RARE - josalggui/MaRGE GitHub Wiki

MRI Sequence Documentation: 3D Imaging with RARE

Overview

This sequence is designed to acquire 3D images using the Rapid Acquisition with Relaxation Enhancement (RARE) technique. The sequence involves multiple echo trains. The acquisition process includes noise measurement before each excitation to ensure optimal signal-to-noise ratio (SNR).

RARE

Sequence Description

The sequence generates 3D images by acquiring data in the k-space for all three spatial dimensions (x, y, z). The scan is repeated multiple times and averaged to improve image quality. The acquisition involves both a pre-excitation and inversion recovery pulse, with refocusing pulses interleaved in each repetition.

Key Components of the Sequence:

  1. Noise Measurement: A noise measurement is performed before each excitation to estimate the baseline noise level. 2Matrix Size: The matrix size must be even for all directions (readout, phase, slice). 3Rotation and Axis Configuration: The sequence provides flexibility in selecting the rotation axis and defining the direction of readout, phase, and slice axes (TODO).

Sequence Steps:

  1. Excitation and Refocusing: Each scan starts with an excitation RF pulse, followed by one or more refocusing pulses to generate echoes. The angle of excitation and refocusing pulses can be adjusted to optimize image contrast.
  2. Echo Train Acquisition: The sequence acquires multiple echoes per repetition, and the length of the echo train is user-configurable.
  3. Phase Encoding and Readout: The sequence uses phase encoding for spatial localization in the phase direction, while the readout gradient determines the k-space sampling along the readout direction.

Parameters

The following parameters control the behavior of the MRI sequence. Adjusting these settings affects the quality and speed of the imaging process.

  • Number of scans: The number of scans to average for improving image quality. More scans lead to a higher SNR.
  • FOV[x, y, z] (cm): Defines the field of view along the x, y, and z axes of the scanner in centimeters.
  • dFOV[x, y, z] (mm): Displacement of the field of view from the center of the previously selected image in millimeters.
  • nPoints[rd, ph, sl]: Matrix size along readout (rd), phase (ph), and slice (sl) directions. These must be even numbers.
  • Angle(º): Rotation angle with respect to the selected rotation axis in degrees.
  • Rotation axis: Axis around which the image is rotated, specified as (x, y, z).
  • Axis[rd, ph, sl]: Defines the axis for readout, phase, and slice directions. For example, [2, 1, 0] means that z is for readout, y for phase, and x for slice.
  • Larmor frequency offset (kHz): Frequency offset in kHz from the calibrated Larmor frequency (stored in configs/hw_config.py).
  • Excitation flip angle (º): The flip angle for the excitation pulse.
  • Refocusing flip angle (º): The flip angle for the refocusing pulse.
  • RF excitation time (us): Duration of the excitation RF pulse in microseconds.
  • RF refocusing time (us): Duration of the refocusing RF pulse in microseconds.
  • Echo spacing (ms): Time between echoes in milliseconds.
  • Pre-excitation time (ms): Time between a 90º pulse and the excitation pulse in milliseconds. Set to 0 to skip this RF pulse.
  • Inversion time (ms): Time between the 180º pulse and the excitation pulse in milliseconds. Set to 0 to skip this RF pulse.
  • Repetition time (ms): Time between excitations in milliseconds.
  • Echo train length: The number of echoes acquired per repetition.
  • Acquisition time (ms): The total acquisition window duration in milliseconds.
  • Sweep mode: Select phase acquisition order:
    • 0 -> From -kmax to kmax
    • 1 -> From 0 to kmax
    • 2 -> From kmax to 0
  • Dummy pulses: The number of dummy pulses used to reach a steady state.
  • Rd gradient time (ms): Duration of the readout gradient in milliseconds. Recommended to be slightly longer than the acquisition time.
  • Rd dephasing time (ms): Duration of the dephasing readout gradient in milliseconds.
  • Ph gradient time (ms): Duration of phase and slice gradient pulses for de-phasing and re-phasing.
  • Rd preemphasis: The factor by which the de-phasing readout gradient is multiplied to center the echoes.
  • RF phase (º): The phase of the excitation RF pulse in degrees.
  • Shimming: Offset applied to the three gradient axes for field uniformity correction.
  • Partial fourier fraction: A fraction between 0.5 and 1 to select the portion of slices to acquire.
  • Echo time shift (ms): A shift in the acquisition window in milliseconds.
  • Unlock image orientation: Set to 1 to show images in their raw orientation or 0 for the Physio V1.0 scanner orientation.

Important Notes:

  • nPoints must always be even numbers for consistency in the Fourier transform process.
  • The echo train length (etl) should be chosen such that the number of phases divided by etl is an even number.
  • An exception occurs when the etl equals the number of phases.
  • A schematic of the sweep modes along the phase direction can be seen in the image above.

SequenceRun and SequenceAnalysis

sequenceRun:

The sequenceRun function is responsible for executing the MRI sequence. The function controls the flow of data collection, managing the timing and synchronization of RF pulses, gradients, and acquisitions. The run is repeated based on the user-defined number of scans and other parameters like repetition time and echo train length.

sequenceAnalysis:

Once the data is acquired, sequenceAnalysis processes the raw data. The primary steps in the analysis include: 1Fourier Transform: The k-space data is transformed into the image domain via a Fourier transform. 2Reconstruction: After processing, the image is reconstructed, correcting for any phase or frequency discrepancies.

Output and Results

The final output consists of 3D images that represent the scanned object. These images are saved in standard formats like DICOM, which include detailed metadata such as slice thickness, field of view, and other scanning parameters. Additionally, intermediate data like k-space matrices and reconstruction logs may be stored for further analysis.