import os
import sys
import numpy as np
import nibabel as nb
import nighresjava
from ..io import load_volume, save_volume, load_mesh_geometry, save_mesh_geometry
from ..utils import _output_dir_4saving, _fname_4saving,_check_available_memory
[docs]def mesh_to_levelset(surface_mesh, reference_image,
save_data=False, overwrite=False,
output_dir=None, file_name=None):
"""Mesh to levelset
Creates a signed distance function from a triangulated mesh using pseudonormals.
Parameters
----------
surface_mesh: mesh
Mesh model of the surface
reference_image: niimg
Image of the dimensions and resolutions corresponding to the mesh
save_data: bool, optional
Save output data to file (default is False)
overwrite: bool, optional
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)
* result (niimg): Levelset function representing the mesh (_m2l-lvl)
Notes
----------
Ported from original Java module by Christine Tardif and Pierre-Louis Bazin.
References
----------
"""
print("\nMesh to Levelset")
# make sure that saving related parameters are correct
if save_data:
output_dir = _output_dir_4saving(output_dir, surface_mesh)
lvl_file = os.path.join(output_dir,
_fname_4saving(module=__name__,file_name=file_name,
rootfile=surface_mesh,
suffix='m2l-lvl'))
if overwrite is False \
and os.path.isfile(lvl_file) :
print("skip computation (use existing results)")
output = {'result': lvl_file}
return output
# start virtual machine if not running
try:
mem = _check_available_memory()
nighresjava.initVM(initialheap=mem['init'], maxheap=mem['max'])
except ValueError:
pass
# initiate class
algorithm = nighresjava.SurfaceMeshToLevelsetPseudoNormals()
# load the data
mesh = load_mesh_geometry(surface_mesh)
algorithm.setSurfacePoints(nighresjava.JArray('float')(
(mesh['points'].flatten('C')).astype(float)))
algorithm.setSurfaceTriangles(nighresjava.JArray('int')(
(mesh['faces'].flatten('C')).astype(int).tolist()))
ref_img = load_volume(reference_image)
hdr = ref_img.header
aff = ref_img.affine
resolution = [x.item() for x in hdr.get_zooms()]
dimensions = hdr.get_data_shape()
algorithm.setResolutions(resolution[0], resolution[1], resolution[2])
algorithm.setDimensions(dimensions[0], dimensions[1], dimensions[2])
# execute class
try:
algorithm.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
# collect outputs
lvl_data = np.reshape(np.array(algorithm.getLevelsetImage(),
dtype=np.float32), dimensions, 'F')
# create the mesh dictionary
header['cal_min'] = np.nanmin(lvl_data)
header['cal_max'] = np.nanmax(lvl_data)
lvl = nb.Nifti1Image(lvl_data, affine, header)
if save_data:
save_volume(lvl_file, lvl)
return {'result': lvl_file}
else:
return {'result': lvl}