Source code for dipy.sims.voxel

import numpy as np
from numpy import dot
from scipy.special import jn

from dipy.core.geometry import sphere2cart, vec2vec_rotmat
from dipy.reconst.utils import dki_design_matrix
from dipy.testing.decorators import warning_for_keywords

# Diffusion coefficients for white matter tracts, in mm^2/s
#
# Based roughly on values from:
#
#   Pierpaoli, Basser, "Towards a Quantitative Assessment of Diffusion
#   Anisotropy", Magnetic Resonance in Medicine, 1996; 36(6):893-906.
#
diffusion_evals = np.array([1500e-6, 400e-6, 400e-6])


def _check_directions(angles):
    """
    Helper function to check if direction ground truth have the right format
    and are in cartesian coordinates

    Parameters
    ----------
    angles : array (K,2) or (K, 3)
        List of K polar angles (in degrees) for the sticks or array of K
        sticks as unit vectors.

    Returns
    -------
    sticks : (K,3)
        Sticks in cartesian coordinates.
    """
    angles = np.array(angles)
    if angles.shape[-1] == 3:
        sticks = angles
    else:
        sticks = [
            sphere2cart(1, np.deg2rad(pair[0]), np.deg2rad(pair[1])) for pair in angles
        ]
        sticks = np.array(sticks)

    return sticks


def _add_gaussian(sig, noise1, noise2):
    """
    Helper function to add_noise

    This one simply adds one of the Gaussians to the sig and ignores the other
    one.
    """
    return sig + noise1


def _add_rician(sig, noise1, noise2):
    """
    Helper function to add_noise.

    This does the same as abs(sig + complex(noise1, noise2))

    """
    return np.sqrt((sig + noise1) ** 2 + noise2**2)


def _add_rayleigh(sig, noise1, noise2):
    r"""Helper function to add_noise.

    The Rayleigh distribution is $\sqrt\{Gauss_1^2 + Gauss_2^2}$.

    """
    return sig + np.sqrt(noise1**2 + noise2**2)


