MRI Artifact Correction via Neural Network Synthetic Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Magnetic Resonance Imaging (MRI) systems face challenges in reducing artifacts and accelerating image acquisition due to subject motion and spurious RF signals, which corrupt magnetic resonance images and require lengthy data acquisition times.
Innovation Solution
A medical system employing an image generating neural network that takes reference magnetic resonance image data from a different configuration and generates synthetic data simulating images from a first configuration, allowing for the reconstruction of corrected images from measured k-space data, using techniques such as regularization terms or synthetic k-space data modification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If measured k-space data is acquired using a first configuration of the magnetic resonance imaging system, then image acquisition speed is improved, but image quality deteriorates due to artifacts from subject motion and spurious RF signals
Solution Approach 1:
The system performs preliminary action by generating synthetic reference magnetic resonance image data from a second configuration before reconstruction. This synthetic data serves as prior knowledge that is incorporated into the reconstruction process of the measured k-space data from the first configuration, thereby pre-preparing correction information to mitigate artifacts from subject motion and spurious RF signals.
2Measurement precision
If extensive data sampling is performed to improve image quality, then measurement precision is improved, but loss of time increases due to lengthy acquisition requirements
Solution Approach 1:
The system creates a copy by generating synthetic reference magnetic resonance image data from a second configuration that replicates the anatomical information. This synthetic copy is then used as prior knowledge in the reconstruction process, allowing the system to achieve high measurement precision without requiring extensive direct sampling of the actual k-space data, thereby reducing acquisition time.
3Reliability
If an image-to-image neural network is trained to generate images with reduced artifacts, then image quality is improved, but device complexity increases
Solution Approach 1:
The system introduces an intermediary approach by using a generative adversarial network (GAN) that includes a generator and a discriminator. The generator creates synthetic reference image data while the discriminator evaluates its authenticity. This intermediary GAN framework acts as a mediator between the measured k-space data and the final reconstructed image, improving image quality by separating the artifact reduction function into a dedicated neural network component.
Data Source
AI summary
Disclosed herein is a medical system (100, 300) comprising a memory (110) storing machine executable instructions (120) and an image generating neural network (122). The image generating neural network is configured for outputting synthetic magnetic resonance image data (128) in response to receiving reference magnetic resonance image data (126) as input. The synthetic magnetic resonance image data is a simulation of magnetic resonance image data acquired according to a first configuration of a magnetic resonance imaging system when the reference magnetic resonance image data is acquired according to a second configuration of the magnetic resonance imaging system. Execution of the machine executable instructions causes a computational system (106) to: receive (200) measured k-space data (124) acquired according to the first configuration of the magnetic resonance imaging system; receive (202) the reference magnetic resonance image data; receive (204) the synthetic magnetic resonance image data by inputting the reference magnetic resonance image data into the image generating neural network; and reconstruct (206) corrected magnetic resonance image data (132) from the measured k-space data and the synthetic magnetic resonance image data.