MR image reconstruction from generalized projections.

Area of Science:

• Medical Imaging
• Computational Imaging
• Signal Processing

Background:

• Multidimensional imaging trajectories with multichannel systems and nonlinear gradients result in lengthy image reconstruction times.
• Existing methods face computational complexity challenges, hindering efficient data processing.

Purpose of the Study:

• To present novel methods for reducing the computational complexity of iterative time-domain image reconstruction algorithms.
• To decrease the complexity from O(N(4)) to O(N(3)) for faster processing.

Main Methods:

• Two distinct methods are introduced, differing in filter incorporation within the reconstruction algorithm.
• Method 1: A weighted time-domain approach without thresholding.
• Method 2: An equivalent method to the time-domain approach.

Main Results:

• Image reconstruction speed increased by up to two orders of magnitude for high-resolution data.
• Method 1 (weighted reconstruction) showed less sensitivity to thresholding but required more iterations.
• Method 2 demonstrated efficiency and speed improvements.

Conclusions:

• Fast and accurate image reconstruction methods are provided for complex spatial encoding strategies.
• These methods are particularly efficient for high-resolution anatomical imaging, especially with nonlinear gradient fields.