Deep Reinforcement Learning Agent for Medical Imaging Parameter Optimization

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional medical imaging systems require labor-intensive and time-consuming manual configuration of imaging parameters to optimize image quality, which varies by user preference, making them inefficient and costly.

Innovation Solution

A Deep Reinforcement Learning (DRL) agent is used to map user-selected images to optimal imaging parameters, iteratively adjusting settings to meet quality thresholds for sharpness, fuzziness, blurring, noise, dynamic range, contrast, and brightness, leveraging deep learning methods for quantification and inverse reinforcement learning to automate the process.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If manual configuration of imaging parameters is used, then image quality can be optimized according to user preferences, but the process becomes labor intensive and time consuming

Engineering Contradiction:

Improveimage quality optimizationVSAvoidparameter configuration time

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The system uses deep reinforcement learning agents that automatically learn and optimize imaging parameters through self-directed interaction with the imaging system, eliminating the need for manual configuration by medical professionals while maintaining personalized quality optimization

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The invention transforms the discrete manual adjustment of multiple imaging parameters into a continuous automated optimization process where the DRL agent dynamically changes parameters based on learned patterns and real-time feedback, converting a complex manual task into an efficient automated system

Inventive Principle:

Principle #35Parameter changes

2Manufacturing precision

If manual configuration of imaging parameters is used, then image quality can be optimized according to user preferences, but the process becomes expensive

Engineering Contradiction:

Improveimage quality optimizationVSAvoidcost

Core Design Contradiction:

Manufacturing precisionVSLoss of energy

Solution Approach 1:

The invention replaces the mechanical process of manual parameter adjustment with an intelligent software-based DRL system, substituting human expertise and labor with automated machine learning algorithms that achieve the same optimization goals without the associated labor costs

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Manufacturing precision

If numerous imaging parameters are adjusted manually, then image quality can be optimized, but the complexity of the configuration process increases

Engineering Contradiction:

Improveimage qualityVSAvoidparameter configuration complexity

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The invention segments the complex parameter optimization problem into manageable components by using multiple specialized DRL agents, each responsible for learning and optimizing specific parameter groups or aspects of image quality, making the overall complex system more tractable and maintainable

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS10957037B2Smart imaging using artificial intelligence

Publication Date: 2021.03.23 SIEMENS HEALTHINEERS AG

AI summary

Systems and methods are provided for determining a set of imaging parameters for an imaging system. A selection of an image is received from a set of images. A modification of certain quality measures is received for the selected image. The modified selected image is mapped to a set of imaging parameters of an imaging system based on the certain quality measures using a trained Deep Reinforcement Learning (DRL) agent.