[docs] @warning_for_keywords() def add_noise(signal, snr, S0, *, noise_type="rician", rng=None): r"""Add noise of specified distribution to the signal from a single voxel. Parameters ---------- signal : 1-d ndarray The signal in the voxel. snr : float The desired signal-to-noise ratio. (See notes below.) If `snr` is None, return the signal as-is. S0 : float Reference signal for specifying `snr`. noise_type : string, optional The distribution of noise added. Can be either 'gaussian' for Gaussian distributed noise, 'rician' for Rice-distributed noise (default) or 'rayleigh' for a Rayleigh distribution. rng : numpy.random.Generator, optional Random number generator for the noise. If ``None``, uses NumPy's default random generator. Returns ------- signal : array, same shape as the input Signal with added noise. Notes ----- SNR is defined here, following :footcite:p:`Descoteaux2007`, as ``S0 / sigma``, where ``sigma`` is the standard deviation of the two Gaussian distributions forming the real and imaginary components of the Rician noise distribution (see :footcite:p:`Gudbjartsson1995`). References ---------- .. footbibliography:: Examples -------- >>> signal = np.arange(800).reshape(2, 2, 2, 100) >>> signal_w_noise = add_noise(signal, 10., 100., noise_type='rician') """ if snr is None: return signal if rng is None: rng = np.random.default_rng() sigma = S0 / snr noise_adder = { "gaussian": _add_gaussian, "rician": _add_rician, "rayleigh": _add_rayleigh, } noise1 = rng.normal(0, sigma, size=signal.shape) if noise_type == "gaussian": noise2 = None else: noise2 = rng.normal(0, sigma, size=signal.shape) return noise_adder[noise_type](signal, noise1, noise2)
[docs] @warning_for_keywords() def sticks_and_ball( gtab, *, d=0.0015, S0=1.0, angles=((0, 0), (90, 0)), fractions=(35, 35), snr=20 ): """Simulate the signal for a Sticks & Ball model. See :footcite:p:`Behrens2007` for a definition of the Sticks & Ball model. Parameters ---------- gtab : GradientTable Signal measurement directions. d : float, optional Diffusivity value. S0 : float, optional Unweighted signal value. angles : array (K, 2) or (K, 3), optional List of K polar angles (in degrees) for the sticks or array of K sticks as unit vectors. fractions : array-like, optional Percentage of each stick. Remainder to 100 specifies isotropic component. snr : float, optional Signal to noise ratio, assuming Rician noise. If set to None, no noise is added. Returns ------- S : (N,) ndarray Simulated signal. sticks : (M,3) Sticks in cartesian coordinates. References ---------- .. footbibliography:: """ fractions = [f / 100.0 for f in fractions] f0 = 1 - np.sum(fractions) S = np.zeros(len(gtab.bvals)) sticks = _check_directions(angles) for i, g in enumerate(gtab.bvecs[1:]): S[i + 1] = f0 * np.exp(-gtab.bvals[i + 1] * d) + np.sum( [ fractions[j] * np.exp(-gtab.bvals[i + 1] * d * np.dot(s, g) ** 2) for (j, s) in enumerate(sticks) ] ) S[i + 1] = S0 * S[i + 1] S[gtab.b0s_mask] = S0 S = add_noise(S, snr, S0) return S, sticks
[docs] def callaghan_perpendicular(q, radius): """Calculates the perpendicular diffusion signal E(q) in a cylinder of radius R using the Soderman model. Assumes that the pulse length is infinitely short and the diffusion time is infinitely long. See :footcite:p:`Soderman1995` for details about the Soderman model. Parameters ---------- q : array, shape (N,) q-space value in 1/mm radius : float cylinder radius in mm Returns ------- E : array, shape (N,) signal attenuation References ---------- .. footbibliography:: """ # Eq. [6] in the paper numerator = (2 * jn(1, 2 * np.pi * q * radius)) ** 2 denom = (2 * np.pi * q * radius) ** 2 E = np.divide(numerator, denom, out=np.zeros_like(q), where=denom != 0) return E
[docs] @warning_for_keywords() def gaussian_parallel(q, tau, *, D=0.7e-3): r"""Calculates the parallel Gaussian diffusion signal. Parameters ---------- q : array, shape (N,) q-space value in 1/mm tau : float diffusion time in s D : float, optional diffusion constant Returns ------- E : array, shape (N,) signal attenuation """ return np.exp(-((2 * np.pi * q) ** 2) * tau * D)
