Accelerated reconstruction of dictionary-based T₂ relaxation maps based on dictionary compression and gradient...

Area of Science:

• Medical Imaging
• Quantitative MRI
• Image Reconstruction

Background:

• Quantitative T2-relaxation maps are vital for clinical diagnosis and monitoring.
• Current methods for generating T2 maps are hindered by lengthy acquisition and processing times.
• The EMC platform simulates signal curves for T2 mapping but requires intensive computation.

Purpose of the Study:

• To present and evaluate methods for accelerating the reconstruction of EMC-based T2 relaxation maps.
• To reduce the computationally intensive post-processing associated with EMC T2 mapping.
• To maintain the accuracy of T2 values despite accelerated reconstruction.

Main Methods:

• Two post-processing acceleration approaches were investigated: dictionary compression via principal component analysis (PCA) and a gradient-descent search algorithm.
• The optimal MATLAB C++ compiler was identified to further enhance processing speed.
• Relative error was calculated to assess the accuracy of the proposed acceleration techniques.

Main Results:

• The gradient-descent method achieved perfect agreement with exhaustive search matching.
• PCA-based acceleration resulted in a root mean square error (RMSE) of up to 4% compared to exhaustive matching.
• A significant overall acceleration of 16x was achieved with gradient descent, plus an additional 7x acceleration using the optimal compiler.

Conclusions:

• Post-processing of EMC-based T2 relaxation maps can be substantially accelerated.
• The proposed methods accelerate reconstruction without compromising the accuracy of T2 values.
• Faster T2 map generation facilitates improved clinical diagnosis and patient management.