Spine Registration Helper Functions

spine_reg.draw(I, xI=None, fig=None, function=numpy.sum, **kwargs)[source]

Generate a plot of ‘I’ along the 3 cardinal planes - Coronal, sagittal, and transverse.

Parameters:

Iarray

A 3D image volume

xIlist of array

A list of the coordinates along each dimension of I

figmatplotlib.figure

A figure on which new plots will be generated

functionPython function

Default - np.sum(), will produce a Maximum Intensity Projection; The function used to generate the 3 views of I.

Returns:

axsnp.array of matplotlib.axes

Each element of axs contains 1 of the 3 cardinal views of I

spine_reg.fromblocks(fphiIpp, Jdsize)[source]

Parameters:

fphiIpp

Jdsizetorch.Tensor

The shape of the downsampled target data

Returns:

output

spine_reg.getslice(I, ax)[source]

Return a 2D slice of the 3D image volume ‘I’

Parameters:

Iarray

A 3D image volume

axint

Options : -1, -2, -3; The axis of I from which a slice should be extracted

Returns:

A subset of I along the desired axis

spine_reg.interp(xI, I, Xs, **kwargs)[source]

Interpolate …

Parameters:

xIlist of array

A list of the coordinates along each dimension of I

Iarray

A 3D image volume

Xs

Returns:

outputtorch.Tensor

spine_reg.interp1d(xI, squish, Xs, dd, **kwargs)[source]

Hack for 1D interpolation

Parameters:

xIlist of array

A list of the coordinates along each dimension of I

squish

Xs

dddict

A Python dictionary with 2 keys ‘device’ and ‘dtype’, which specifies those 2 PyTorch paremeters for computation

Returns:

output

spine_reg.measure_matching_dot(qIU, wIU, phiiQJU, wphiiQJU, SigmaQIU)[source]

Parameters:

qIU

wIU

phiiQJU

wphiiQJU

SigmaQIU

Returns:

output

spine_reg.phii_from_v(xv, v)[source]

Parameters:

xvlist of torch.Tensor

vtorch.Tensor

Returns:

phiitorch.Tensor

spine_reg.toblocks(Jd, blocksize)[source]

Parameters:

Jdtorch.Tensor

A 3D image volume with 1 additional batch dimension

blocksizeint

Returns:

Jdpp