MR-MED: A Time-Efficient Multi-Echo DWI Sequence for Prostate Luminal Water Imaging
Abstract
Purpose
To develop and evaluate a time-efficient pulse sequence – MR-MED (multi-readout reduced field-of-view (rFOV) multi-spin-echo EPI diffusion-weighted imaging (MR-MED) sequence for prostate luminal water imaging (LWI). This sequence acquires diffusion-weighted signals at multiple echo times following a single excitation, reducing scan time by a factor of three compared to conventional techniques. MR-MED aims to improve quantitative assessment of luminal water fraction (LWF), a biomarker of aggressive prostate cancer, and enhance the clinical feasibility of LWI for diagnosis and risk stratification.
Methods
The MR-MED sequence was validated on phantoms with SNR analysis, followed by evaluations on healthy volunteers at 3 Tesla. In both phantom and human scans, MR-MED was compared to a conventional EPI sequence using similar parameters with triple the scan time. A custom reconstruction pipeline with Marchenko-Pastur Principal Component Analysis-based denoising was employed before extracting five LWI parameters using nonlinear least-squares fitting.
Results
MR-MED achieved sufficient SNR (>15) with a total scan time of approximately 4.5 minutes at b-values up to 1500 s/mm2 and TE over 100 ms, supporting its suitability for quantitative diffusion-relaxometry modeling for LWI. While the SNR was lower in MR-MED compared to the conventional acquisition—as expected due to reduced scan time—the 30% SNR reduction is consistent with theoretical predictions for rFOV acquisitions. In human scans, MR-MED yielded reasonable luminal fractions between 14% and 22%, consistent with literature for the study cohort. Further, stable parameter estimates of T2 and ADC across volunteers in MR-MED suggest that MR-MED can produce reliable quantitative results and be ready for possible clinical deployment.
Conclusion
This study presents MR-MED, a time-efficient, multi-echo DWI sequence for LWI of the prostate that combines rFOV excitation, multiple spin echoes, and multi-echo EPI readouts to achieve a three-fold acceleration over conventional acquisition while preserving spatial resolution and enabling robust quantitative modeling.