Enhancing intravoxel incoherent motion parameter mapping in the brain using k-b PCA.

Area of Science:

• Magnetic Resonance Imaging (MRI)
• Medical Physics
• Neuroimaging

Background:

• Intravoxel Incoherent Motion (IVIM) imaging provides valuable diffusion and perfusion parameters in the brain.
• Parallel imaging in IVIM MRI can amplify local noise, degrading image quality and parameter accuracy.
• Existing reconstruction methods struggle with significant undersampling, limiting scan efficiency.

Purpose of the Study:

• To develop and evaluate a constrained image reconstruction approach, k-b PCA, to address noise amplification in undersampled IVIM brain imaging.
• To compare the performance of k-b PCA against conventional CG-SENSE reconstruction for both retrospective and prospective undersampling.
• To assess the impact of k-b PCA on the accuracy of IVIM parameter mapping (D, f, D*).

Main Methods:

• IVIM imaging was performed on a 3T MR system with multi-b-value diffusion encoding.
• Data were retrospectively undersampled (2-6x) and reconstructed using CG-SENSE and the proposed k-b PCA.
• Prospective single-shot echo planar imaging (EPI) acquisition with k-b PCA was also performed in volunteers.

Main Results:

• k-b PCA significantly outperformed CG-SENSE, reducing reconstruction errors, especially at higher undersampling factors (≥3x).
• k-b PCA effectively removed undersampling artifacts up to sixfold undersampling, with median relative errors for image magnitude and IVIM parameters substantially lower than CG-SENSE.
• Prospective undersampled acquisitions using k-b PCA demonstrated reduced image artifacts and noise compared to parallel imaging alone.

Conclusions:

• Constrained reconstruction using k-b PCA enhances IVIM parameter mapping accuracy from undersampled MRI data.
• The k-b PCA method offers a robust solution for accelerated IVIM acquisition, improving scan efficiency without compromising image quality.
• This approach facilitates more reliable diffusion and perfusion quantification in neuroimaging applications.