class designer.plotting.snrplot.makesnr(dwilist, noisepath=None, maskpath=None)

Bases: object

Class object that computes and prints SNR plots for any number of input DWIs

Parameters

  • dwilist (list of str) – List of 4D DWI (nifti-format) paths to evaluate and plot

  • noisepath (str) – Path to noise map from “dwidenoise”

  • maskpath (str, optional) – Path to brain mask

__init__ : constructs makesnr class
getuniquebval : creates a list of unique B-values for the purpose of

SNR computation

computesnr : performs SNR computation
histcount : bins SNR values
makeplot : creates and saves SNR plot from bin counts
computesnr()

Computes SNR of all DWIs in class object

Returns

snr_dwi – Numpy array of SNR across all DWI.

Return type

ndarray

getuniquebval()

Creates a list of unique B-values for the purpose of SNR computation. In the calculation of SNR, B0 signal can be averaged becase they are not associated to any direction. This is not true for non-B0 values however, because every 3D volume represents a different direction. To compute SNR appropriately, differences in gradients have to be accounted. This function creates a list of B-values in the order they need to appear for the calculation of SNR.

Returns

b_list – Numpy vector containing list of B-values to be used in SNR calculation

Return type

ndarray

histcount(nbins=100)

Bins SNR into nbins and returns various counting properties

Parameters

  • nbins (int)

  • Number of bins to plot

Returns

  • count (ndarray) – Array of count of voxels in bins

  • binval (ndarray) – Array of bin values

  • unibvals (ndarray) – Array containing all unique B-values detected

makeplot(path, smooth=True, smoothfactor=5)

Creates and saves SNR plot to a path as SNR.png

Parameters

  • path (str) – Directory to save the plot in

  • smooth (bool, optional) – Specify whether to interpolate and smooth (Default: True)

  • smoothfactor (int, optional) – Smoothing factor to apply (Default: 5)

Returns

None

Return type

Writes out image into directory as SNR.png

designer.plotting.snrplot.vectorize(img, mask)

Returns vectorized image based on brain mask, requires no input parameters If the input is 1D or 2D, unpatch it to 3D or 4D using a mask If the input is 3D or 4D, vectorize it using a mask Classification: Function

Parameters

  • img (ndarray) – 1D, 2D, 3D or 4D image array to vectorize

  • mask (ndarray) – 3D image array for masking

Returns

  • vec (N X number_of_voxels vector or array, where N is the number) – of DWI volumes

  • Usage

  • —–

  • vec = vectorize(img) if there’s no mask

  • vec = vectorize(img, mask) if there’s a mask