[docs] @warning_for_keywords() def cylinders_and_ball_soderman( gtab, tau, *, radii=(5e-3, 5e-3), D=0.7e-3, S0=1.0, angles=((0, 0), (90, 0)), fractions=(35, 35), snr=20, ): """Calculates the three-dimensional signal attenuation E(q) originating from within a cylinder of radius R using the Soderman approximation. The diffusion signal is assumed to be separable perpendicular and parallel to the cylinder axis :footcite:p:`Assaf2004`. This function is basically an extension of the ball and stick model. Setting the radius to zero makes them equivalent. See :footcite:p:`Soderman1995` for details about the Soderman model. Parameters ---------- gtab : GradientTable Signal measurement directions. tau : float diffusion time in s radii : array-like, optional cylinder radius in mm D : float, optional diffusion constant S0 : float, optional Unweighted signal value. angles : array (K, 2) or (K, 3), optional List of K polar angles (in degrees) for the sticks or array of K sticks as unit vectors. fractions : array-like, optional Percentage of each stick. Remainder to 100 specifies isotropic component. snr : float, optional Signal to noise ratio, assuming Rician noise. If set to None, no noise is added. Returns ------- E : array, shape (N,) signal attenuation References ---------- .. footbibliography:: """ qvals = np.sqrt(gtab.bvals / tau) / (2 * np.pi) qvecs = qvals[:, None] * gtab.bvecs q_norm = np.sqrt(np.einsum("ij,ij->i", qvecs, qvecs)) fractions = [f / 100.0 for f in fractions] f0 = 1 - np.sum(fractions) S = np.zeros(len(gtab.bvals)) sticks = _check_directions(angles) for i, f in enumerate(fractions): q_par = abs(np.dot(qvecs, sticks[i])) q_perp = np.sqrt(q_norm**2 - q_par**2) S_cylinder = callaghan_perpendicular(q_perp, radii[i]) * gaussian_parallel( q_par, tau, D=D ) S += f * S_cylinder S += f0 * np.exp(-gtab.bvals * D) S *= S0 S[gtab.b0s_mask] = S0 S = add_noise(S, snr, S0) return S, sticks
[docs] @warning_for_keywords() def single_tensor(gtab, S0=1, *, evals=None, evecs=None, snr=None, rng=None): """Simulate diffusion-weighted signals with a single tensor. See :footcite:p:`Stejskal1965`, :footcite:p:`Descoteaux2008b`. Parameters ---------- gtab : GradientTable Table with information of b-values and gradient directions g. Note that if gtab has a btens attribute, simulations will be performed according to the given b-tensor B information. S0 : double, optional Strength of signal in the presence of no diffusion gradient (also called the ``b=0`` value). evals : (3,) ndarray, optional Eigenvalues of the diffusion tensor. By default, values typical for prolate white matter are used. evecs : (3, 3) ndarray, optional Eigenvectors of the tensor. You can also think of this as a rotation matrix that transforms the direction of the tensor. The eigenvectors need to be column wise. snr : float, optional Signal to noise ratio, assuming Rician noise. None implies no noise. rng : numpy.random.Generator, optional Random number generator for the noise. If ``None``, uses NumPy's default random generator. Returns ------- S : (N,) ndarray Simulated signal: ``S(b, g) = S_0 e^(-b g^T R D R.T g)``, if gtab.tens=None ``S(B) = S_0 e^(-B:D)``, if gtab.tens information is given References ---------- .. footbibliography:: """ if rng is None: rng = np.random.default_rng() if evals is None: evals = diffusion_evals if evecs is None: evecs = np.eye(3) out_shape = gtab.bvecs.shape[: gtab.bvecs.ndim - 1] gradients = gtab.bvecs.reshape(-1, 3) R = np.asarray(evecs) S = np.zeros(len(gradients)) D = dot(dot(R, np.diag(evals)), R.T) if gtab.btens is None: for i, g in enumerate(gradients): S[i] = S0 * np.exp(-gtab.bvals[i] * dot(dot(g.T, D), g)) else: for i, b in enumerate(gtab.btens): S[i] = S0 * np.exp(-np.sum(b * D)) S = add_noise(S, snr, S0, rng=rng) return S.reshape(out_shape)
