recon package¶
Module contents¶
This module includes many of the useful MRI reconstruction functions. At the moment it includes stuff like FFTs, sum of squares, registration, and interpolation.
Submodules¶
recon.bart module¶
-
recon.bart.bart(nargout, cmd, *args)¶ This is the python interface for the bart toolbox. If you want more information about individual bart functions, go for
bart [command] -h
- Parameters
nargout (int) – The number of outputs you expect
cmd (str) – The bart command to put in (exclude ‘bart’)
*args (variable) – The inputs into bart, and this can vary
- Returns
output – The output from bart
- Return type
array_like
recon.fftc module¶
Centered FFTs
-
recon.fftc.fft2c(data)¶ Centered FFT in two dimensions (dim0, dim1). All other dimensions are left alone.
- Parameters
data (array_like) – The input to the FFT, must be 2D
- Returns
dim1 – Output from the FFT
- Return type
array_like
-
recon.fftc.fftc(data, dim)¶ Centered FFT. User specifies the dimension of the transform.
- Parameters
data (array_like) – The input to the FFT
dim (int) – The dimension to take the FFT
- Returns
Result – The result of the FFT
- Return type
array_like
-
recon.fftc.ifft2c(data)¶ Centered iFFT in two dimensions (dim0, dim1). All other dimensions are left alone.
- Parameters
data (array_like) – The input to the iFFT, must be 2D
- Returns
dim1 – Output from the iFFT
- Return type
array_like
-
recon.fftc.ifftc(data, dim)¶ Centered iFFT. User specifies the dimension of the transform.
- Parameters
data (array_like) – The input to the iFFT
dim (int) – The dimension to take the iFFT
- Returns
Result – The result of the iFFT
- Return type
array_like
recon.imtools module¶
These are some basic tools for image processing. This includes basic sinc interpolation, image cropping, zeropadding, and registration.
-
recon.imtools.add_rician(v, s)¶ r = ricernd(v, s) returns random sample(s) from the Rice (aka Rician) distribution with parameters v and s. (either v or s may be arrays, if both are, they must match in size)
R ~ Rice(v, s) if R = sqrt(X^2 + Y^2), where X ~ N(v*cos(a), s^2) and Y ~ N(v*sin(a), s^2) are independent normal distributions (any real a). Note that v and s are not the mean and standard deviation of R!
The size of Y is the common size of the input arguments. A scalar input functions as a constant matrix of the same size as the other inputs.
- Parameters
v (array_like) – Array to add noise to
s (float) – The Rician noise parameter
- Returns
imOut – Magnitude image with Rician noise
- Return type
array_like
-
recon.imtools.crop(im, cropy, cropx)¶ Crops or pads image to the specified dimensions. This assums that the original is up to 4D (dim0, dim1, dim2, dim3) and crops in dim0 and dim1.
- Parameters
im (array_like) – Array to be cropped. Can be up to 4D
cropy (int) – Extent of cropping in dim0
cropx (int) – Extent of cropping in dim1
- Returns
im – Output image
- Return type
array_like
-
recon.imtools.interpolate(im, cropy, cropx)¶ This is sinc interpolation (in Fourier domain) along the first two dimensions of an array.
- Parameters
im (Input image) – Image to be interpolated. Can be complex or real and can have up to four dimensions
cropy (int) – The extent of interpolation in y (first dimension)
cropx (int) – The extent of interpolation in x (second dimension)
- Returns
im – Interpolated array
- Return type
array_like
-
recon.imtools.register_2D(data, regType='SyNRA', verbose=True)¶ This is a 2D registration for a stack of images using ANTs
- Parameters
data (array_like) – A stack of 2D images in a 3D array, with registration occurring on third dimension
regType (str) – Description of ANTs registration
verbose (bool) – Verbose output
- Returns
dataReg – Array of 2D images that have been registered
- Return type
array_like
-
recon.imtools.register_2d_model(data, numPrecontrastBlood=5, numPrecontrastTissue=7, numProtonDensity=10, sliceCheck=4, frameCheck=31)¶ This function is a wrapper for the matlab implentation of 2D registration for 4D perfusion datasets. This relies on the physiological model to handle the contrast changes from frame to frame. It uses a spline deformable model on the backend.
- Parameters
data (array_like) – Unregistered 4D data to be registered in the (y, x, t, z) format (while data is 4D, registration happens only in 2D)
numPrecontrastBlood (int) – Number of pre contrast blood images
numPrecontrastTissue (int) – Number of pre contrast tissue images
numProtonDensity (int) – Number of proton density images
- Returns
dataReg – Registered data array (4D)
- Return type
array_like
-
recon.imtools.register_3D(data, regType='SyNRA', verbose=True)¶ This is a 3D registration for a stack of images using ANTs
- Parameters
data (array_like) – A stack of 3D images in a 4D array, with registration occurring on third dimension
regType (str) – Description of ANTs registration
verbose (bool) – Verbose output
- Returns
dataReg – Array of 3D images that have been registered
- Return type
array_like
-
recon.imtools.register_3d_model(data, numPrecontrastBlood=5, numPrecontrastTissue=7, numProtonDensity=10, sliceCheck=4, frameCheck=31)¶ This function is a wrapper for the matlab implentation of 3D registration for 4D perfusion datasets. This relies on the physiological model to handle the contrast changes from frame to frame. It uses a spline deformable model on the backend.
- Parameters
data (array_like) – Unregistered 4D data to be registered in the (y, x, t, z) format
numPrecontrastBlood (int) – Number of pre contrast blood images
numPrecontrastTissue (int) – Number of pre contrast tissue images
numProtonDensity (int) – Number of proton density images
- Returns
dataReg – Registered data array (4D)
- Return type
array_like
recon.sos module¶
Sum of squares coil combination
-
recon.sos.sos(data, dim)¶ Sum-of-squares combination. User specificies the dimension of the combination.