Ctrl+K

Setting up the Calgary-Campinas Dataset

Setting up the Calgary-Campinas Dataset#

Imports#

from functools import partial

import numpy as np
import matplotlib.pyplot as plt


from direct.data.datasets import CalgaryCampinasDataset
from direct.data.mri_transforms import build_mri_transforms
from direct.data.transforms import fft2, ifft2, modulus, root_sum_of_squares
from direct.common.subsample import CalgaryCampinasMaskFunc

Define forward and backward transforms#

forward_operator = partial(fft2, dim=(1,2), centered=False)
backward_operator = partial(ifft2, dim=(1,2), centered=False)

Define sub-sampling function#

mask_func = CalgaryCampinasMaskFunc(accelerations=[5, 10])

Define MRI transforms#

transforms = build_mri_transforms(
    forward_operator=forward_operator,
    backward_operator=backward_operator,
    mask_func=mask_func,
    scaling_key="masked_kspace",
)

Create the Dataset#

# root pointing to the Calgary-Campinas data directory
root = "<...>"
dataset = CalgaryCampinasDataset(
    root = root,
    transform=transforms,
    crop_outer_slices=True,
)

Investigate a sample#

sample = dataset[20]

print("Sample keys: ", sample.keys())

Visualizations#

plt.imshow(sample["target"], cmap='gray')
plt.axis("off")
plt.title("Target")
plt.show()

plt.imshow(sample["sampling_mask"].squeeze(), cmap='gray')
plt.axis("off")
plt.title("Sub-Sampling Mask")
plt.show()

masked_kspace = sample["masked_kspace"]
print("K-space shape: (Coils, Height, Width, Complex) = ", tuple(masked_kspace.shape))

backprojected_kspace = backward_operator(masked_kspace)

fig, axs = plt.subplots(1, masked_kspace.shape[0], figsize=(20,50))

for i in range(masked_kspace.shape[0]):
    axs[i].imshow(modulus(backprojected_kspace[i]))
    axs[i].axis("off")
    axs[i].set_title(f"Coil {i+1}")
plt.show()
Contents