AI Agent Planning for Liver Tumor Thermal Ablation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Manual planning of liver tumor thermal ablation is time-consuming and prone to incomplete ablation, with conventional automatic approaches being computationally expensive and having high inference times.

Innovation Solution

The use of AI agents trained with deep reinforcement learning to determine optimal positions for ablation electrodes, iteratively updating their positions based on a defined state and cumulative reward to achieve 100% tumor coverage while satisfying clinical constraints, without requiring manual annotations during training.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If manual planning is used for thermal ablation, then the clinician can visualize and control the process, but the planning becomes time-consuming and may lead to incomplete tumor ablation

Engineering Contradiction:

Improvetumor ablation completenessVSAvoidplanning time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent replaces the manual mechanical visualization and planning process with an AI-based automatic planning system. The AI agent automatically determines electrode positions and ablation parameters by processing medical images and optimizing treatment plans, eliminating the time-consuming manual manipulation of images and planning tools while ensuring complete tumor coverage through algorithmic optimization.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system enables self-service automatic planning where the AI agent independently analyzes patient-specific anatomical data, identifies tumor boundaries, optimizes electrode placement, and generates treatment plans without requiring continuous clinician intervention. This autonomous capability reduces planning time while maintaining or improving ablation completeness compared to manual methods.

Inventive Principle:

Principle #25Self-service

2Productivity

If conventional automatic planning approaches are used, then the planning time is reduced, but the computational cost increases and inference time per patient becomes high

Engineering Contradiction:

Improveplanning speedVSAvoidcomputational energy consumption

Core Design Contradiction:

ProductivityVSUse of energy by moving object

Solution Approach 1:

The patent implements preliminary action by pre-training the AI agent on extensive datasets of anatomical variations and tumor configurations. This pre-training phase captures general patterns and knowledge, enabling the system to perform rapid, energy-efficient inference for individual patients without requiring computationally intensive processing during actual treatment planning. The pre-processed knowledge base allows fast deployment with minimal real-time computational resources.

Inventive Principle:

Principle #10Preliminary action

3Extent of automation

If conventional automatic planning approaches are used, then automation is achieved, but the inference time per patient remains high

Engineering Contradiction:

Improveplanning automationVSAvoidinference time per patient

Core Design Contradiction:

Extent of automationVSLoss of time

Solution Approach 1:

The patent employs parameter changes by optimizing the AI model's architecture and training parameters to achieve faster inference. This includes selecting appropriate network depths, activation functions, and training hyperparameters that balance model accuracy with computational speed. The optimized parameters enable the system to maintain high automation levels while reducing inference time to clinically acceptable levels.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS20240115320A1Automatic planning and guidance of liver tumor thermal ablation using ai agents trained with deep reinforcement learning

Publication Date: 2024.04.11 SIEMENS HEALTHINEERS AG

AI summary

Systems and methods for determining an optimal position of one or more ablation electrodes are provided. A current state of an environment is defined based on a mask of one or more anatomical objects and one or more current positions of one or more ablation electrodes. The one or more anatomical objects comprise one or more tumors. For each particular AI (artificial intelligence) agent of one or more AI agents, one or more actions for updating the one or more current positions of a respective ablation electrode of the one or more ablation electrodes in the environment are determined based on the current state using the particular AI agent. A next state of the environment is defined based on the mask and the one or more updated positions of the respective ablation electrode. The steps of determining the one or more actions and defining the next state are repeated for a plurality of iterations using 1) the next state as the current state and 2) the one or more updated positions as the one or more current positions to determine one or more final positions of the respective ablation electrode for performing a thermal ablation on the one or more tumors. The one or more final positions of each respective ablation electrode are output.