3D DCE-MRI in the liver

Fit a linear 2-compartment model to a 3D DCE dataset.

Note: this is a large real-world dataset, and the computation may take several hours. In order to test this with a smaller dataset, you can use the DCE_small dataset.

Setup

import os
import time
import shutil
import numpy as np
import mdreg

# We use zarray data in this example
data = mdreg.fetch_zarr('DCE')

# Variables used in this script
tacq = data.attrs['time']
aif = data.attrs['aif']
spacing = data.attrs['spacing']
t0, t1 = 100, 150

# Path for output
results_path = os.path.join(os.getcwd(), 'tmp')

# Check the data
anim = mdreg.plot.animation(
    data[:, :, :, t0:t1],
    title='DCE Data',
    vmin=0,
    vmax=0.9*np.max(data[...,0]),
)

Once Loop Reflect

Perform motion correction

We fit the DCE data using a linearised 2-compartment model, using motion correction with default settings:

t = time.time()

coreg, fit, transfo, pars = mdreg.fit(
    data,
    fit_image={
        'func': mdreg.fit_2cm_lin,
        'time': tacq,                   # Acquisition times
        'aif': aif,                     # Signal-time curve in the aorta
        'baseline': 5,                  # Nr of precontrast samples
    },
    maxit=3,
    path=results_path,
    verbose=2,
)

print(f"Computation time: {round(time.time()-t)} seconds.")

Initializing..
Iteration 1: fitting signal model
Fitting deformation field (iteration 1)
Calculation time for iteration 1: 225.01013048887253 min
Iteration 2: fitting signal model
Fitting deformation field (iteration 2)
Calculation time for iteration 2: 227.00593881607057 min
Iteration 3: fitting signal model
Fitting deformation field (iteration 3)
Calculation time for iteration 3: 226.34060509999594 min
Total calculation time: 682.6465732375781 min
Computation time: 40966 seconds.

Check the result

anim = mdreg.plot.animation(
    coreg[:, :, :, t0:t1],
    title='DCE motion corrected',
    vmin=0,
    vmax=0.9*np.max(data[...,0]),
)

Once Loop Reflect

Cleanup disk

shutil.rmtree(results_path)