API Reference

The following section outlines the API of shimming-toolbox.

unwrap

Wrapper to FSL Prelude (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FUGUE/Guide#PRELUDE_.28phase_unwrapping.29)

shimmingtoolbox.unwrap.prelude.prelude(wrapped_phase, affine, mag=None, mask=None, threshold=None, is_unwrapping_in_2d=False)

wrapper to FSL prelude

This function enables phase unwrapping by calling FSL prelude on the command line. A mask can be provided to mask the phase image provided. 2D unwrapping can be turned off. The output path can be specified. The temporary niis can optionally be saved.

Parameters
  • wrapped_phase (numpy.ndarray) – 2D or 3D radian numpy array to perform phase unwrapping. (2 pi interval)

  • affine (numpy.ndarray) – 2D array containing the transformation coefficients. Can be calculated by using: nii = nib.load(“nii_path”) affine = nii.affine

  • mag (numpy.ndarray) – 2D or 3D magnitude numpy array corresponding to the phase array

  • mask (numpy.ndarray, optional) – numpy array of booleans with shape of complex_array to mask during phase unwrapping

  • threshold – Threshold value for automatic mask generation (Use either mask or threshold, not both)

  • is_unwrapping_in_2d (bool, optional) – prelude parameter to unwrap slice by slice

Returns

3D array with the shape of complex_array of the unwrapped phase output from prelude

Return type

numpy.ndarray

masking

Image mask with shape API

shimmingtoolbox.masking.shapes.shape_cube(data, len_dim1, len_dim2, len_dim3, center_dim1=None, center_dim2=None, center_dim3=None)

Creates a cube mask. Returns mask with the same shape as data.

Parameters
  • data (numpy.ndarray) – Data to mask, must be 3 dimensional array.

  • len_dim1 (int) – Length of the side of the square along first dimension (in pixels).

  • len_dim2 (int) – Length of the side of the square along second dimension (in pixels).

  • len_dim3 (int) – Length of the side of the square along third dimension (in pixels).

  • center_dim1 (int) – Center of the square along first dimension (in pixels). If no center is provided, the middle is used.

  • center_dim2 (int) – Center of the square along second dimension (in pixels). If no center is provided, the middle is used.

  • center_dim3 (int) – Center of the square along third dimension (in pixels). If no center is provided, the middle is used.

Returns

Mask with booleans. True where the cube is located and False in the background.

Return type

numpy.ndarray

shimmingtoolbox.masking.shapes.shape_square(data, len_dim1, len_dim2, center_dim1=None, center_dim2=None)

Creates a square mask. Returns mask with the same shape as data.

Parameters
  • data (numpy.ndarray) – Data to mask, must be 2 dimensional array.

  • len_dim1 (int) – Length of the side of the square along first dimension (in pixels).

  • len_dim2 (int) – Length of the side of the square along second dimension (in pixels).

  • center_dim1 (int) – Center of the square along first dimension (in pixels). If no center is provided, the middle is used.

  • center_dim2 (int) – Center of the square along second dimension (in pixels). If no center is provided, the middle is used.

Returns

Mask with booleans. True where the square is located and False in the background.

Return type

numpy.ndarray

shimmingtoolbox.masking.shapes.shapes(data, shape, **kargs)

Wrapper to different shape masking functions.

Parameters
  • data (numpy.ndarray) – Data to mask.

  • shape (str) – Shape to mask, implemented shapes include: {‘square’, ‘cube’}.

  • **kargs – Refer to the specific function in this file for the specific arguments for each shape. See example section for more details.

Returns

Mask with booleans. True where the shape is located and False in the background.

Return type

numpy.ndarray

Examples

>>> dummy_data = np.ones([4,3])
>>> dummy_mask = shapes(dummy_data, 'square', center_dim1=1, center_dim2=1, len_dim1=1, len_dim2=3)

coils

siemens_basis

shimmingtoolbox.coils.siemens_basis.siemens_basis(x, y, z)

The function first wraps shimmingtoolbox.coils.spherical_harmonics to generate 1st and 2nd order spherical harmonic basis fields at the grid positions given by arrays X,Y,Z. Following Siemens convention, basis is then:

  • Reordered along the 4th dimension as X, Y, Z, Z2, ZX, ZY, X2-Y2, XY

  • Rescaled to Hz/unit-shim, where “unit-shim” refers to the measure displayed in the Adjustments card of the Syngo console UI, namely:

    • 1 micro-T/m for X,Y,Z gradients (= 0.042576 Hz/mm)

    • 1 micro-T/m^2 for 2nd order terms (= 0.000042576 Hz/mm^2)

The returned basis is thereby in the form of ideal “shim reference maps”, ready for optimization.