[docs] @warning_for_keywords() def multi_tensor( gtab, mevals, *, S0=1.0, angles=((0, 0), (90, 0)), fractions=(50, 50), snr=20, rng=None, ): r"""Simulate a Multi-Tensor signal. Parameters ---------- gtab : GradientTable Table with information of b-values and gradient directions. Note that if gtab has a btens attribute, simulations will be performed according to the given b-tensor information. mevals : array (K, 3) each tensor's eigenvalues in each row S0 : float, optional Unweighted signal value (b0 signal). angles : array (K, 2) or (K, 3), optional List of K tensor directions in polar angles (in degrees) or unit vectors fractions : array-like, optional Percentage of the contribution of each tensor. The sum of fractions should be equal to 100%. snr : float, optional Signal to noise ratio, assuming Rician noise. If set to None, no noise is added. rng : numpy.random.Generator, optional Random number generator for the noise. If ``None``, uses NumPy's default random generator. Returns ------- S : (N,) ndarray Simulated signal. sticks : (M,3) Sticks in cartesian coordinates. Examples -------- >>> import numpy as np >>> from dipy.sims.voxel import multi_tensor >>> from dipy.data import get_fnames >>> from dipy.core.gradients import gradient_table >>> from dipy.io.gradients import read_bvals_bvecs >>> fimg, fbvals, fbvecs = get_fnames(name='small_101D') >>> bvals, bvecs = read_bvals_bvecs(fbvals, fbvecs) >>> gtab = gradient_table(bvals, bvecs=bvecs) >>> mevals=np.array(([0.0015, 0.0003, 0.0003],[0.0015, 0.0003, 0.0003])) >>> e0 = np.array([1, 0, 0.]) >>> e1 = np.array([0., 1, 0]) >>> S = multi_tensor(gtab, mevals) """ if rng is None: rng = np.random.default_rng() if np.round(np.sum(fractions), 2) != 100.0: raise ValueError("Fractions should sum to 100") fractions = [f / 100.0 for f in fractions] S = np.zeros(len(gtab.bvals)) sticks = _check_directions(angles) for i in range(len(fractions)): S = S + fractions[i] * single_tensor( gtab, S0=S0, evals=mevals[i], evecs=all_tensor_evecs(sticks[i]), snr=None ) return add_noise(S, snr, S0, rng=rng), sticks
[docs] @warning_for_keywords() def multi_tensor_dki( gtab, mevals, *, S0=1.0, angles=((90.0, 0.0), (90.0, 0.0)), fractions=(50, 50), snr=20, ): r"""Simulate the diffusion-weight signal, diffusion and kurtosis tensors based on the DKI model See :footcite:p:`NetoHenriques2015` for further details. Parameters ---------- gtab : GradientTable Gradient table. mevals : array (K, 3) eigenvalues of the diffusion tensor for each individual compartment S0 : float, optional Unweighted signal value (b0 signal). angles : array (K,2) or (K,3), optional List of K tensor directions of the diffusion tensor of each compartment in polar angles (in degrees) or unit vectors fractions : float (K,), optional Percentage of the contribution of each tensor. The sum of fractions should be equal to 100%. snr : float, optional Signal to noise ratio, assuming Rician noise. If set to None, no noise is added. Returns ------- S : (N,) ndarray Simulated signal based on the DKI model. dt : (6,) elements of the diffusion tensor. kt : (15,) elements of the kurtosis tensor. Notes ----- Simulations are based on multicompartmental models which assumes that tissue is well described by impermeable diffusion compartments characterized by their only diffusion tensor. Since simulations are based on the DKI model, coefficients larger than the fourth order of the signal's taylor expansion approximation are neglected. Examples -------- >>> import numpy as np >>> from dipy.sims.voxel import multi_tensor_dki >>> from dipy.data import get_fnames >>> from dipy.core.gradients import gradient_table >>> from dipy.io.gradients import read_bvals_bvecs >>> fimg, fbvals, fbvecs = get_fnames(name='small_64D') >>> bvals, bvecs = read_bvals_bvecs(fbvals, fbvecs) >>> bvals_2s = np.concatenate((bvals, bvals * 2), axis=0) >>> bvecs_2s = np.concatenate((bvecs, bvecs), axis=0) >>> gtab = gradient_table(bvals_2s, bvecs=bvecs_2s) >>> mevals = np.array([[0.00099, 0, 0],[0.00226, 0.00087, 0.00087]]) >>> S, dt, kt = multi_tensor_dki(gtab, mevals) References ---------- .. footbibliography:: """ if np.round(np.sum(fractions), 2) != 100.0: raise ValueError("Fractions should sum to 100") fractions = [f / 100.0 for f in fractions] # S = np.zeros(len(gtab.bvals)) sticks = _check_directions(angles) # computing a 3D matrix