Display 3D trajectories, with a grid interface #193
Conversation
This add "get_interpolators_from_fieldmap", supporting both real (B0 map only) and complex (Zmap = R2* + 1j * B0map) fields in arbitrary dimension (2D and 3D). The routine supports both numpy / cupy arrays and torch tensors on CPU and GPU, the latter requiring cupy due to limitations in torch.histogram / torch.histogramdd (pytorch/pytorch#69519). Calculation uses time segmentation with uniform time samples and LS coefficients (using histogram). Based on MIRT (https://github.com/JeffFessler/mirt/blob/main/mri/mri_exp_approx.m) and its Python porting from MIRTORCH (https://github.com/guanhuaw/MIRTorch/blob/master/mirtorch/linear/mri.py) and SigPy (https://github.com/mikgroup/sigpy/blob/main/sigpy/mri/util.py).
In addition, add off-resonance example.
Avoid torch cuda test case if cupy is not available.
remove deprecated get_grad Co-authored-by: Chaithya G R <[email protected]>
|
Please base yourself from master :) |
|
Naa, ill wait for #180 to go in :P |
|
I will get a CLI also :P |
|
I seemed to have forgotten to add the example. Added it now. @Daval-G you are good to go for review whenever you get time. |
|
@chaithyagr I will have a look over the week-end |
| @@ -0,0 +1,99 @@ | |||
| """ | |||
| ====================== | |||
| 3D Trajectory Display | |||
There was a problem hiding this comment.
| 3D Trajectory Display | |
| 3D Trajectory display |
There was a problem hiding this comment.
Also why is it a GPU example ?
There was a problem hiding this comment.
@chaithyagr Sorry I misplaced this comment, but could you answer it ?
There was a problem hiding this comment.
I think the reason why it's a GPU example is cause we use spread and interpolate kernels which currently is only supported on GPU backends. I'll add support for it in finufft soon
| def create_grid(grid_type, title="", **kwargs): | ||
| fig, axs = plt.subplots(3, 3, figsize=(10, 10)) | ||
| for i, (name, traj) in enumerate(trajectories.items()): | ||
| grid = get_gridded_trajectory( | ||
| traj, (64, 64, 64), grid_type=grid_type, traj_params=traj_params, **kwargs | ||
| ) | ||
| plot_slices(axs[:, i], grid, title=name) | ||
| plt.tight_layout() | ||
| plt.suptitle(title) | ||
| plt.show() |
There was a problem hiding this comment.
Why not provide trajectories and traj_params as arguments but keep grid_type and title ? It makes it a bit of a pain to read here and below
There was a problem hiding this comment.
I moved title below. But I dont know if I understand you?
| data = grid_op.raw_op.adj_op( | ||
| np.tile(np.linspace(1, 10, shots.shape[1]), (shots.shape[0],)), | ||
| None, | ||
| True, | ||
| ) |
There was a problem hiding this comment.
Don't we want to provide actual time values ? It should be more explicit in the documentation
There was a problem hiding this comment.
Also the output shape doesn't correspond to shape as expected when reading the documentation. I don't understand why this is part of that function since it doesn't relate to a grid
There was a problem hiding this comment.
It is just relative to get an understanding. Given that adjoint is not inverse, even if we give exact values, we cant expect it to be reflected the same way on image domain.
The shape would still be same as expected as we use grid_op which is set based on shape, right? Am I missing something?
|
|
||
| Parameters | ||
| ---------- | ||
| shots : ndarray |
There was a problem hiding this comment.
The shots name isn't conventional in that part of the package
There was a problem hiding this comment.
What would you recommend? traj ?
| def get_gridded_trajectory( | ||
| shots: np.ndarray, | ||
| shape: Tuple, | ||
| grid_type: str = "density", | ||
| osf: int = 1, | ||
| backend: str = "gpunufft", | ||
| traj_params: dict = None, | ||
| turbo_factor: int = 176, | ||
| elliptical_samp: bool = True, | ||
| threshold: float = 1e-3, | ||
| ): |
There was a problem hiding this comment.
Overall, why have all of those features gathered into only one function ? Some of them are not related to grids.
There was a problem hiding this comment.
Also please add more comments to your code
There was a problem hiding this comment.
I think, all the features are related to gridding and how you want to "color" the trajectories on the grid.
| data = grid_op.raw_op.adj_op( | ||
| np.repeat( | ||
| np.linspace(1, 10, turbo_factor), samples.shape[0] // turbo_factor + 1 | ||
| )[: samples.shape[0]], | ||
| None, | ||
| True, | ||
| ) |
There was a problem hiding this comment.
Same problem here, the output shape doesn't seem to match the documented output
There was a problem hiding this comment.
Same as above, isnt data.shape same as shape we give as input?
| if elliptical_samp: | ||
| data[ | ||
| np.linalg.norm( | ||
| np.meshgrid( | ||
| *[np.linspace(-1, 1, sh) for sh in shape], indexing="ij" | ||
| ), | ||
| axis=0, | ||
| ) | ||
| > 1 | ||
| ] = 0 |
| @@ -0,0 +1,99 @@ | |||
| """ | |||
| ====================== | |||
| 3D Trajectory Display | |||
There was a problem hiding this comment.
Also why is it a GPU example ?
Co-authored-by: Guillaume Daval-Frérot <[email protected]>
|
@Daval-G is there something else remaining or am I missing something here? |
No description provided.