Compressed Sensing fMRI via Variable-Density Spiral Sampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Achieving high spatial resolution in functional magnetic resonance imaging (fMRI) while maintaining temporal resolution and contrast-to-noise ratio (CNR) is challenging due to the trade-offs inherent in existing methods.
Innovation Solution
The method employs a randomized, variable-density spiral acquisition scheme with a high spatial sparsifying transform and fast iterative shrinkage thresholding algorithm (FISTA) for real-time fMRI, utilizing a discrete cosine transform (DCT) and optimizing regularization parameters to achieve high acceleration factors and improved CNR.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high spatial resolution is achieved through conventional fMRI methods, then spatial resolution is improved, but temporal resolution decreases and contrast-to-noise ratio decreases
Solution Approach 1:
The patent applies partial sampling in k-space by acquiring only a subset of k-space lines according to a randomized variable-density spiral trajectory. This partial action allows acceleration by not collecting all Nyquist-required samples, while compressed sensing reconstruction recovers the full image from these reduced samples, thereby improving temporal resolution while maintaining spatial resolution.
Solution Approach 2:
The patent changes the sampling trajectory parameters from conventional Cartesian or spiral grids to a randomized variable-density spiral pattern. This parameter change in the acquisition scheme enables higher acceleration factors by optimizing where samples are taken in k-space, allowing faster acquisition without sacrificing spatial resolution when combined with compressed sensing reconstruction.
2Measurement precision
If high spatial resolution is achieved through conventional fMRI methods, then spatial resolution is improved, but contrast-to-noise ratio decreases
Solution Approach 1:
The patent introduces compressed sensing reconstruction as an intermediary processing step between partial k-space sampling and final image formation. This intermediary uses the spatial sparsifying transform and iterative optimization to recover high-quality images from undersampled data, thereby maintaining contrast-to-noise ratio despite reduced sampling that enables faster acquisition.
Solution Approach 2:
The patent applies a spatial sparsifying transform (such as wavelet transform or total variation regularization) as a preliminary step in the reconstruction process. This preliminary action identifies and preserves important image features while discarding redundant information, which improves contrast-to-noise ratio by enhancing meaningful signals relative to noise in the reconstructed high-resolution images.
3Productivity
If sampling acceleration is increased, then temporal resolution is improved, but image quality and contrast-to-noise ratio worsen
Solution Approach 1:
The patent implements an iterative feedback mechanism in the compressed sensing reconstruction algorithm (such as FISTA or conjugate gradient). The reconstruction process repeatedly refines the image estimate by comparing reconstructed images with the acquired undersampled data and adjusting parameters to minimize error, thereby maintaining contrast-to-noise ratio even at high acceleration factors where simple sampling would fail.
4Ease of manufacture
If conventional sampling schemes are used, then acquisition is simpler, but temporal resolution and acceleration factor are limited
Solution Approach 1:
The patent employs a dynamic randomized variable-density spiral sampling trajectory that adapts sampling density across different k-space regions and time points. This dynamic approach concentrates samples where most information is needed (central k-space) while using fewer samples in peripheral regions, enabling higher acceleration factors and improved temporal resolution compared to static uniform sampling schemes.
Data Source
Figure 1A~1E
Figure 1F
Figure 2A
AI summary
The present disclosure provides methods and systems for high-resolution functional magnetic resonance imaging (fMRI), including real-time high-resolution functional MRI methods and systems.