Parameters
  • orders (numpy.ndarray) – not yet implemented Degrees of the desired terms in the series expansion, specified as a vector of non-negative integers ([0:1:n] yields harmonics up to n-th order)

  • x (numpy.ndarray) – 3-D arrays of grid coordinates, “Right->Left” grid coordinates in the patient coordinate system (i.e. DICOM reference, units of mm)

  • y (numpy.ndarray) – 3-D arrays of grid coordinates (same shape as x). “Anterior->Posterior” grid coordinates in the patient coordinate system (i.e. DICOM reference, units of mm)

  • z (numpy.ndarray) – 3-D arrays of grid coordinates (same shape as x). “Inferior->Superior” grid coordinates in the patient coordinate system (i.e. DICOM reference, units of mm)

Returns

4-D array of spherical harmonic basis fields

Return type

numpy.ndarray

Notes

For now, orders is, in fact, ignored: fixed as [1:2]—which is suitable for the Prisma (presumably other Siemens systems as well) however, the 3rd-order shims of the Terra should ultimately be accommodated too. (Requires checking the Adjustments/Shim card to see what the corresponding terms and values actually are). So, for now, basis will always be returned with 8 terms along the 4th dim.

spherical_harmonics

shimmingtoolbox.coils.spherical_harmonics.spherical_harmonics(orders, x, y, z)

Returns an array of spherical harmonic basis fields with the order/degree index along the 4th dimension.

Parameters
  • orders (numpy.ndarray) – Degrees of the desired terms in the series expansion, specified as a vector of non-negative integers (np.array(range(0, 3)) yields harmonics up to (n-1)-th order). Must be non negative.

  • x (numpy.ndarray) – 3-D arrays of grid coordinates

  • y (numpy.ndarray) – 3-D arrays of grid coordinates (same shape as x)

  • z (numpy.ndarray) – 3-D arrays of grid coordinates (same shape as x)

Returns

4d basis set of spherical harmonics with order/degree ordered along 4th dimension

Return type

numpy.ndarray

Examples

Initialize grid positions

>>> [x, y, z] = np.meshgrid(np.array(range(-10, 11)), np.array(range(-10, 11)), np.array(range(-10, 11)), indexing='ij')

0th-to-2nd order terms inclusive

>>> orders = np.array(range(0, 3))
>>> basis = spherical_harmonics(orders, x, y, z)

Notes

  • basis[:, :, :,0] corresponds to the 0th-order constant term (globally=unity)
    • 0: c

  • basis[:, :, :, 1:3] to 1st-order linear terms
    • 1: y

    • 2: z

    • 3: x

  • basis[:, :, :, 4:8] to 2nd-order terms
    • 4: xy

    • 5: zy

    • 6: z2

    • 7: zx

    • 8: x2y2

Based on

numerical_model

Create numerical model data for multi-echo B0 field mapping data

This module is for numerically simulating multi-echo B0 field mapping data. It considers features like: background B0 field, flip angle, echo time, and noise.

Typical usage example:

from shimmingtoolbox.simulate import *

b0_sim = NumericalModel(model="shepp-logan")

# Generate a background B0
b0_field = 13 # (Hz)
b0_sim.generate_deltaB0("linear", [0.0, b0_field])

# Simulate the signal data
FA = 15 # (degrees)
TE = [0.003, 0.015] # (seconds)
SNR = 50
b0_sim.simulate_measurement(FA, TE, SNR)

# Save simulation as NIfTI file (JSON sidecar also exported with parameters)
b0_sim.save('Phase', 'b0_mapping_data.nii', format='nifti')
class shimmingtoolbox.simulate.numerical_model.NumericalModel(model=None, num_vox=128)

Multi-echo B0 field mapping data numerical simulator.

Simulate multi-echo B0 field mapping data in the presence of a B0 field. Can simulate data under ideal conditions or with noise. Export simulations in a NIfTI or .mat file formats.

gamma

Gyromagnetic ratio in rad * Hz / Tesla.

Type

float

field_strength

Static field strength in Tesla.

Type

float

handedness

Orientation of the cross-product for the Larmor equation. The value of this attribute is MRI vendor-dependent.

measurement

Simulated measurement data array.

proton_density

Default assumed brain proton density in %.

T2_star

Default assumed brain T2* values in seconds at 3T.

generate_deltaB0(field_type, params)

Generates a background B0 field.

Defines the starting volume. Sets the background B0 field to zeros.

Parameters
  • field_type – Type of field to be generated. Available implementations are: 'linear'.

  • params – List of parameters defining the field for the selected field type. If field_type = 'linear', then params are [m b] where m (Hz/pixel) is the slope and b is the floor field (Hz).

save(data_type, file_name, format=None)

Exports simulated data to a file with a JSON sidecar.

Resets the measurement class attribute to zero before simulating. Simulates the signal for each echo-time provided. If defined, adds noise to the complex simulated signal measurements using an SNR value.

Parameters
  • data_type – Export data type. “Magnitude”, “Phase”, “Real”, or “Imaginary”.

  • file_name – Filename of exported file, with or without file extension.

