# encoding: utf-8 # module dipy.segment.mrf # from C:\Users\xukai\Downloads\发票2\venv\Lib\site-packages\dipy\segment\mrf.cp311-win_amd64.pyd # by generator 1.147 # no doc # imports import builtins as __builtins__ # import numpy as np # C:\Users\xukai\Downloads\发票2\venv\Lib\site-packages\numpy\__init__.py # functions def __pyx_unpickle_Enum(*args, **kwargs): # real signature unknown pass # classes class ConstantObservationModel(object): """ Observation model assuming that the intensity of each class is constant. The model parameters are the means $\mu_{k}$ and variances $\sigma_{k}$ associated with each tissue class. According to this model, the observed intensity at voxel $x$ is given by $I(x) = \mu_{k} + \eta_{k}$ where $k$ is the tissue class of voxel $x$, and $\eta_{k}$ is a Gaussian random variable with zero mean and variance $\sigma_{k}^{2}$. The observation model is responsible for computing the negative log-likelihood of observing any given intensity $z$ at each voxel $x$ assuming the voxel belongs to each class $k$. It also provides a default parameter initialization. """ def initialize_param_uniform(self, *args, **kwargs): # real signature unknown """ Initializes the means and variances uniformly The means are initialized uniformly along the dynamic range of `image`. The variances are set to 1 for all classes Parameters ---------- image : array 3D structural image nclasses : int number of desired classes Returns ------- mu : array 1 x nclasses, mean for each class sigma : array 1 x nclasses, standard deviation for each class. Set up to 1.0 for all classes. """ pass def negloglikelihood(self, *args, **kwargs): # real signature unknown """ Computes the gaussian negative log-likelihood of each class at each voxel of `image` assuming a gaussian distribution with means and variances given by `mu` and `sigmasq`, respectively (constant models along the full volume). The negative log-likelihood will be written in `nloglike`. Parameters ---------- image : ndarray 3D gray scale structural image mu : ndarray mean of each class sigmasq : ndarray variance of each class nclasses : int number of classes Returns ------- nloglike : ndarray 4D negloglikelihood for each class in each volume """ pass def prob_image(self, *args, **kwargs): # real signature unknown """ Conditional probability of the label given the image Parameters ---------- img : ndarray 3D structural gray-scale image nclasses : int number of tissue classes mu : ndarray 1 x nclasses, current estimate of the mean of each tissue class sigmasq : ndarray 1 x nclasses, current estimate of the variance of each tissue class P_L_N : ndarray 4D probability map of the label given the neighborhood. Previously computed by function prob_neighborhood Returns ------- P_L_Y : ndarray 4D probability of the label given the input image """ pass def seg_stats(self, *args, **kwargs): # real signature unknown """ Mean and standard variation for N desired tissue classes Parameters ---------- input_image : ndarray 3D structural image seg_image : ndarray 3D segmented image nclass : int number of classes (3 in most cases) Returns ------- mu, var : ndarrays 1 x nclasses dimension Mean and variance for each class """ pass def update_param(self, *args, **kwargs): # real signature unknown """ Updates the means and the variances in each iteration for all the labels. This is for equations 25 and 26 of Zhang et. al., IEEE Trans. Med. Imag, Vol. 20, No. 1, Jan 2001. Parameters ---------- image : ndarray 3D structural gray-scale image P_L_Y : ndarray 4D probability map of the label given the input image computed by the expectation maximization (EM) algorithm mu : ndarray 1 x nclasses, current estimate of the mean of each tissue class. nclasses : int number of tissue classes Returns ------- mu_upd : ndarray 1 x nclasses, updated mean of each tissue class var_upd : ndarray 1 x nclasses, updated variance of each tissue class """ pass def __init__(self, *args, **kwargs): # real signature unknown """ Initializes an instance of the ConstantObservationModel class """ pass __weakref__ = property(lambda self: object(), lambda self, v: None, lambda self: None) # default """list of weak references to the object (if defined)""" __dict__ = None # (!) real value is "mappingproxy({'__module__': 'dipy.segment.mrf', '__doc__': '\\n Observation model assuming that the intensity of each class is constant.