Is the gold standard always the ifft2 of the complete k-space data, without any downsample or subsampling?

For a brain data of size 14×20×768×368 can we say that we get one slice out of the 20 images and we get volume space with 14 slices?

Is the gold standard always the ifft2 of the complete k-space data, without any downsample or subsampling?

For a brain data of size 14×20×768×368 can we say that we get one slice out of the 20 images and we get volume space with 14 slices?

Hello @creator2020, that is close. You also need to reduce the coil dimension. The simplest approach is the square root sum of squares. In totality, with the `numpy`

API, it looks something like this.

```
image = np.sqrt(
np.sum(
np.abs(
np.fft.fftshift(
np.fft.ifftn(np.fft.ifftshift(kspace, axes=(2, 3)), axes=(2, 3)),
axes=(2, 3),
)
)
** 2,
axis=1,
)
)
```