatment planning, much like skilled human planners. Significance: Better interpretability of the agent's decision-making process may enhance clinician trust and inspire new strategies for automatic treatment planning.

Problem

• Uncovering opaque AI decision-making in radiotherapy treatment planning
• Analyzing how AI agents tune treatment planning parameters automatically
• Explaining AI-driven dose-violation mitigation in prostate cancer radiotherapy

Innovation

• Used explainable AI to analyze ACER agent decisions
• Applied attribution analysis to dose-volume histogram inputs
• Systematically examined TPP-tuning strategies for radiotherapy planning