Low-Field MR Image Optimization Using High-Quality Reference Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Magnetic resonance (MR) images often have low resolution and low signal-to-noise ratio (SNR), which are unsatisfactory for clinical needs, particularly when using volume transmitting coils (VTC) or low-field MRI devices.
Innovation Solution
An optimization model is employed that utilizes a deep feature extraction component and correlation search component to enhance image quality by transferring high-quality image features to low-quality MR images, using a reference image with higher quality to improve resolution and reduce noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If volume transmitting coils (VTC) or low-field MRI devices are used, then device complexity and cost are reduced, but image resolution and signal-to-noise ratio deteriorate
Solution Approach 1:
The patent introduces a high-resolution reference image as an intermediary to transfer quality information to the low-quality initial image. The optimization model uses the reference image to guide the enhancement process, effectively mediating between the limited capabilities of VTC/low-field devices and the requirement for high-quality clinical images.
Solution Approach 2:
The patent changes the parameter space by transforming the image from low-resolution to high-resolution through the optimization model. The model learns the mapping between different image quality parameters, enabling transformation of images captured with simpler devices into images that meet clinical quality standards.
2Device complexity
If volume transmitting coils (VTC) or low-field MRI devices are used, then device complexity and cost are reduced, but signal-to-noise ratio deteriorates
Solution Approach 1:
The patent converts the harmful noise and low signal-to-noise ratio characteristics into beneficial information by using the optimization model to learn and separate signal from noise. The model leverages the statistical patterns in noisy images to reconstruct cleaner, higher-quality images, effectively turning the weakness of VTC/low-field devices into an opportunity for intelligent enhancement.
Solution Approach 2:
The optimization model acts as an intermediary that processes and cleanses the noisy signals from VTC or low-field devices. By introducing this computational intermediary, the system can achieve high signal-to-noise ratio outputs even from devices that inherently produce noisy images.
3Measurement precision
If deep learning optimization models are used to enhance image quality, then image resolution and signal-to-noise ratio are improved, but computational complexity and processing time increase
Solution Approach 1:
The patent performs preliminary action by pre-training the optimization model on large datasets of paired low-quality and high-quality images. This pre-training phase captures the essential transformations needed for image enhancement, so that during actual clinical use, the model can quickly process images without requiring extensive computational resources in real-time.
Solution Approach 2:
The patent uses copying by training the model on synthetic or simulated image pairs that replicate the characteristics of VTC and low-field MRI images. This allows the model to learn enhancement patterns from abundant training data without requiring equally abundant clinical scan pairs, reducing the computational burden of model development and deployment.
Data Source
AI summary
Systems and methods for image optimization are provided. The systems may obtain an initial image of a target object. The systems may also obtain a correlation reference image that is generated based on a reference image associated with the target object. The reference image may have a second image quality higher than a first image quality of the initial image, and the correlation reference image may have a third image quality lower than the second image quality. The systems may further determine an optimized image of the initial image by inputting the initial image and the correlation reference image to an optimization model. The optimization model may refer to a machine learning model that is configured for high-resolved and noise-reduced reconstruction using priori information existing in the reference image. The optimized image may have a fourth image quality higher than the first image quality.