Deep Learning Imaging System Configuration for Diagnostic Accuracy

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Healthcare facilities face challenges in providing quality care due to economic, technological, and administrative hurdles, including limited access to imaging systems, complex equipment, and the need for standardized radiation exposure dose management, which complicates effective image acquisition and diagnosis.

Innovation Solution

The implementation of a deep learning-based imaging system configuration apparatus that includes a training learning device to learn imaging system configuration parameters from prior image acquisitions and a deployed learning device to provide optimized settings for image acquisition, using a deep learning network to process feedback and update configurations for improved image quality and diagnostic accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual configuration of imaging system parameters is used, then system complexity is reduced, but diagnostic accuracy and image quality deteriorate

Engineering Contradiction:
Improvediagnostic accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The imaging system automatically configures acquisition parameters using deep learning networks that learn optimal settings from training data. The system serves itself by selecting parameters based on patient characteristics and clinical indications without requiring manual expert intervention, thereby maintaining high diagnostic accuracy while reducing operational complexity.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system dynamically adjusts imaging parameters (such as radiation dose, acquisition time, and technical settings) based on learned patterns from training data. The deep learning network selects optimal parameter combinations tailored to each specific examination type and patient condition, improving image quality and diagnostic accuracy through data-driven parameter optimization.

Inventive Principle:
Principle #35Parameter changes

2Object-affected harmful factors

If standardized radiation exposure management is implemented, then patient safety is improved, but image acquisition flexibility deteriorates

Engineering Contradiction:
Improveradiation exposureVSAvoidimage acquisition flexibility
Core Design Contradiction:
Object-affected harmful factorsVSAdaptability or versatility

Solution Approach 1:

The deep learning network optimizes radiation exposure parameters within standardized safety limits by learning the minimum necessary dose for each examination type and patient condition. The system adjusts technical parameters such as tube current, voltage, and acquisition protocols to achieve diagnostic quality images at the lowest feasible radiation levels, maintaining both patient safety and acquisition flexibility.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The system uses feedback from training data consisting of prior image acquisitions and their associated outcomes to continuously refine radiation dose selection. By analyzing historical data on image quality, diagnostic accuracy, and patient characteristics, the system learns to prescribe optimal radiation doses that satisfy both safety constraints and diagnostic requirements for each specific case.

Inventive Principle:
Principle #23Feedback

3Manufacturing precision

If deep learning networks are trained with extensive feedback data, then image quality improves, but processing time and computational resources increase

Engineering Contradiction:
Improveimage qualityVSAvoidprocessing time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

The deep learning networks are trained offline using extensive feedback data from prior image acquisitions before deployment. This preliminary training phase allows the system to learn optimal parameter selections and image quality patterns in advance, so that during actual clinical operation, the pre-trained networks can rapidly select parameters without requiring real-time extensive processing, thus achieving high image quality with minimal processing delay.

Inventive Principle:
Principle #10Preliminary action

4Productivity

If automated parameter selection is implemented, then operational efficiency is improved, but ease of operation by physicians deteriorates

Engineering Contradiction:
Improveoperational efficiencyVSAvoidease of operation
Core Design Contradiction:
ProductivityVSEase of operation

Solution Approach 1:

The system performs automated parameter selection and configuration without requiring manual input from physicians, thereby improving operational efficiency. The deep learning network independently selects optimal acquisition parameters based on patient characteristics and examination type, reducing the time and expertise required for system operation while maintaining or improving image quality and diagnostic accuracy.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP3545526B1Deep learning medical systems and methods for image acquisition
Publication Date: 2024.05.22 GENERAL ELECTRIC CO
  • EP3545526B1 patent drawingFigure 1
  • EP3545526B1 patent drawingFigure 2
  • EP3545526B1 patent drawingFigure 3

AI summary

Methods and apparatus for improved deep learning for image acquisition are provided. An imaging system configuration apparatus includes a training learning device including a first processor to implement a first deep learning network (DLN) to learn a first set of imaging system configuration parameters based on a first set of inputs from a plurality of prior image acquisitions to configure at least one imaging system for image acquisition, the training learning device to receive and process feedback including operational data from the plurality of image acquisitions by the at least one imaging system. The example apparatus includes a deployed learning device including a second processor to implement a second DLN, the second DLN generated from the first DLN of the training learning device, the deployed learning device configured to provide a second imaging system configuration parameter to the imaging system in response to receiving a second input for image acquisition.