Deep Learning Quantitative MRI from Single Qualitative Scan
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional clinical MRI requires multiple scans with different imaging parameters to obtain quantitative MR relaxation parametric maps, making the process time-consuming and incompatible with standard clinical protocols.
Innovation Solution
A deep learning-based method using hierarchical convolutional neural networks to derive quantitative MR parametric maps from a single qualitative MR image acquired with standard imaging protocols, leveraging statistical and spatial a priori information to produce T1, T2, and proton density maps without the need for additional scans or modified protocols.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple qualitative MRI scans with different imaging parameters are performed to obtain quantitative relaxation parametric maps, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The method performs preliminary action by acquiring multiple qualitative MRI images with different imaging parameters (TR, TE, flip angle) before quantitative map generation. These pre-acquired images with varying contrasts provide the necessary data foundation for subsequent quantitative parameter calculation, enabling accurate T1 and T2 relaxation time measurement without requiring dedicated quantitative scanning sequences
Solution Approach 2:
The method creates a copy of quantitative information from qualitative images. By processing multiple qualitative MRI images that were acquired for diagnostic purposes, the system generates quantitative relaxation parametric maps as derived products. This copying approach extracts quantitative measurements from existing qualitative data without requiring separate quantitative scanning
2Measurement precision
If multiple qualitative MRI scans are performed to ensure adequate sampling of signal evolution, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The method achieves multi-functionality by using a single qualitative MRI scanning protocol that serves dual purposes: generating diagnostic images for clinical evaluation and providing data for quantitative relaxation parameter calculation. The same imaging sequence and parameters used for diagnostic imaging are leveraged to extract T1 and T2 maps, eliminating the need for separate quantitative imaging protocols and reducing overall system complexity
3Productivity
If standard clinical imaging protocols are modified to implement MR fingerprinting, then productivity is improved, but ease of operation deteriorates
Solution Approach 1:
The method performs preliminary action by acquiring multiple qualitative MRI images with different imaging parameters (TR, TE, flip angle) before quantitative map generation. These pre-acquired images with varying contrasts provide the necessary data foundation for subsequent quantitative parameter calculation, enabling accurate T1 and T2 relaxation time measurement without requiring dedicated quantitative scanning sequences
Solution Approach 2:
Instead of modifying the pulse sequence to encode quantitative information directly during scanning (MR fingerprinting approach), this method inverts the conventional approach by acquiring standard qualitative images and then deriving quantitative parameters through post-processing. This reversal maintains compatibility with existing clinical workflows while achieving quantitative measurement goals
Data Source
AI summary
A method for quantitative magnetic resonance imaging (MRI) includes [800] performing an MRI scan using a conventional pulse sequence to obtain a qualitative MR image; and [802] applying the qualitative MR image as input to a deep convolutional neural network (CNN) to produce a quantitative magnetic resonance (MR) relaxation parametric map. The qualitative MR image is the only image input to the deep neural network to produce the quantitative MR relaxation parametric map. The conventional pulse sequence may be a Spoiled Gradient Echo sequence, a Fast Spin Echo sequence, a Steady-State Free Precession sequence, or other sequence that is commonly used in current clinical practice.