Image reconstruction of compressed sensing MRI using graph-based redundant wavelet transform.

Area of Science:

• Medical Imaging
• Signal Processing
• Computer Vision

Background:

• Compressed sensing magnetic resonance imaging (CS-MRI) accelerates image acquisition by leveraging sparse image representations.
• The choice of sparsification method significantly impacts the quality of reconstructed MRI images.
• Existing methods face challenges in artifact reduction and minimizing reconstruction errors.

Purpose of the Study:

• To introduce a novel graph-based redundant wavelet transform for sparse representation in iterative MRI reconstructions.
• To improve the quality of accelerated magnetic resonance imaging through enhanced sparse representation.
• To evaluate the performance of the proposed method against state-of-the-art reconstruction techniques.

Main Methods:

• Developed a graph-based redundant wavelet transform where image patches are vertices and differences are edges.
• Utilized a shortest path algorithm on the graph to minimize the total difference among image patches.
• Employed an l1 norm regularized formulation solved via an alternating-direction minimization with continuation algorithm.

Main Results:

• The proposed graph-based transform effectively sparsifies magnetic resonance images for iterative reconstruction.
• Experimental results show superior performance compared to several state-of-the-art reconstruction methods.
• Demonstrated significant improvements in artifact removal and reduction of reconstruction errors on tested datasets.

Conclusions:

• The graph-based redundant wavelet transform offers a promising approach for sparse representation in CS-MRI.
• This method enhances reconstruction quality, reduces artifacts, and minimizes errors in accelerated MRI.
• The findings suggest a potential advancement for faster and more accurate magnetic resonance imaging.