3 d@sdZddlZddlmZmZmZmZddlmZddl m Z Gdd d eZ Gd d d eZ Gd d d eZ Gddde ZGddde ZGddde ZGddde ZdS)z The stats module provides higher-level interfaces to some of the operations that can be performed with the niftyseg stats (seg_stats) command-line program. N) TraitedSpecFiletraitsCommandLineInputSpec)NiftySegCommand)get_custom_pathc@sFeZdZdZeddddddZeddddd Zd Zej d ded Z dS) StatsInputz$Input Spec for seg_stats interfaces.rz%sTzimage to operate on)positionargstrexists mandatorydescz-m %sz!statistics within the masked area)r r r rz8Only estimate statistics if voxel is larger than z-t %f)r r rN) __name__ __module__ __qualname____doc__rin_fileZ mask_filerrFloatZ larger_voxelrrB/tmp/pip-build-7vycvbft/nipype/nipype/interfaces/niftyseg/stats.pyr sr c@seZdZdZejddZdS) StatsOutputz%Output Spec for seg_stats interfaces.zOutput array from seg_stats)rN)rrrrrArrayoutputrrrrr!srcsDeZdZdZedddZeZeZ ddZ fddZ d d Z Z S) StatsCommanda Base Command Interface for seg_stats interfaces. The executable seg_stats enables the estimation of image statistics on continuous voxel intensities (average, standard deviation, min/max, robust range, percentiles, sum, probabilistic volume, entropy, etc) either over the full image or on a per slice basis (slice axis can be specified), statistics over voxel coordinates (location of max, min and centre of mass, bounding box, etc) and statistics over categorical images (e.g. per region volume, count, average, Dice scores, etc). These statistics are robust to the presence of NaNs, and can be constrained by a mask and/or thresholded at a certain level. Z seg_statsZ NIFTYSEGDIR)Zenv_dircCs\g}xH|jdD]:}|jdr qt|dkr.qdd|jD}|j|qWtj|jS)N #rcSsg|] }t|qSr)float).0srrr Asz.StatsCommand._parse_stdout..)split startswithlenappendnparrayZsqueeze)selfstdoutoutZ string_linelinerrr _parse_stdout:s  zStatsCommand._parse_stdoutcs"tt|j|}|j|j|_|S)N)superr_run_interfacer/r,r)r+ZruntimeZ new_runtime) __class__rrr1EszStatsCommand._run_interfacecCs|jj}|j|d<|S)Nr) output_specgetr)r+outputsrrr _list_outputsJs  zStatsCommand._list_outputs)rrrrr Z_cmdr input_specrr3r/r1r6 __classcell__rr)r2rr's   rc@sDeZdZdZejdddddddd d d d d ddddddddddZdS)UnaryStatsInputz*Input Spec for seg_stats unary operations.rRar#vZvlZvpnr)enexXcBZxvoxZxdimz-%sTamOperation to perform: * r - The range of all voxels. * R - The robust range (assuming 2% outliers on both sides) of all voxels * a - Average of all voxels * s - Standard deviation of all voxels * v - Volume of all voxels above 0 (<# voxels> * ) * vl - Volume of each integer label (<# voxels per label> x ) * vp - Volume of all probabilsitic voxels (sum() x ) * n - Count of all voxels above 0 (<# voxels>) * np - Sum of all fuzzy voxels (sum()) * e - Entropy of all voxels * ne - Normalized entropy of all voxels * x - Location (i j k x y z) of the smallest value in the image * X - Location (i j k x y z) of the largest value in the image * c - Location (i j k x y z) of the centre of mass of the object * B - Bounding box of all nonzero voxels [ xmin xsize ymin ysize zmin zsize ] * xvox - Output the number of voxels in the x direction. Replace x with y/z for other directions. * xdim - Output the voxel dimention in the x direction. Replace x with y/z for other directions. )r r rrN)rrrrrEnum operationrrrrr9Ps.r9c@seZdZdZeZdS) UnaryStatsaUnary statistical operations. See Also -------- `Source code `__ -- `Documentation `__ Examples -------- >>> import copy >>> from nipype.interfaces import niftyseg >>> unary = niftyseg.UnaryStats() >>> unary.inputs.in_file = 'im1.nii' >>> # Test v operation >>> unary_v = copy.deepcopy(unary) >>> unary_v.inputs.operation = 'v' >>> unary_v.cmdline 'seg_stats im1.nii -v' >>> unary_v.run() # doctest: +SKIP >>> # Test vl operation >>> unary_vl = copy.deepcopy(unary) >>> unary_vl.inputs.operation = 'vl' >>> unary_vl.cmdline 'seg_stats im1.nii -vl' >>> unary_vl.run() # doctest: +SKIP >>> # Test x operation >>> unary_x = copy.deepcopy(unary) >>> unary_x.inputs.operation = 'x' >>> unary_x.cmdline 'seg_stats im1.nii -x' >>> unary_x.run() # doctest: +SKIP N)rrrrr9r7rrrrrHs$rHc@sbeZdZdZejdddddddd d d d d dddZed dd ddgddZej dd ddgddZ dS)BinaryStatsInputz+Input Spec for seg_stats Binary operations.psassZsvpaldZnccZnmiZVlZNlTz-%srEaOperation to perform: * p - - The th percentile of all voxels intensity (float=[0,100]) * sa - - Average of all voxels * ss - - Standard deviation of all voxels * svp - - Volume of all probabilsitic voxels (sum() x ) * al - - Average value in for each label in * d - - Calculate the Dice score between all classes in and * ncc - - Normalized cross correlation between and * nmi - - Normalized Mutual Information between and * Vl - - Volume of each integer label . Save to file. * Nl - - Count of each label . Save to file. )rr r rz%s operand_valuez&second image to perform operation with)r r rr xorrz%.8f operand_filezvalue to perform operation with)r rr rQrN) rrrrrrFrGrrRrrPrrrrrIs:rIc@seZdZdZeZdS) BinaryStatsaBinary statistical operations. See Also -------- `Source code `__ -- `Documentation `__ Examples -------- >>> import copy >>> from nipype.interfaces import niftyseg >>> binary = niftyseg.BinaryStats() >>> binary.inputs.in_file = 'im1.nii' >>> # Test sa operation >>> binary_sa = copy.deepcopy(binary) >>> binary_sa.inputs.operation = 'sa' >>> binary_sa.inputs.operand_value = 2.0 >>> binary_sa.cmdline 'seg_stats im1.nii -sa 2.00000000' >>> binary_sa.run() # doctest: +SKIP >>> # Test ncc operation >>> binary_ncc = copy.deepcopy(binary) >>> binary_ncc.inputs.operation = 'ncc' >>> binary_ncc.inputs.operand_file = 'im2.nii' >>> binary_ncc.cmdline 'seg_stats im1.nii -ncc im2.nii' >>> binary_ncc.run() # doctest: +SKIP >>> # Test Nl operation >>> binary_nl = copy.deepcopy(binary) >>> binary_nl.inputs.operation = 'Nl' >>> binary_nl.inputs.operand_file = 'output.csv' >>> binary_nl.cmdline 'seg_stats im1.nii -Nl output.csv' >>> binary_nl.run() # doctest: +SKIP N)rrrrrIr7rrrrrSs$rS)rnumpyr)baserrrrrZ niftyreg.baser r rrr9rHrIrSrrrrs  )3)4