import os
import sys
import numpy
import nibabel
import nighresjava
from ..io import load_volume, save_volume
from ..utils import _output_dir_4saving, _fname_4saving, _check_available_memory
[docs]def laminar_regional_approximation(profile_surface_image, intensity_image, roi_image,
save_data=False, overwrite=False, output_dir=None,
file_name=None):
'''Sampling data on multiple intracortical layers
Parameters
-----------
profile_surface_image: niimg
4D image containing levelset representations of different intracortical
surfaces on which data should be sampled
intensity_image: niimg
Image from which data should be sampled
roi_image: niimg
Label image of the region to sample
save_data: bool
Save output data to file (default is False)
overwrite: bool
Overwrite existing results (default is False)
output_dir: str, optional
Path to desired output directory, will be created if it doesn't exist
file_name: str, optional
Desired base name for output files with file extension
(suffixes will be added)
Returns
-----------
dict
Dictionary collecting outputs under the following keys
(suffix of output files in brackets)
* weights (niimg): weight image, representing the weighting of profiles in the estimation (_lra-weight)
* degree (niimg): degree image, representing the degree of profiles in the estimation (_lra-deg)
* residuals (niimg): residuals image, representing the residual error in the estimation (_lra-res)
* median ([float]): the estimated median profile (_lra-med)
* perc25 ([float]): the estimated 25 percentile profile (_lra-p25)
* perc75 ([float]): the estimated 75 percentile profile (_lra-p75)
Notes
----------
Original Java module by Pierre-Louis Bazin
'''
print('\nProfile regional approximation')
# make sure that saving related parameters are correct
if save_data:
output_dir = _output_dir_4saving(output_dir, intensity_image)
weight_file = os.path.join(output_dir,
_fname_4saving(module=__name__,file_name=file_name,
rootfile=intensity_image,
suffix='lra-weight'))
deg_file = os.path.join(output_dir,
_fname_4saving(module=__name__,file_name=file_name,
rootfile=intensity_image,
suffix='lra-deg'))
res_file = os.path.join(output_dir,
_fname_4saving(module=__name__,file_name=file_name,
rootfile=intensity_image,
suffix='lra-res'))
perc25_file = os.path.join(output_dir,
_fname_4saving(module=__name__,file_name=file_name,
rootfile=intensity_image,
suffix='lra-p25',ext='txt'))
median_file = os.path.join(output_dir,
_fname_4saving(module=__name__,file_name=file_name,
rootfile=intensity_image,
suffix='lpa-med',ext='txt'))
perc75_file = os.path.join(output_dir,
_fname_4saving(module=__name__,file_name=file_name,
rootfile=intensity_image,
suffix='lpa-p75',ext='txt'))
if overwrite is False \
and os.path.isfile(weight_file) and os.path.isfile(sample_file) \
and os.path.isfile(median_file) and os.path.isfile(iqr_file) :
print("skip computation (use existing results)")
output = {'weights': weight_file,
'best': numpy.loadtxt(sample_file),
'median': numpy.loadtxt(median_file),
'iqr': numpy.loadtxt(iqr_file)}
return output
# start VM if not already running
try:
mem = _check_available_memory()
nighresjava.initVM(initialheap=mem['init'], maxheap=mem['max'])
except ValueError:
pass
# initate class
sampler = nighresjava.LaminarProfileAveraging()
# load the data
surface_img = load_volume(profile_surface_image)
surface_data = surface_img.get_data()
hdr = surface_img.header
aff = surface_img.affine
resolution = [x.item() for x in hdr.get_zooms()]
dimensions = surface_data.shape
intensity_data = load_volume(intensity_image).get_data()
roi_data = load_volume(roi_image).get_data()
# pass inputs
sampler.setIntensityImage(nighresjava.JArray('float')(
(intensity_data.flatten('F')).astype(float)))
sampler.setProfileSurfaceImage(nighresjava.JArray('float')(
(surface_data.flatten('F')).astype(float)))
sampler.setRoiMask(nighresjava.JArray('int')(
(roi_data.flatten('F')).astype(int).tolist()))
sampler.setResolutions(resolution[0], resolution[1], resolution[2])
sampler.setDimensions(dimensions[0], dimensions[1],
dimensions[2], dimensions[3])
# execute class
try:
sampler.execute()
except:
# if the Java module fails, reraise the error it throws
print("\n The underlying Java code did not execute cleanly: ")
print(sys.exc_info()[0])
raise
return
# collecting outputs
weight_data = numpy.reshape(numpy.array(
sampler.getProfileWeights(),
dtype=numpy.float32), (dimensions[0],dimensions[1],dimensions[2]), 'F')
sample = numpy.array(sampler.getSampleProfile(), dtype=numpy.float32)
median = numpy.array(sampler.getMedianProfile(), dtype=numpy.float32)
iqr = numpy.array(sampler.getIqrProfile(), dtype=numpy.float32)
hdr['cal_max'] = numpy.nanmax(weight_data)
weights = nibabel.Nifti1Image(weight_data, aff, hdr)
if save_data:
save_volume(weight_file, weights)
numpy.savetxt(sample_file, sample)
numpy.savetxt(median_file, median)
numpy.savetxt(iqr_file, iqr)
return {'weights': weight_file}
else:
return {'weights': weights}