Related Experiment Videos
A rapid look-up table method for reconstructing MR images from arbitrary K-space trajectories
1Department of Radiology and Biomedical Engineering, Case Western Reserve University and University Hospitals of Cleveland, OH 44106, USA.
IEEE Transactions on Medical Imaging
|May 9, 2001
Summary
Look-up tables (LUTs) significantly accelerate k-space data reconstruction. These methods enable real-time gridding for faster image processing.
Area of Science:
- Medical imaging
- Computational algorithms
- Signal processing
Background:
- Look-up tables (LUTs) are widely used to enhance algorithm speed.
- Extending LUTs to reconstruct nonuniformly sampled k-space data is a key challenge.
Purpose of the Study:
- To implement and evaluate LUT-based gridding algorithms for k-space data reconstruction.
- To assess the speed improvements offered by LUTs compared to traditional methods.
Main Methods:
- Implemented discrete Fourier transform (DFT) and convolution-based gridding algorithms using LUTs in C++.
- Utilized precalculated weight arrays for DFT and k-space neighborhood weighting for convolution.
- Tested on a radial k-space acquisition sequence (180 views, 256 points/view).
Main Results:
- LUT-based gridding achieved reconstruction in 36.2 ms (approx. 800 ns/point).
- This represents a 45x speed improvement over non-LUT methods (1639.2 ms).
- Larger convolution kernels (4x4 Kaiser-Bessel) were feasible with LUTs.
Conclusions:
- LUTs offer substantial speedups for k-space data gridding and reconstruction.
- These methods facilitate concurrent acquisition and gridding for real-time processing.
- Future real-time image reconstruction is enabled by these table-based computations.