Application of perceptual difference model on regularization techniques of parallel MR imaging.

Area of Science:

• Medical Imaging
• Magnetic Resonance Imaging (MRI)
• Image Reconstruction

Background:

• Parallel MRI accelerates imaging or enhances signal-to-noise ratio (SNR) using multiple receiver coils.
• Image quality is often limited by sensitivity map inaccuracies, k-space noise, and ill-conditioned matrices.
• Tikhonov regularization is a common technique to address ill-conditioning, requiring careful selection of regularization maps and parameters.

Purpose of the Study:

• To evaluate the suitability of the Perceptual Difference Model (PDM) for assessing image quality in parallel MRI.
• To compare different methods for selecting regularization maps and parameters in parallel MRI reconstruction.
• To optimize parameters within adaptive regularization methods using PDM as an objective function.

Main Methods:

• Utilized the Perceptual Difference Model (PDM), a quantitative image quality assessment tool.
• Compared four distinct methods for selecting the regularization map and four for the regularization parameter.
• Applied PDM as an objective function to optimize parameters in a spatially adaptive regularization method.

Main Results:

• PDM demonstrated a high correlation with human image quality ratings, confirming its suitability for parallel MRI.
• The best image reconstructions were achieved using a generalized series (GS) regularization map combined with a spatially adaptive regularization parameter.
• PDM effectively optimized key parameters within the spatially adaptive regularization technique.

Conclusions:

• The Perceptual Difference Model (PDM) is a valuable tool for comprehensive experimentation in parallel MRI.
• PDM facilitates the design and optimization of advanced reconstruction methods for parallel MRI.
• The combination of GS regularization and spatially adaptive parameters, guided by PDM, yields superior parallel MRI reconstructions.