MRI Navigator-Based Parameter Optimization for High-Throughput Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for analyzing industrial samples using MRI techniques face challenges in achieving high throughput while maintaining quality, particularly in reducing the time required for sample analysis without compromising image quality.
Innovation Solution
A method involving a preparation MRI experiment to derive optimal scan settings using a machine learning module, which updates MRI experimental parameters for a subsequent MRI experiment, allowing for precise targeting of features and optimization of acquisition parameters to enhance image quality and reduce acquisition time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the number of samples is reduced to increase processing rate, then productivity increases, but measurement precision deteriorates
Solution Approach 1:
The system performs a preliminary navigation MRI scan before the main experiment to pre-identify the region of interest and pre-optimize acquisition parameters. This preliminary action enables the main scan to focus resources on the relevant area, maintaining high image quality while reducing overall processing time and enabling higher throughput.
Solution Approach 2:
Instead of uniformly scanning the entire sample with high resolution, the system applies local quality optimization by identifying and focusing computational and imaging resources specifically on the region containing the feature of interest. This selective approach maintains measurement precision for the critical area while reducing processing requirements for the overall sample.
2Productivity
If shot repetition time is reduced to increase processing rate, then productivity increases, but measurement precision deteriorates
Solution Approach 1:
The navigation scan performs preliminary optimization of acquisition parameters including shot repetition time based on the specific sample characteristics and region of interest. This pre-optimization enables the main experiment to use tailored parameters that maintain measurement precision while achieving higher processing rates than standard protocols.
3Productivity
If the field of view is reduced to increase processing rate, then productivity increases, but measurement precision deteriorates
Solution Approach 1:
The navigation MRI scan pre-identifies the exact location and extent of the region of interest, enabling the system to reduce the field of view for the main experiment to only the necessary area. This preliminary localization maintains measurement precision for the feature of interest while significantly reducing processing time and enabling higher throughput.
Solution Approach 2:
The system applies local quality by concentrating imaging resources and computational power on the specific region containing the feature of interest rather than uniformly processing the entire sample. This selective focus maintains high measurement precision for the critical area while reducing overall processing requirements.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
The invention relates to a Method for automated non-invasive analysis of a predetermined feature in a multitude of industrial samples (100) of a predefined sample type, the method comprising the steps of: b) recording in a preparation MRI experiment first MRI data of the industrial sample (100) with the MR scanner (102) to obtain a navigator information comprising the recorded first MRI data, c) analysing the navigator information with an inference module (200) for deriving one or more MR experimental parameters for use in a subsequent MR experiment for the analysis of the predetermined feature in the industrial sample (100) using a machine learning module (202), d) recording in a subsequent MR experiment MR data of the industrial sample (100) with the MR scanner(102) using the derived one or more MR experimental parameters, and e) analysing the predetermined feature in the industrial sample (100) using the MR data of the industrial sample (100), wherein the machine learning module (202) is trained for deriving one or more MR experimental parameters for use in a subsequent MR experiment for the analysis of the predetermined feature in the industrial sample (100) by analysing navigator information using a training set comprising navigator information of different training samples and corresponding MR experimental parameters used in a subsequent MR experiment for the analysis of the predetermined feature in the training sample.