\\n The model parameters are the means $\\\\mu_{k}$ and variances $\\\\sigma_{k}$\\n associated with each tissue class. According to this model, the observed\\n intensity at voxel $x$ is given by $I(x) = \\\\mu_{k} + \\\\eta_{k}$ where $k$\\n is the tissue class of voxel $x$, and $\\\\eta_{k}$ is a Gaussian random\\n variable with zero mean and variance $\\\\sigma_{k}^{2}$. The observation\\n model is responsible for computing the negative log-likelihood of\\n observing any given intensity $z$ at each voxel $x$ assuming the voxel\\n belongs to each class $k$. It also provides a default parameter\\n initialization.\\n ', '__init__': , 'initialize_param_uniform': , 'seg_stats': , 'negloglikelihood': , 'prob_image': , 'update_param': , '__dict__': , '__weakref__': })" class IteratedConditionalModes(object): # no doc def icm_ising(self, *args, **kwargs): # real signature unknown """ Executes one iteration of the ICM algorithm for MRF MAP estimation. The prior distribution of the MRF is a Gibbs distribution with the Potts/Ising model with parameter `beta`: https://en.wikipedia.org/wiki/Potts_model Parameters ---------- nloglike : ndarray 4D shape, nloglike[x, y, z, k] is the negative log likelihood of class k at voxel (x, y, z) beta : float positive scalar, it is the parameter of the Potts/Ising model. Determines the smoothness of the output segmentation. seg : ndarray 3D initial segmentation. This segmentation will change by one iteration of the ICM algorithm Returns ------- new_seg : ndarray 3D final segmentation energy : ndarray 3D final energy """ pass def initialize_maximum_likelihood(self, *args, **kwargs): # real signature unknown """ Initializes the segmentation of an image with given neg-loglikelihood Initializes the segmentation of an image with neglog-likelihood field given by `nloglike`. The class of each voxel is selected as the one with the minimum neglog-likelihood (i.e. maximum-likelihood segmentation). Parameters ---------- nloglike : ndarray 4D shape, nloglike[x, y, z, k] is the likelihhood of class k for voxel (x, y, z) Returns ------- seg : ndarray 3D initial segmentation """ pass def prob_neighborhood(self, *args, **kwargs): # real signature unknown """ Conditional probability of the label given the neighborhood Equation 2.18 of the Stan Z. Li book (Stan Z. Li, Markov Random Field Modeling in Image Analysis, 3rd ed., Advances in Pattern Recognition Series, Springer Verlag 2009.) Parameters ---------- seg : ndarray 3D tissue segmentation derived from the ICM model beta : float scalar that determines the importance of the neighborhood and the spatial smoothness of the segmentation. Usually between 0 to 0.5 nclasses : int number of tissue classes Returns ------- PLN : ndarray 4D probability map of the label given the neighborhood of the voxel. """ pass def __init__(self, *args, **kwargs): # real signature unknown pass __weakref__ = property(lambda self: object(), lambda self, v: None, lambda self: None) # default """list of weak references to the object (if defined)""" __dict__ = None # (!) real value is "mappingproxy({'__module__': 'dipy.segment.mrf', '__init__': , 'initialize_maximum_likelihood': , 'icm_ising': , 'prob_neighborhood': , '__dict__': , '__weakref__': , '__doc__': None})" # variables with complex values __loader__ = None # (!) real value is '<_frozen_importlib_external.ExtensionFileLoader object at 0x0000022D82CBF610>' __spec__ = None # (!) real value is "ModuleSpec(name='dipy.segment.mrf', loader=<_frozen_importlib_external.ExtensionFileLoader object at 0x0000022D82CBF610>, origin='C:\\\\Users\\\\xukai\\\\Downloads\\\\??2\\\\venv\\\\Lib\\\\site-packages\\\\dipy\\\\segment\\\\mrf.cp311-win_amd64.pyd')" __test__ = {}