Single-shot T₂ mapping via multi-echo-train multiple overlapping-echo detachment planar imaging and multitask deep...

Area of Science:

• Magnetic Resonance Imaging
• Biomarker Development
• Quantitative Imaging

Background:

• Quantitative MRI offers valuable clinical biomarkers but suffers from long scan times and motion artifacts.
• Existing single-shot T2 mapping methods like MOLED have limited accuracy due to a narrow echo time range, especially for tissues with long T2 values.

Purpose of the Study:

• To develop a novel single-shot method, METMOLED, for accurate T2 quantification across a wide range of T2 values.
• To address signal degeneration caused by refocusing pulse imperfections without additional measurements.

Main Methods:

• Integrated multiple echo-train techniques into MOLED to extend the echo time range.
• Employed multitask deep learning (U-Net) for synchronous reconstruction of T2, B1, and spin density maps from METMOLED data.
• Trained the network on synthetic data to learn the signal-to-map relationship and correct for pulse imperfections.

Main Results:

• METMOLED demonstrated improved quantitative accuracy and tissue detail compared to MOLED in digital brains, phantoms, and human brains.
• Significant improvements were observed in tissues with long T2 values and in high temporal resolution applications.
• The method effectively corrected signal deviations from non-ideal refocusing pulses and showed high repeatability (CV=1.65%).

Conclusions:

• METMOLED overcomes echo-train length limitations, enabling unbiased T2 quantification over an extensive range.
• The technique corrects for refocusing pulse inaccuracies without extra measurements or post-processing, maintaining its single-shot nature.
• METMOLED offers a promising advancement for accurate and efficient T2 quantification in clinical MRI.