A Physics-Based Algorithm to Universally Standardize Routinely Obtained Clinical T₂-Weighted Images.

Area of Science:

• Medical Imaging
• Quantitative MRI
• Machine Learning in Radiology

Background:

• Clinical T2-weighted (T2w) MRI data suffers from site-specific variations due to hardware and acquisition parameters.
• Quantitative T2 mapping offers standardization but faces slow adoption in clinical practice.
• Existing reconstruction methods struggle to derive quantitative maps from routine T2w data.

Purpose of the Study:

• To demonstrate that routine T2-weighted (T2w) complex raw MRI datasets can be utilized as quantitative mapping sequences.
• To introduce a novel constrained reconstruction method for quantitative T2 mapping from standard T2w images.
• To enable the creation of large, standardized MRI datasets without altering clinical protocols.

Main Methods:

• Developed and applied the expanding-constrained alternating minimization for parameter mapping (e-CAMP) algorithm.
• e-CAMP utilizes stepwise initialization, a linearized exponential model, and a phase conjugacy constraint.
• Validated the method using simulations, retrospectively undersampled brain images, and prospectively acquired phantom and brain T2w images.

Main Results:

• The e-CAMP method successfully reconstructs quantitative T2 maps directly from vendor T2w single image data.
• Demonstrated the method's performance across various datasets, including phantom and in-vivo brain imaging.
• Quantitative T2 maps derived from routine T2w data are shown to be accurate and reliable.

Conclusions:

• Routine T2w MRI datasets can be repurposed for quantitative T2 mapping.
• The e-CAMP approach facilitates the generation of standardized quantitative MRI data.
• This method has the potential to significantly expand large-scale MRI data analysis capabilities without protocol modifications.