Medical Scanner Self-Optimizing Image Acquisition

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Medical image acquisition optimization is a complex process due to numerous input parameters, requiring extensive operator input and time, and optimal settings for one study cannot be easily reused for dissimilar studies, necessitating automated parameter selection techniques.

Innovation Solution

A medical scanner employs a deep reinforcement learning framework to optimize image acquisition by using available parameters, including those stored in database systems, where an artificial agent learns to select optimal parameter values through a reward system, refining its model based on user feedback to adapt to different imaging scenarios.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If manual parameter selection is used to optimize image acquisition, then image quality can be improved, but scanning time and operator input requirements increase significantly

Engineering Contradiction:

Improveimage qualityVSAvoidscanning time

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The system employs machine learning models that automatically select optimal parameters without operator intervention. The model learns from historical scan data and user feedback to autonomously determine parameter settings, enabling the system to serve itself in optimizing image acquisition parameters.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent utilizes machine learning to dynamically adjust multiple parameters including scan duration, resolution, contrast medium dosage, and reconstruction algorithms. The system changes parameter combinations based on learned patterns from similar studies, automatically optimizing the parameter set for each specific imaging scenario.

Inventive Principle:

Principle #35Parameter changes

2Manufacturing precision

If manual parameter selection is used to optimize image acquisition, then desired imaging results can be achieved, but operator input and time requirements increase

Engineering Contradiction:

Improveimaging resultsVSAvoidoperator input

Core Design Contradiction:

Manufacturing precisionVSEase of operation

Solution Approach 1:

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces manual operator decision-making with automated machine learning algorithms. The system substitutes human cognitive processes with computational models that analyze study requirements and automatically select parameter combinations, eliminating the need for operator expertise in parameter optimization.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Productivity

If automated parameter selection is implemented, then scanning time is reduced, but adaptability to different imaging scenarios may be compromised

Engineering Contradiction:

Improvescanning efficiencyVSAvoidadaptability to different studies

Core Design Contradiction:

ProductivityVSAdaptability or versatility

Solution Approach 1:

The system employs dynamic parameter adjustment through machine learning models that adapt to different study types. The model continuously learns from new data and adjusts its parameter selection strategy based on the specific characteristics of each imaging scenario, maintaining adaptability while automating the process.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The patent incorporates feedback mechanisms where user responses to acquired images are used to refine and update the machine learning model. This feedback loop enables the system to learn from actual outcomes and improve its adaptability to different imaging scenarios over time, ensuring optimal performance across diverse applications.

Inventive Principle:

Principle #23Feedback

4Manufacturing precision

If study-specific parameters are used for each imaging task, then optimal results are achieved, but parameter reuse across different studies becomes difficult

Engineering Contradiction:

Improveoptimal imaging resultsVSAvoidparameter reuse

Core Design Contradiction:

Manufacturing precisionVSAdaptability or versatility

Solution Approach 1:

The system creates a universal machine learning model that can handle multiple imaging modalities and study types. The model learns generalizable patterns from diverse training data, enabling it to transfer knowledge across different study types. This allows parameters optimized for one study type to be effectively adapted to similar studies through the model's learned understanding of imaging parameters.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentEP3279820B1Medical scanner teaches itself to optimize clinical protocols and image acquisition

Publication Date: 2021.12.29 SIEMENS HEALTHCARE GMBH

Figure 1
Figure 2
Figure 3

AI summary

A computer-implemented method for identifying an optimal set of parameters for medical image acquisition includes receiving a set of input parameters corresponding to a medical imaging scan of a patient and using a model of operator parameter selection to determine a set of optimal target parameter values for a medical image scanner based on the set of input parameters. The medical imaging scan of the patient is performed using the set of optimal target parameter values to acquire one or more images and feedback is collected from one or more users in response to acquisition of the one or more images. This feedback is used to update the model of operator parameter selection, thereby yielding an updated model of operator parameter selection.