Local RF Coil Position Detection Using Trained Neural Networks

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current methods for determining the position of local radio-frequency coils in magnetic resonance examinations face inaccuracies due to signal variations from different tissue types, distance from the isocenter, and noise, leading to incorrect localization and maintenance challenges.

Innovation Solution

A computer-implemented method using a trained artificial neural network to process magnetic resonance data, providing a result dataset with position information of local radio-frequency coils, independent of sensor systems, and capable of handling coils outside homogeneity and linearity volumes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If the position of local radio-frequency coils is determined from the distribution of measured signal as a function of location, then the position can be automatically detected, but inaccuracies and falsification of position occur due to signal variations from different tissue types, distance from isocenter, and noise

Engineering Contradiction:

Improveautomatic detection of coil positionVSAvoidaccuracy of coil position determination

Core Design Contradiction:

Extent of automationVSMeasurement precision

Solution Approach 1:

The patent introduces a trained neural network as an intermediary between the raw magnetic resonance data and the position determination. This neural network processes the complex signal distribution data, learning to distinguish true coil position signals from variations caused by different tissue types, distance effects, and noise, thereby resolving the contradiction between automated detection and measurement accuracy

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent transforms the position determination problem by changing from direct signal-based position calculation to a learned parameter mapping through neural network training. The neural network learns optimal parameter relationships from training data, enabling accurate position determination despite signal variations, thus maintaining both automation and precision

Inventive Principle:

Principle #35Parameter changes

2Adaptability or versatility

If local radio-frequency coils are positioned outside the homogeneity and linearity volumes of the magnetic resonance apparatus, then more flexible positioning is achieved, but signal quality decreases and noise dominates leading to incorrect localization

Engineering Contradiction:

Improveflexibility in coil positioningVSAvoidaccuracy of position determination for coils outside standard regions

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent applies preliminary action by training the neural network in advance with data that includes examples of coils positioned outside homogeneity and linearity volumes. This pre-training enables the network to recognize and correctly interpret signals from coils in these challenging positions, allowing flexible positioning while maintaining determination accuracy

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent uses copying by creating a trained model that captures the relationship between signal characteristics and coil positions from training data. This trained neural network copy can then generalize to new situations, including coils positioned outside standard regions, without requiring additional hardware or manual calibration

Inventive Principle:

Principle #26Copying

3Productivity

If different local radio-frequency coils are used for different magnetic resonance measurements, then optimal coverage for each organ is achieved, but complex management and maintenance of multiple coils is required

Engineering Contradiction:

Improveefficiency of magnetic resonance measurementsVSAvoidmaintenance effort for multiple coils

Core Design Contradiction:

ProductivityVSEase of repair

Solution Approach 1:

The patent applies self-service by enabling the system to automatically detect, identify, and manage multiple local radio-frequency coils without manual intervention. The trained neural network autonomously processes signal data to determine coil positions and manages coil selection, eliminating the need for manual tracking and maintenance of multiple coils while maintaining measurement efficiency

Inventive Principle:

Principle #25Self-service

Data Source

PatentUS11719776B2Provision of position information of a local RF coil

Publication Date: 2023.08.08 SIEMENS HEALTHINEERS AG

AI summary

A computer-implemented method for provision of a result dataset having position information of a local radio-frequency coil, including: providing input data having at least magnetic resonance data, which is acquired by means of the local radio-frequency coil; determining a result dataset by applying a trained function to the input data, wherein the result dataset comprises position information for determining the position of the local radio-frequency coil; and providing the result dataset.