omni.masks#

Functions for image mask creation.

Functions

`compute_weight_mask`(brain, brain_mask[, ...])	Compute a weight mask for a given brain mask.
`generate_noise_mask`(img, mask[, size, ...])	Generates a mask identifying noise and signal voxels.
`lda`(X, labels)	Linear Discriminant Analysis for time series data.
`make_regression_mask`(output_prefix, epi, ...)	Make regression mask.

omni.masks.compute_weight_mask(brain: Nifti1Image, brain_mask: Nifti1Image, eye_mask: Optional[Nifti1Image] = None, dilation_size: int = 15) → Nifti1Image[source]#

Compute a weight mask for a given brain mask.

This function computes a weight mask for a given brain mask. The weight mask is computed by dilating the brain mask by dilation_size and then finding the non-brain portions through the use of otsu’s method + thresholding + morphology

The weight mask is weighted by the number of voxels in brain vs. non-brain.

Parameters:

brain: nib.Nifti1Image: Brain image.
brain_mask: nib.Nifti1Image: Brain mask.
eye_mask: nib.Nifti1Image, optional: Eye mask.
dilation_size: int: Size to dilate brain mask by.

Returns:

nib.Nifti1Image: Weight mask.

omni.masks.generate_noise_mask(img: Nifti1Image, mask: Nifti1Image, size: int = 2, iterations: int = 20, sigma: float = 3) → Tuple[ndarray, ndarray][source]#

Generates a mask identifying noise and signal voxels.

Parameters:

img: nib.Nifti1Image: Image to construct noise mask on.
mask: nib.Nifti1Image: Mask outlining a prior guess between noise/signal voxels.
size: int: Size to dilate noise mask by.
iterations: int: Number of iterations to run LDA.
sigma: float: Size of smoothing kernel for weight mask.

Returns:

np.ndarray: Noise mask.
np.ndarray: Signal weight mask.

omni.masks.lda(X: ndarray, labels: ndarray) → ndarray[source]#

Linear Discriminant Analysis for time series data.

Parameters:

X: np.ndarray: The data in a numpy array. Where rows are samples, cols are features.
labels: np.ndarray: Labels for data. Corresponds to samples.

Returns:

np.ndarray: Projected data.

omni.masks.make_regression_mask(output_prefix: str, epi: str, anat_bet_mask: str, anat_weight_mask: str, affine: str, iaffine: str, warp: str, iwarp: str, noise_mask_dilation_size: int = 2, noise_mask_iterations: int = 20, noise_mask_sigma: float = 2)[source]#

Make regression mask.

Parameters:

output_prefix: str: Set prefix for output files.
epi: str: EPI file to apply LDA.
anat_bet_mask: str: Anatomical brain mask.
anat_weight_mask: str: Anatomical weight mask.
affine: str: Affine transform (anat to func) (afni).
iaffine: str: Inverse affine transform (func to anat) (afni).
warp: str: Forward warp (anat to func).
iwarp: str: Inverse warp (func to anat).
noise_mask_dilation_size: int: Size to dilate noise mask by.
noise_mask_iterations: int: Number of iterations to run LDA.
noise_mask_sigma: float: Size of smoothing kernel for weight mask.

Returns:

str: Regression mask.