  • format – File format for exported data. If no value given, will attempt to extract format from filename file extension, otherwise default to NIfTI.

simulate_measurement(FA, TE, SNR=None)

Simulates a multi-echo measurement for field mapping

Resets the measurement class attribute to zero before simulating. Simulates the signal for each echo-time provided. If defined, adds noise to the complex simulated signal measurements using an SNR value.

Parameters
  • FA – Flip angle in degrees.

  • TE – Echo-times in seconds. Can be either a single value, list, or array.

  • SNR – Signal-to-noise ratio used to define noise. If not set, no noise is added to the measurements.

misc

shimmingtoolbox.download.download_data(urls)

Download the binaries from a URL and return the destination filename Retry downloading if either server or connection errors occur on a SSL connection

Parameters

urls – list of several urls (mirror servers) or single url (string)

shimmingtoolbox.download.install_data(url, dest_folder, keep=False)

Download a data bundle from a URL and install in the destination folder.

Parameters
  • url – URL or sequence thereof (if mirrors).

  • dest_folder – destination directory for the data (to be created).

  • keep – whether to keep existing data in the destination folder.

Returns

NoneType

Note

The function tries to be smart about the data contents.

Examples:

If the archive only contains a README.md, and the destination folder is ${dst}, ${dst}/README.md will be created. Note: an archive not containing a single folder is commonly known as a “tarbomb” because it puts files anywhere in the current working directory.

If the archive contains a ${dir}/README.md, and the destination folder is ${dst}, ${dst}/README.md will be created. Note: typically the package will be called ${basename}-${revision}.zip and contain a root folder named ${basename}-${revision}/ under which all the other files will be located. The right thing to do in this case is to take the files from there and install them in ${dst}.

  • Uses download_data() to retrieve the data.

  • Uses unzip() to extract the bundle.

shimmingtoolbox.download.unzip(compressed, dest_folder)

Extract compressed file to the dest_folder. Can handle .zip, .tar.gz. If none of this extension is found, simply copy the file in dest_folder.

Parameters
  • compressed – the compressed .zip or .tar.gz file

  • dest_folder – the destination dir that expanded files are written to

shimmingtoolbox.load_nifti.get_acquisition_times(nii_data, json_data)

Return the acquisition timestamps from a json sidecar. This assumes BIDS convention.

Parameters
  • nii_data (nibabel.Nifti1Image) – Nibabel object containing the image timeseries.

  • json_data (dict) – Json dict corresponding to a nifti sidecar.

Returns

Acquisition timestamps in ms.

Return type

numpy.ndarray

shimmingtoolbox.load_nifti.load_nifti(path_data, modality='phase')

Load data from a directory containing NIFTI type file with nibabel. :param path_data: Path to the directory containing the file(s) to load :type path_data: str :param modality: Modality to read nifti (can be phase or magnitude) :type modality: str

Returns

List containing headers for every Nifti file dict: List containing all information in JSON format from every Nifti image numpy.ndarray: 5D array of all acquisition in time (x, y, z, echo, volume)

Return type

nibabel.Nifti1Image.Header

Note

If ‘path’ is a folder containing niftis, directly output niftis. It ‘path’ is a folder containing acquisitions, ask the user for which acquisition to use.

shimmingtoolbox.load_nifti.read_nii(nii_path, auto_scale=True)

Reads a nifti file and returns the corresponding image and info. Also returns the associated json data. :param nii_path: direct path to the .nii or .nii.gz file that is going to be read :type nii_path: str :param auto_scale: Tells if scaling is done before return :type auto_scale: bool, optional

Returns

Objet containing various data about the nifti file (returned by nibabel.load) json_data (dict): Contains the different fields present in the json file corresponding to the nifti file image (ndarray): Image contained in the read nifti file. Siemens phase images are rescaled between 0 and 2pi.

Return type

info (Nifti1Image)

shimmingtoolbox.utils.add_suffix(fname, suffix)

Add suffix between end of file name and extension.

Parameters
  • fname – absolute or relative file name. Example: t2.nii

  • suffix – suffix. Example: _mean

Return

file name string with suffix. Example: t2_mean.nii

Examples:

  • add_suffix(t2.nii, _mean) -> t2_mean.nii

  • add_suffix(t2.nii.gz, a) -> t2a.nii.gz

shimmingtoolbox.utils.iso_times_to_ms(iso_times)

Convert dicom acquisition times to ms

Parameters

iso_times (numpy.ndarray) – 1D array of time strings from dicoms. Suported formats: “HHMMSS.mmmmmm” or “HH:MM:SS.mmmmmm”

Returns

1D array of times in milliseconds

Return type

numpy.ndarray

shimmingtoolbox.utils.run_subprocess(cmd)

Wrapper for subprocess.run() that enables to input cmd as a full string (easier for debugging).

Parameters

cmd (string) – full command to be run on the command line

shimmingtoolbox.utils.st_progress_bar(*args, **kwargs)

Thin wrapper around tqdm.tqdm which checks SCT_PROGRESS_BAR muffling the progress bar if the user sets it to no, off, or false (case insensitive).