Module for computing digitally reconstructed radiographs



 DRR (volume:np.ndarray, spacing:np.ndarray, height:int, delx:float,
      width:int|None=None, dely:float|None=None,
      p_subsample:float|None=None, reshape:bool=True,

Torch module that computes differentiable digitally reconstructed radiographs.

Type Default Details
volume np.ndarray CT volume
spacing np.ndarray Dimensions of voxels in the CT volume
height int Height of the rendered DRR
delx float X-axis pixel size
width int | None None Width of the rendered DRR (if not provided, set to height)
dely float | None None Y-axis pixel size (if not provided, set to delx)
p_subsample float | None None Proportion of pixels to randomly subsample
reshape bool True Return DRR with shape (b, h, w)
params torch.Tensor | None None The parameters of the camera, including SDR, rotations, and translations.

X-ray pose parameters

The viewing angle for the DRR (known generally in computer graphics as pose parameters) is parameterized by the following:

  • sdr : Source-to-Detector radius (half of the source-to-detector distance)
  • theta : Azimuthal angle
  • phi : Polar angle
  • gamma : Plane rotation angle
  • bx : X-dir translation
  • by : Y-dir translation
  • bz : Z-dir translation

Translational (bx, by, bz) and rotational (theta, phi, gamma) parameters are grouped. The rotational pose parameters are detailed in Spherical Coordiantes Tutorial.



 DRR.forward ()

Forward call if DRR has been initialized with params.

The forward pass of the DRR module can be used if the module has viewing angle stored as nn.Parameters. Either pass parameters to the constructor or use _update_params to set them.

If params are passed to the DRR constructor, they are instantiated as nn.Parameters of the module. This makes them able to be optimized with any PyTorch optimizer.



 DRR.project (sdr:float, theta:float, phi:float, gamma:float, bx:float,
              by:float, bz:float)

Instead of initializing the DRR module with parameters, they can be passed to project to compute a DRR. This acts as a differentiable renderer (i.e., reconstruction), but is not differentiable with respect to the pose parameters (i.e., registration).

Note: .project doesn’t support batched DRR synthesis.