Single anatomical EPI with SNAKE-fMRI

This examples walks through the elementary components of SNAKE.

Here we proceed step by step and use the Python interface. A more integrated alternative is to use the CLI snake-main

# Imports
import numpy as np

from snake.core.phantom import Phantom
from snake.core.sampling import EPI3dAcquisitionSampler
from snake.core.simulation import GreConfig, SimConfig, default_hardware
from snake.toolkit.plotting import axis3dcut

Setting up the base simulation Config. This configuration holds all key parameters for the simulation, describing the scanner parameters.

sim_conf = SimConfig(
    max_sim_time=6,
    seq=GreConfig(TR=50, TE=30, FA=3),
    hardware=default_hardware,
)
sim_conf.hardware.n_coils = 8
sim_conf.fov.res_mm = (3, 3, 3)
sim_conf

SimConfig

max_sim_time(float)	6
seq(GreConfig)	GreConfig TR (float) TE (float) FA (float) 50 30 3
hardware(HardwareConfig)	HardwareConfig gmax (float) smax (float) n_coils (int) dwell_time_ms (float) raster_time_ms (float) field (float) 40 200 8 0.001 0.005 3.0
fov(FOVConfig)	FOVConfig size (ThreeFloats) offset (ThreeFloats) angles (ThreeFloats) res_mm (ThreeFloats) (181, 217, 181) (-90.25, -126.25, -72.25) (0, 0, 0) (3, 3, 3)
rng_seed(int)	19290506

Creating the base Phantom

The simulation acquires the data describe in a phantom. A phantom consists of fuzzy segmentation of head tissue, and their MR intrinsic parameters (density, T1, T2, T2*, magnetic susceptibilities)

Here we use Brainweb reference mask and values for convenience.

phantom = Phantom.from_brainweb(
    sub_id=4, sim_conf=sim_conf, output_res=1, tissue_file="tissue_7T"
)

# Here are the tissue availables and their parameters
phantom.affine

No matrix size found in the header. The header is probably missing.

array([[1., 0., 0., 0.],
       [0., 1., 0., 0.],
       [0., 0., 1., 0.],
       [0., 0., 0., 1.]], dtype=float32)

Setting up Acquisition Pattern and Initializing Result file.

# The next piece of simulation is the acquisition trajectory.
# Here nothing fancy, we are using a EPI (fully sampled), that samples a 3D
# k-space (this akin to the 3D EPI sequence of XXXX)

sampler = EPI3dAcquisitionSampler(accelz=1, acsz=0.1, orderz="top-down")

Acquisition with Cartesian Engine

The generated file example_EPI.mrd does not contains any k-space data for now, only the sampling trajectory. let’s put some in. In order to do so, we need to setup the acquisition engine that models the MR physics, and get sampled at the specified k-space trajectory.

SNAKE comes with two models for the MR Physics:

• model=”simple” :: Each k-space shot acquires a constant signal, which is the image contrast at TE.

• model=”T2s” :: Each k-space shot is degraded by the T2* decay induced by each tissue.

# Here we will use the "simple" model, which is faster.
#
# SNAKE's Engine are capable of simulating the data in parallel, by distributing
# the shots to be acquired to a set of processes. To do so , we need to specify
# the number of jobs that will run in parallel, as well as the size of a job.
# Setting the job size and the number of jobs can have a great impact on total
# runtime and memory consumption.
#
# Here, we have a single frame to acquire with 60 frames (one EPI per slice), so
# a single worker will do.

from snake.core.engine import EPIAcquisitionEngine


engine = EPIAcquisitionEngine(model="simple")

engine(
    "example_EPI.mrd",
    sampler=sampler,
    phantom=phantom,
    sim_conf=sim_conf,
    worker_chunk_size=16,
    n_workers=4,
)

Traceback (most recent call last):
  File "/volatile/github-ci-mind-inria/gpu_mind_runner/_work/snake-fmri/snake-fmri/examples/anatomical/example_anat_EPI.py", line 98, in <module>
    engine(
  File "/volatile/github-ci-mind-inria/gpu_mind_runner/_work/snake-fmri/snake-fmri/src/snake/core/engine/base.py", line 299, in __call__
    raise exc
  File "/volatile/github-ci-mind-inria/gpu_mind_runner/_work/snake-fmri/snake-fmri/src/snake/core/engine/base.py", line 296, in __call__
    f_chunk = str(future.result())
  File "/volatile/github-ci-mind-inria/gpu_mind_runner/_work/_tool/Python/3.10.16/x64/lib/python3.10/concurrent/futures/_base.py", line 451, in result
    return self.__get_result()
  File "/volatile/github-ci-mind-inria/gpu_mind_runner/_work/_tool/Python/3.10.16/x64/lib/python3.10/concurrent/futures/_base.py", line 403, in __get_result
    raise self._exception
IndexError: index 1 is out of bounds for axis 0 with size 1

Simple reconstruction

Getting k-space data is nice, but SNAKE also provides rudimentary reconstruction tools to get images (and check that we didn’t mess up the acquisition process). This is available in the companion package snake.toolkit.

Loading the .mrd file to retrieve all information can be done using the ismrmd python package, but SNAKE provides convenient dataloaders, which are more efficient, and take cares of managing underlying files access. As we are showcasing the API, we will do things manually here, and use only core SNAKE.

from snake.mrd_utils import CartesianFrameDataLoader

with CartesianFrameDataLoader("example_EPI.mrd") as data_loader:
    mask, kspace_data = data_loader.get_kspace_frame(0)

with CartesianFrameDataLoader("example_EPI.mrd") as data_loader:
    phantom = data_loader.get_phantom()

sim_conf.fov.affine

Reconstructing a Single Frame of fully sampled EPI boils down to performing a 3D IFFT:

from scipy.fft import fftshift, ifftn, ifftshift

axes = (-3, -2, -1)
image_data = ifftshift(
    ifftn(fftshift(kspace_data, axes=axes), axes=axes, norm="ortho"), axes=axes
)

# Take the square root sum of squares to get the magnitude image (SSOS)
image_data = np.sqrt(np.sum(np.abs(image_data) ** 2, axis=0))

import matplotlib.pyplot as plt


fig, ax = plt.subplots()

axis3dcut(image_data.squeeze().T, None, None, cbar=False, cuts=(0.5, 0.5, 0.5), ax=ax)
plt.show()