def mitkglobaltractography(i: str,
out_folder: str,
parameters: str,
mask: str = None):
"""
Perform MITK global tractography. Save a paramter file for usage using the MITK Diffusion GUI application.
http://mitk.org/wiki/MitkDiffusion
i: input image (tensor
out_folder: output folder
parameters: parameter file (.gtp)
mask: binary mask image (optional)
"""
os.makedirs(out_folder, exist_ok=True)
tracts = out_folder + os.path.basename(i).split('.')[0] + '_Global_mitk.trk'
runner = CmdInterface('MitkGlobalTractography')
runner.add_arg(key='-i', arg=i, check_input=True)
runner.add_arg(key='-o', arg=tracts, check_output=True)
runner.add_arg(key='--parameters', arg=parameters)
if mask is not None:
runner.add_arg(key='--mask', arg=mask, check_input=True)
runner.run()
return tracts
def recon_qball(input_dwi, out_folder, sh_order: int, do_flip_bvecs: bool):
""" Perform analytical q-ball reconstruction with solid angle consideration and output the resulting
spherical harmonics ODFs.
"""
os.makedirs(out_folder, exist_ok=True)
if do_flip_bvecs:
flipper = CmdInterface(flip_bvecs)
flipper.add_arg('input_dwi', input_dwi, check_input=True)
input_dwi = input_dwi.replace(os.path.dirname(input_dwi), out_folder)
input_dwi = input_dwi.replace('.nii.gz', '_flipped.nii.gz')
flipper.add_arg('output_dwi', input_dwi, check_output=True)
flipper.run()
qball_recon = CmdInterface('MitkQballReconstruction')
qball_recon.add_arg(key='-i', arg=input_dwi, check_input=True)
qball_recon.add_arg(key='-o', arg=out_folder + 'odf_qball_mitk.nii.gz', check_output=True)
qball_recon.add_arg(key='--sh_order', arg=sh_order)
qball_recon.run()
return out_folder + 'odf_qball_mitk.nii.gz'
def recon_tensor(input_dwi: str, out_folder: str, do_flip_bvecs: bool = False):
""" Perform diffusion tesnor modelling of the signal.
"""
os.makedirs(out_folder, exist_ok=True)
if do_flip_bvecs:
flipper = CmdInterface(flip_bvecs)
flipper.add_arg('input_dwi', input_dwi, check_input=True)
input_dwi = input_dwi.replace(os.path.dirname(input_dwi), out_folder)
input_dwi = input_dwi.replace('.nii.gz', '_flipped.nii.gz')
flipper.add_arg('output_dwi', input_dwi, check_output=True)
flipper.run()
tensor_recon = CmdInterface('MitkTensorReconstruction')
tensor_recon.add_arg(key='-i', arg=input_dwi, check_input=True)
tensor_recon.add_arg(key='-o', arg=out_folder + 'tensors_mitk.dti', check_output=True)
tensor_recon.run()
return out_folder + 'tensors_mitk.dti'
from cmdint import CmdInterface
'''This example shows how to setup the telegram logging service. You will receive a telegram message each time a
command execution with CmdInterface started or finished, including the logfile after the command has finished.
'''
# How to get a telegram bot token and chat_id: https://github.com/python-telegram-bot/python-telegram-bot/wiki/Introduction-to-the-API
# Steps:
# 1. Search for botfather in the telegram search an start a conversation
# 2. Type /newbot and follow the instructions of BotFather
# 3. Save your token
# 4. Search for your bot in the teegram search and start a conversion. SImply send "hello" or something.
# 5. Get the chat id by browsing the adress
# https://api.telegram.org/bot[YOUR-TOKEN-HERE-WITHOUT-BRACKETS]/getUpdates
# The number shown in the field "id" is your chat ID
CmdInterface.set_telegram_logger(token='my-telegram-bot-token',
chat_id='my-chat-with-bot',
caption='TEST')
# create instance of CmdInterface with the name of the command to be called (here "ls")
test = CmdInterface('ls')
# add keyword based argument
test.add_arg(key='-l', arg='/')
# run command
test.run()
def my_python_function(my_arg: str):
for i in range(10):
print(my_arg + ' ' + str(i))
# another dummy with progress bar
def my_python_function_with_progress():
bar = ProgressBar(10)
for i in range(10):
time.sleep(1)
bar.next()
# set output logfile and tell CmdInterface to delete a potentially present old logfile
CmdInterface.set_static_logfile('python_function_logging.json',
delete_existing=True)
# create instance of CmdInterface that calls the previously defined python function "my_python_function"
test_my_python_function = CmdInterface(my_python_function)
test_my_python_function.add_arg('my_arg', 'BLABLA')
test_my_python_function.run()
# another call with our progress bar function
# the progress bar in the output logfile is updated continuously
test_my_python_function_with_progress = CmdInterface(
my_python_function_with_progress)
test_my_python_function_with_progress.run()
# anonymize logfile
CmdInterface.anonymize_log(out_log_name='python_function_logging.json')
# set output logfile and tell CmdInterface to delete a potentially present old logfile
CmdInterface.set_static_logfile('io_file_logging.json', delete_existing=True)
CmdInterface.set_throw_on_error(do_throw=False)
# we want to log if a run is not necessary. by default this is disabled since we don't want to swamp our
# logfile if we call the script multiple times
CmdInterface.set_immediate_return_on_run_not_necessary(do_return=False)
