AI K-Space Reconstruction from Sparse MRI Scans
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing MR imaging techniques using sparse-sampled acquisition schemes suffer from reduced MR image quality due to a lack of spatial information, while full sampling results in ghosting and artifacts.
Innovation Solution
A convolutional neural network-based AI engine is trained to reconstruct a complete MR K-space data matrix from multiple incomplete matrices using sparse-sampled scans, employing unique sampling patterns and supervised learning to improve image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If sparse sampling is used to accelerate MRI acquisition, then acquisition time is reduced, but image quality deteriorates due to lack of spatial information
Solution Approach 1:
The patent divides the K-space data into multiple incomplete matrices from different sparse-sampled scans, processes them through separate AI reconstruction paths, and combines the results to form a complete reconstructed image. This segmentation allows parallel processing and leverages multiple low-quality measurements to achieve high-quality output.
Solution Approach 2:
The patent introduces an AI engine as an intermediary component that receives incomplete K-space data matrices, applies deep learning models to reconstruct complete matrices, and generates high-quality MR images. This intermediary transforms the relationship between sparse sampling and image quality by using learned patterns to fill in missing information.
2Manufacturing precision
If full K-space sampling is used to ensure complete spatial information, then image quality is maintained, but acquisition time increases
Solution Approach 1:
The patent applies partial action by using sparse-sampled scans that acquire only a subset of K-space data (incomplete matrices). Instead of requiring full sampling, the system uses multiple partial measurements and combines them through AI reconstruction to achieve quality equivalent to full sampling, thereby reducing acquisition time.
3Productivity
If sparse sampling is used to reduce the number of readouts, then scan time is shortened, but ghosting and artifacts increase
Solution Approach 1:
The patent converts the harmful effects of sparse sampling (ghosting and artifacts) into beneficial outcomes by using AI reconstruction. The AI engine learns to recognize and eliminate these artifacts while preserving genuine image features, transforming the problematic incomplete data into high-quality images that would not be possible through conventional reconstruction methods.
Solution Approach 2:
The patent employs feedback mechanisms in the AI training process, where the model is exposed to ground truth images and adjusts its reconstruction parameters based on the difference between predicted and actual images. This feedback loop enables the system to iteratively improve artifact removal and image quality enhancement.
Data Source
AI summary
A first artificial intelligence (AI) engine receives a plurality of incomplete magnetic resonance (MR) K-space data matrices of an object scanned by an MR device. Each of the incomplete MR K-space data matrices comprises complex values and is the result of a corresponding san of the object by the MR device using a sparse-sampled MR scan acquisition sequence. Each sparse-sample MR scan acquisition sequence employs a unique sampling pattern. The first AI engine reconstructs a complete MR K-space data matrix of the scanned object, corresponding to a complete MR K-space acquisition. The reconstruction is based on the data in the plurality of incomplete MR K-space data matrices.