Machine-Learned MRI Configuration from Updated Scanner Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Magnetic Resonance Imaging (MRI) suffers from non-reproducibility and non-quantitativeness due to scanner-induced imperfections and patient-induced effects, leading to variability in image quality and hindered quantification of relaxation parameters.
Innovation Solution
A machine learning model (MLM) is used to update an initial scanner model based on actual MRI measurement data, predicting an updated scanner model to determine an optimal configuration for MRI, mitigating these imperfections and effects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If very fast and very accurate adjustment and calibration scans are performed to incorporate scanner-induced imperfections and patient-induced effects, then the quantitative accuracy and reproducibility of MRI results improve, but the scanning time and complexity increase significantly
Solution Approach 1:
The system performs preliminary characterization of scanner imperfections (eddy currents, gradient non-linearities, B0 variations) and patient-specific effects (coil loading, anatomy-induced B0 variations) before the actual diagnostic scan. This preliminary action creates correction maps and updated scanner models that are then applied during routine scanning, eliminating the need for time-consuming calibration scans before each diagnostic examination.
Solution Approach 2:
The system creates a digital copy or model of the scanner's actual performance characteristics (scanner model) that includes all imperfections and variations. This digital model is updated using measurement data and then used to predict and correct for scanner-induced imperfections and patient-induced effects, replacing the need for repeated physical calibration scans.
2Measurement precision
If traditional physics-based correction methods are used to account for eddy currents and gradient non-linearities, then the quantitative accuracy improves, but the device complexity and computational requirements increase
Solution Approach 1:
The system replaces complex physics-based correction algorithms with a machine learning model that has been trained to predict scanner imperfections. Instead of implementing and computing complex electromagnetic field models for eddy currents and gradient non-linearities, the trained MLM directly predicts the correction needed based on input measurement data, simplifying the computational approach while maintaining accuracy.
Solution Approach 2:
The system changes the approach from calculating physical parameters (eddy current decay constants, gradient non-linearity coefficients) to using a data-driven model that learns the relationship between measurement data and correction parameters. The MLM transforms the problem from physics-based parameter estimation to pattern recognition, changing the fundamental parameters from physical constants to learned model weights.
3Reliability
If scanner models are frequently updated to reflect current scanner performance and patient-specific effects, then the reproducibility of MRI results improves, but the computational processing time and resource requirements increase
Solution Approach 1:
The system performs preliminary training of the machine learning model using extensive measurement data collected under various conditions. This preliminary training phase creates a pre-configured model that can then be quickly applied to new patients without requiring extensive real-time computation, separating the heavy computational work from the time-critical scanning process.
Solution Approach 2:
The system automatically updates the scanner model using measurement data acquired during routine operation, without requiring manual intervention or extensive processing. The MLM self-adjusts by learning from the data, automatically incorporating scanner drift and patient-specific effects into the model, reducing the need for manual recalibration and extensive computational reprocessing.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach improves the reproducibility and quantitative evaluation of MRI results by providing an up-to-date scanner model for accurate data acquisition and image reconstruction, enhancing image quality and quantification.
Implementation Method 1
an initial scanner model for an MRI scanner is received, the initial scanner model specifying a deviation of a main magnetic field from a predefined target main magnetic field
Implementation Method 2
Magnetic Resonance Imaging (MRI)
Implementation Method 3
A machine learning model (MLM) is used to update an initial scanner model based on actual MRI measurement data, predicting an updated scanner model
Data Source
AI summary
For automatically determining a configuration for magnetic resonance imaging (MRI), an initial scanner model for an MRI scanner is received. The initial scanner model specifies a deviation of a main magnetic field from a predefined target main magnetic field and/or a deviation of a magnetic field gradient from a predefined target gradient field. MRI measurement data measured by using the MRI scanner is received. A first updated scanner model is generated by applying a trained first machine learning model (MLM) to first input data that depends on the initial scanner model and the MRI measurement data. The configuration for MRI is determined depending on the first updated scanner model.