Low-Field MRI Distortion Correction Using Paired High-Resolution Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
MRI systems with high magnetic fields pose challenges for surgical interventions due to limited physical access and restrictions on electrical and mechanical components, necessitating improved image distortion correction for low-field MRIs.
Innovation Solution
A system using a pre-trained generative adversarial network to correct image distortions in low-field MRIs by inputting low-field MRIs into a generator model paired with high-resolution images, allowing for real-time distortion correction during surgical procedures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-field strength MRI systems are used, then image quality and resolution are improved, but physical access to the patient and usage of electrical/mechanical components are restricted
Solution Approach 1:
The system separates the imaging function into two independent parts: a low-field MRI scanner that provides physical access and a machine learning model that performs distortion correction. This segmentation allows the MRI system to operate in a low-field configuration for accessibility while using computational methods to achieve high-quality images.
Solution Approach 2:
The patent replaces the mechanical constraint of high-field MRI hardware with a computational solution. Instead of using a high-field scanner that physically restricts access, the system uses a low-field scanner combined with a generative adversarial network (GAN) to correct distortions, substituting mechanical limitations with algorithmic processing.
2Ease of operation
If low-field strength MRI systems are used, then physical access and component usage are improved, but image distortion increases
Solution Approach 1:
The system incorporates feedback through the discriminator model in the GAN, which continuously evaluates the generated images and provides feedback to the generator model. This feedback mechanism iteratively improves the distortion correction, ensuring that the low-field MRI images are accurately corrected to match the reference high-resolution images.
Solution Approach 2:
The generator model creates synthetic high-resolution images by copying and transforming the low-field MRI images. It learns the mapping between distorted low-field images and their corresponding undistorted high-resolution versions, effectively copying the structural information while correcting the distortions.
3Measurement precision
If machine learning models are trained with paired high-resolution images, then distortion correction accuracy is improved, but training time and computational resources increase
Solution Approach 1:
The system performs preliminary action by pre-training the GAN model using paired high-resolution and low-field MRI images before actual surgical procedures. This pre-training phase captures the distortion patterns and correction mappings in advance, allowing the model to be ready for rapid inference during time-critical surgical interventions.
Solution Approach 2:
The patent addresses training time by optimizing model parameters such as the number of training epochs, batch sizes, and learning rates. It also utilizes techniques like image pairing strategies and resolution adjustment to improve training efficiency while maintaining correction accuracy.
Data Source
AI summary
Disclosed is a system comprising a database storing a preoperative high-resolution image of an object of interest and a control circuit. The control circuit comprises a processor and a memory. The memory stores instructions executable by the processor to obtain a low-field strength magnetic resonance image of the object of interest. The memory stores further instructions executable by the processor to input the low-field strength MRI of the object of interest into a generator model of a pre-trained generative adversarial network. The generator model is pre-trained with low-field strength MRIs and paired high-resolution images to correct image distortions. The memory stores further instructions executable by the processor to output a distortion-corrected image of the object of interest from the generator model based on the low-field strength MRI and transmit the distortion-corrected image of the object of interest to a user interface.