containing the individual DT components D_comps = np.zeros((len(fractions), 3, 3)) for i in range(len(fractions)): R = all_tensor_evecs(sticks[i]) D_comps[i] = dot(dot(R, np.diag(mevals[i])), R.T) # compute voxel's DT DT = np.zeros((3, 3)) for i in range(len(fractions)): DT = DT + fractions[i] * D_comps[i] dt = np.array([DT[0][0], DT[0][1], DT[1][1], DT[0][2], DT[1][2], DT[2][2]]) # compute voxel's MD MD = (DT[0][0] + DT[1][1] + DT[2][2]) / 3 # compute voxel's KT kt = np.zeros(15) kt[0] = kurtosis_element(D_comps, fractions, 0, 0, 0, 0, DT=DT, MD=MD) kt[1] = kurtosis_element(D_comps, fractions, 1, 1, 1, 1, DT=DT, MD=MD) kt[2] = kurtosis_element(D_comps, fractions, 2, 2, 2, 2, DT=DT, MD=MD) kt[3] = kurtosis_element(D_comps, fractions, 0, 0, 0, 1, DT=DT, MD=MD) kt[4] = kurtosis_element(D_comps, fractions, 0, 0, 0, 2, DT=DT, MD=MD) kt[5] = kurtosis_element(D_comps, fractions, 0, 1, 1, 1, DT=DT, MD=MD) kt[6] = kurtosis_element(D_comps, fractions, 1, 1, 1, 2, DT=DT, MD=MD) kt[7] = kurtosis_element(D_comps, fractions, 0, 2, 2, 2, DT=DT, MD=MD) kt[8] = kurtosis_element(D_comps, fractions, 1, 2, 2, 2, DT=DT, MD=MD) kt[9] = kurtosis_element(D_comps, fractions, 0, 0, 1, 1, DT=DT, MD=MD) kt[10] = kurtosis_element(D_comps, fractions, 0, 0, 2, 2, DT=DT, MD=MD) kt[11] = kurtosis_element(D_comps, fractions, 1, 1, 2, 2, DT=DT, MD=MD) kt[12] = kurtosis_element(D_comps, fractions, 0, 0, 1, 2, DT=DT, MD=MD) kt[13] = kurtosis_element(D_comps, fractions, 0, 1, 1, 2, DT=DT, MD=MD) kt[14] = kurtosis_element(D_comps, fractions, 0, 1, 2, 2, DT=DT, MD=MD) # compute S based on the DT and KT S = dki_signal(gtab, dt, kt, S0=S0, snr=snr) return S, dt, kt
[docs] @warning_for_keywords() def kurtosis_element(D_comps, frac, ind_i, ind_j, ind_k, ind_l, *, DT=None, MD=None): r"""Computes the diffusion kurtosis tensor element (with indexes i, j, k and l) based on the individual diffusion tensor components of a multicompartmental model. Parameters ---------- D_comps : (K,3,3) ndarray Diffusion tensors for all K individual compartment of the multicompartmental model. frac : [float] Percentage of the contribution of each tensor. The sum of fractions should be equal to 100%. ind_i : int Element's index i (0 for x, 1 for y, 2 for z) ind_j : int Element's index j (0 for x, 1 for y, 2 for z) ind_k : int Element's index k (0 for x, 1 for y, 2 for z) ind_l: int Elements index l (0 for x, 1 for y, 2 for z) DT : (3,3) ndarray, optional Voxel's global diffusion tensor. MD : float, optional Voxel's global mean diffusivity. Returns ------- wijkl : float kurtosis tensor element of index i, j, k, l Notes ----- wijkl is calculated using equation 8 given in :footcite:p:`NetoHenriques2015`. References ---------- .. footbibliography:: """ if DT is None: DT = np.zeros((3, 3)) for i in range(len(frac)): DT = DT + frac[i] * D_comps[i] if MD is None: MD = (DT[0][0] + DT[1][1] + DT[2][2]) / 3 wijkl = 0 for f in range(len(frac)): wijkl = wijkl + frac[f] * ( D_comps[f][ind_i][ind_j] * D_comps[f][ind_k][ind_l] + D_comps[f][ind_i][ind_k] * D_comps[f][ind_j][ind_l] + D_comps[f][ind_i][ind_l] * D_comps[f][ind_j][ind_k] ) wijkl = ( wijkl - DT[ind_i][ind_j] * DT[ind_k][ind_l] - DT[ind_i][ind_k] * DT[ind_j][ind_l] - DT[ind_i][ind_l] * DT[ind_j][ind_k] ) / (MD**2) return wijkl
[docs] @warning_for_keywords() def dki_signal(gtab, dt, kt, *, S0=150, snr=None): r"""Simulated signal based on the diffusion and diffusion kurtosis tensors of a single voxel. Simulations are performed assuming the DKI model. See :footcite:p:`NetoHenriques2015` for further details. Parameters ---------- gtab : GradientTable Measurement directions. dt : (6,) ndarray Elements of the diffusion tensor. kt : (15, ) ndarray Elements of the diffusion kurtosis tensor. S0 : float, optional Strength of signal in the presence of no diffusion gradient. snr : float, optional Signal to noise ratio, assuming Rician noise. None implies no noise. Returns ------- S : (N,) ndarray Simulated signal based on the DKI model: .. math:: S=S_{0}e^{-bD+\frac{1}{6}b^{2}D^{2}K} References ---------- .. footbibliography:: """ dt = np.array(dt) kt = np.array(kt) A = dki_design_matrix(gtab) # define vector of DKI parameters MD = (dt[0] + dt[2] + dt[5]) / 3 X = np.concatenate((dt, kt * MD * MD, np.array([-np.log(S0)])), axis=0) # Compute signals based on the DKI model S = np.exp(dot(A, X)) S = add_noise(S, snr, S0) return S
