Pulseq Sequences for Optimized 3D Fast Spin Echo at 7T

This repository contains the Pulseq .seq files associated with the publication:

Peter Dawood et al., “Controlling sharpness, SNR, and specific absorption rate for 3D fast-spin echo at 7T by end-to-end learning,” Magnetic Resonance in Medicine, 2025.
DOI: 10.1002/mrm.30533

---

Overview

The provided sequences were developed within the PyPulseq framework (version 1.3.1post1). They implement a linear reordered T2-weighted 3D Fast Spin Echo sequence with variable flip angle (VFA) trains. The VFAs were optimized using an end-to-end learning framework to balance image sharpness (PSF), signal-to-noise ratio (SNR), and SAR constraints.

The repository contains both imaging .seq files and .seq files for unencoded signal acquisition (i.e., echo train signal responses without phase/partition encoding), which were used for PSF quantification.

Imaging parameters and experimental details are given in the publication.

---

File Structure

/seq_files
    psf_optimized.seq           # VFA scheme optimized for point-spread function
    snr_optimized.seq           # VFA scheme optimized for SNR
    standard.seq                # Standard VFA scheme
    trade_off_optimized.seq     # VFA scheme for trade-off between PSF and SNR
    *_unencoded.seq             # For acquisition of corresponding unencoded signals 

System Limits

The following system limitations were used when generating the sequences:

from pypulseq.opts import Opts 

# set system limitations 
sys = Opts(max_grad=65, grad_unit='mT/m',
        max_slew=150, slew_unit='T/m/s',
        rf_ringdown_time=20e-6, rf_dead_time=100e-6,
        adc_dead_time=10e-6)