Clinical evaluation of free-breathing cardiac multi-parametric mapping using dictionary-based motion correction

Area of Science:

• Cardiovascular Imaging
• Medical Physics
• Magnetic Resonance Imaging

Background:

• Accurate motion correction (MoCo) is crucial for clinically important free-breathing cardiac multi-parametric mapping.
• Existing dictionary-matching and low-rank (DM + LR) MoCo methods face limitations in computational efficiency and clinical validation.
• A modified DM + LR (mDM + LR) approach was developed to enhance computational efficiency and diagnostic performance in free-breathing cardiac T1/T2 mapping.

Purpose of the Study:

• To develop and validate a modified dictionary-matching and low-rank (mDM + LR) motion correction (MoCo) technique.
• To improve computational efficiency and diagnostic performance for free-breathing cardiac T1/T2 mapping.
• To assess the clinical applicability of the mDM + LR MoCo approach.

Main Methods:

• A prospective study included 130 patients with cardiac diseases and 23 healthy controls undergoing 3T cardiac MRI for free-breathing T1/T2 mapping.
• The mDM + LR method utilized a pre-trained multi-layer perceptron (MLP) for signal mapping, achieving a processing time of approximately 25 seconds per sample.
• MoCo accuracy was evaluated against non-MoCo and pTVreg using quantitative metrics (DSC, MCD, RDME), qualitative map scores, and T1/T2 quantification against breath-hold references.

Main Results:

• The mDM + LR method demonstrated superior quantitative performance (higher DSC, lower MCD, lower RDME) and qualitative map scores compared to non-MoCo and pTVreg.
• mDM + LR achieved higher correlations with breath-hold reference values for global T1/T2 quantification (r=0.81/0.80).
• Diagnostic performance was significantly improved with mDM + LR, showing higher specificity (93%/100%) and accuracy (89%/100%) compared to other methods.

Conclusions:

• The modified dictionary-matching and low-rank (mDM + LR) method significantly enhances motion correction in free-breathing cardiac multi-parametric mapping.
• The mDM + LR approach improves quantification accuracy and diagnostic performance.
• The mDM + LR technique is computationally efficient and suitable for clinical application in cardiac T1/T2 mapping.