[docs] @warning_for_keywords() def single_tensor_odf(r, *, evals=None, evecs=None): """Simulated ODF with a single tensor. See :footcite:p:`Aganj2010` for further details. Parameters ---------- r : (N,3) or (M,N,3) ndarray Measurement positions in (x, y, z), either as a list or on a grid. evals : (3,) Eigenvalues of diffusion tensor. By default, use values typical for prolate white matter. evecs : (3, 3) ndarray Eigenvectors of the tensor, written column-wise. You can also think of these as the rotation matrix that determines the orientation of the diffusion tensor. Returns ------- ODF : (N,) ndarray The diffusion probability at ``r`` after time ``tau``. References ---------- .. footbibliography:: """ if evals is None: evals = diffusion_evals if evecs is None: evecs = np.eye(3) out_shape = r.shape[: r.ndim - 1] R = np.asarray(evecs) D = dot(dot(R, np.diag(evals)), R.T) Di = np.linalg.inv(D) r = r.reshape(-1, 3) P = np.zeros(len(r)) for i, u in enumerate(r): P[i] = (dot(dot(u.T, Di), u)) ** (3 / 2) return (1 / (4 * np.pi * np.prod(evals) ** (1 / 2) * P)).reshape(out_shape)
[docs] def all_tensor_evecs(e0): """Given the principle tensor axis, return the array of all eigenvectors column-wise (or, the rotation matrix that orientates the tensor). Parameters ---------- e0 : (3,) ndarray Principle tensor axis. Returns ------- evecs : (3,3) ndarray Tensor eigenvectors, arranged column-wise. """ axes = np.eye(3) mat = vec2vec_rotmat(axes[0], e0) e1 = np.dot(mat, axes[1]) e2 = np.dot(mat, axes[2]) # Return the eigenvectors column-wise: return np.array([e0, e1, e2]).T
[docs] def multi_tensor_odf(odf_verts, mevals, angles, fractions): """Simulate a Multi-Tensor ODF. Parameters ---------- odf_verts : (N,3) ndarray Vertices of the reconstruction sphere. mevals : sequence of 1D arrays, Eigen-values for each tensor. angles : sequence of 2d tuples, Sequence of principal directions for each tensor in polar angles or cartesian unit coordinates. fractions : sequence of floats, Percentages of the fractions for each tensor. Returns ------- ODF : (N,) ndarray Orientation distribution function. Examples -------- Simulate a MultiTensor ODF with two peaks and calculate its exact ODF. >>> import numpy as np >>> from dipy.sims.voxel import multi_tensor_odf, all_tensor_evecs >>> from dipy.data import default_sphere >>> vertices, faces = default_sphere.vertices, default_sphere.faces >>> mevals = np.array(([0.0015, 0.0003, 0.0003],[0.0015, 0.0003, 0.0003])) >>> angles = [(0, 0), (90, 0)] >>> odf = multi_tensor_odf(vertices, mevals, angles, [50, 50]) """ mf = [f / 100.0 for f in fractions] sticks = _check_directions(angles) odf = np.zeros(len(odf_verts)) mevecs = [] for s in sticks: mevecs += [all_tensor_evecs(s)] for j, f in enumerate(mf): odf += f * single_tensor_odf(odf_verts, evals=mevals[j], evecs=mevecs[j]) return odf
[docs] @warning_for_keywords() def single_tensor_rtop(*, evals=None, tau=1.0 / (4 * np.pi**2)): """Simulate a Single-Tensor rtop. See :footcite:p:`Cheng2012` for further details. Parameters ---------- evals : 1D arrays, optional Eigen-values for the tensor. By default, values typical for prolate white matter are used. tau : float, optional diffusion time. By default the value that makes q=sqrt(b). Returns ------- rtop : float, Return to origin probability. References ---------- .. footbibliography:: """ if evals is None: evals = diffusion_evals rtop = 1.0 / np.sqrt((4 * np.pi * tau) ** 3 * np.prod(evals)) return rtop
[docs] @warning_for_keywords() def multi_tensor_rtop(mf, *, mevals=None, tau=1 / (4 * np.pi**2)): """Simulate a Multi-Tensor rtop. See :footcite:p:`Cheng2012` for further details. Parameters ---------- mf : sequence of floats, bounded [0,1] Percentages of the fractions for each tensor. mevals : sequence of 1D arrays, optional Eigen-values for each tensor. By default, values typical for prolate white matter are used. tau : float, optional diffusion time. By default the value that makes q=sqrt(b). Returns ------- rtop : float, Return to origin probability. References ---------- .. footbibliography:: """ rtop = 0 if mevals is None: mevals = [ None, ] * len(mf) for j, f in enumerate(mf): rtop += f * single_tensor_rtop(evals=mevals[j], tau=tau) return rtop
[docs] @warning_for_keywords() def single_tensor_pdf(r, *, evals=None, evecs=None, tau=1 / (4 * np.pi**2)): """Simulated ODF with a single tensor. See :footcite:p:`Cheng2012` for further details. Parameters ---------- r : (N,3) or (M,N,3) ndarray Measurement positions in (x, y, z), either as a list or on a grid. evals : (3,), optional Eigenvalues of diffusion tensor. By default, use values typical for prolate white matter. evecs : (3, 3) ndarray, optional Eigenvectors of the tensor. You can also think of these as the rotation matrix that determines the orientation of the diffusion tensor. tau : float, optional diffusion time. By default the value that makes q=sqrt(b). Returns ------- pdf : (N,) ndarray The diffusion probability at ``r`` after time ``tau``. References ---------- .. footbibliography:: """ if evals is None: evals = diffusion_evals if evecs is None: evecs = np.eye(3) out_shape = r.shape[: r.ndim - 1] R = np.asarray(evecs) D = dot(dot(R, np.diag(evals)), R.T) Di = np.linalg.inv(D) r = r.reshape(-1, 3) P = np.zeros(len(r)) for i, u in enumerate(r): P[i] = (-dot(dot(u.T, Di), u)) / (4 * tau) pdf = (1 / np.sqrt((4 * np.pi * tau) ** 3 * np.prod(evals))) * np.exp(P) return pdf.reshape(out_shape)
[docs] @warning_for_keywords() def multi_tensor_pdf(pdf_points, mevals, angles, fractions, *, tau=1 / (4 * np.pi**2)): """Simulate a Multi-Tensor ODF. See :footcite:p:`Cheng2012` for further details. Parameters ---------- pdf_points : (N, 3) ndarray Points to evaluate the PDF. mevals : sequence of 1D arrays, Eigen-values for each tensor. By default, values typical for prolate white matter are used. angles : sequence, Sequence of principal directions for each tensor in polar angles or cartesian unit coordinates. fractions : sequence of floats, Percentages of the fractions for each tensor. tau : float, optional diffusion time. By default the value that makes q=sqrt(b). Returns ------- pdf : (N,) ndarray, Probability density function of the water displacement. References ---------- .. footbibliography:: """ mf = [f / 100.0 for f in fractions] sticks = _check_directions(angles) pdf = np.zeros(len(pdf_points)) mevecs = [] for s in sticks: mevecs += [all_tensor_evecs(s)] for j, f in enumerate(mf): pdf += f * single_tensor_pdf( pdf_points, evals=mevals[j], evecs=mevecs[j], tau=tau ) return pdf
[docs] @warning_for_keywords() def single_tensor_msd(*, evals=None, tau=1 / (4 * np.pi**2)): """Simulate a Multi-Tensor rtop. See :footcite:p:`Cheng2012` for further details. Parameters ---------- evals : 1D arrays, optional Eigen-values for the tensor. By default, values typical for prolate white matter are used. tau : float, optional diffusion time. By default the value that makes q=sqrt(b). Returns ------- msd : float, Mean square displacement. References ---------- .. footbibliography:: """ if evals is None: evals = diffusion_evals msd = 2 * tau * np.sum(evals) return msd
[docs] @warning_for_keywords() def multi_tensor_msd(mf, *, mevals=None, tau=1 / (4 * np.pi**2)): """Simulate a Multi-Tensor rtop. See :footcite:p:`Cheng2012` for further details. Parameters ---------- mf : sequence of floats, bounded [0,1] Percentages of the fractions for each tensor. mevals : sequence of 1D arrays, optional Eigen-values for each tensor. By default, values typical for prolate white matter are used. tau : float, optional diffusion time. By default the value that makes q=sqrt(b). Returns ------- msd : float, Mean square displacement. References ---------- .. footbibliography:: """ msd = 0 if mevals is None: mevals = [ None, ] * len(mf) for j, f in enumerate(mf): msd += f * single_tensor_msd(evals=mevals[j], tau=tau) return msd