# for this example we dont't want to exit python if a run fails. default is True.
CmdInterface.set_exit_on_error(do_exit=False)
####################################################################################################
# create a new text file using "echo"
step1 = CmdInterface('echo')
step1.add_arg(arg='unus ignis quis vir multum ab audere')
# by setting "check_output" to True, the argument is regarded as an expected output file
# (logged under ['command']['output']['expected']).
# If the file is found BEFORE execution, the execution is skipped, since a run is regarded as unnecessary when all
# outputf iles are already present.
# The run fails if this file is not found AFTER execution
# (logged under: ['command']['output']['missing']).
# If it is found, the run is successful and the file hash is logged to enable unique identification
# (logged under ['command']['output']['found']).
step1.add_arg(key='>', arg='io_file_logging.txt', check_output=True)
step1.run()
####################################################################################################
# copy the previously created file using "cp"
step2 = CmdInterface('cp')
def train_rf(i: str,
t: str,
out_folder: str,
masks: str = None,
wm_masks: str = None,
volume_modification_images: str = None,
additional_feature_images: str = None,
num_trees: int = 30,
max_tree_depth: int = 25,
sample_fraction: float = 0.7,
use_sh_features: bool = 0,
sampling_distance: float = None,
max_wm_samples: int = None,
num_gm_samples: int = None):
"""
Train a random forest for machine learning based tractography.
i: input diffusion-weighted images
t: input training tractograms
o: output random forest (HDF5)
masks: restrict training using a binary mask image (optional)
wm_masks: if no binary white matter mask is specified
volume_modification_images: specify a list of float images that modify the fiber density (optional)
additional_feature_images: specify a list of float images that hold additional features (float) (optional)
num_trees: number of trees (optional)
max_tree_depth: maximum tree depth (optional)
sample_fraction: fraction of samples used per tree (optional)
use_sh_features: use SH features (optional)
sampling_distance: resampling parameter for the input tractogram in mm (determines number of white-matter samples) (optional)
max_wm_samples: upper limit for the number of WM samples (optional)
num_gm_samples: Number of gray matter samples per voxel (optional)
"""
runner = CmdInterface('MitkRfTraining')
runner.add_arg(key='-i', arg=i, check_input=True)
runner.add_arg(key='-t', arg=t, check_input=True)
runner.add_arg(key='-o', arg=out_folder + 'forest_mitk.rf', check_output=True)
if masks is not None:
runner.add_arg(key='--masks', arg=masks)
if wm_masks is not None:
runner.add_arg(key='--wm_masks', arg=wm_masks)
if volume_modification_images is not None:
runner.add_arg(key='--volume_modification_images', arg=volume_modification_images)
if additional_feature_images is not None:
runner.add_arg(key='--additional_feature_images', arg=additional_feature_images)
if num_trees is not None:
runner.add_arg(key='--num_trees', arg=num_trees)
if max_tree_depth is not None:
runner.add_arg(key='--max_tree_depth', arg=max_tree_depth)
if sample_fraction is not None:
runner.add_arg(key='--sample_fraction', arg=sample_fraction)
if use_sh_features is not None:
runner.add_arg(key='--use_sh_features', arg=use_sh_features)
if sampling_distance is not None:
runner.add_arg(key='--sampling_distance', arg=sampling_distance)
if max_wm_samples is not None:
runner.add_arg(key='--max_wm_samples', arg=max_wm_samples)
if num_gm_samples is not None:
runner.add_arg(key='--num_gm_samples', arg=num_gm_samples)
runner.run()
return out_folder + 'forest_mitk.rf'
def track_streamline(input_image: str,
out_folder: str,
algo: str,
num_streamlines: int,
cutoff: float = 0.1,
minlength: int = 30,
maxlength: int = 200,
step: float = None,
angle: float = None):
""" Perform MRtrix streamline tractography.
"""
os.makedirs(out_folder, exist_ok=True)
tracts = out_folder + os.path.basename(input_image).split('.')[0] + '_' + algo + '_mrtrix'
tckgen = CmdInterface('tckgen')
tckgen.add_arg(arg=input_image, check_input=True)
if algo == 'Tensor_Det' or algo == 'Tensor_Prob':
print(algo + ' NOT IMPLEMENTED')
exit()
tckgen.add_arg(key='-fslgrad',
arg=[input_image.replace('.nii.gz', '.bvecs'), input_image.replace('.nii.gz', '.bvals')])
tckgen.add_arg(arg=tracts + '.tck', check_output=True)
tckgen.add_arg('-algorithm', algo)
tckgen.add_arg('-seed_dynamic', input_image)
tckgen.add_arg('-nthreads', 12)
tckgen.add_arg('-select', num_streamlines)
tckgen.add_arg('-minlength', minlength)
tckgen.add_arg('-maxlength', maxlength)
tckgen.add_arg('-cutoff', cutoff)
if step is not None:
tckgen.add_arg('-step', step)
if angle is not None:
tckgen.add_arg('-angle', angle)
tckgen.add_arg('-force')
tckgen.run(version_arg='--version')
postproc = CmdInterface('MitkFiberProcessing')
postproc.add_arg('-i', tracts + '.tck', check_input=True)
postproc.add_arg('--compress', 0.1)
postproc.add_arg('-o', tracts + '.trk', check_output=True)
postproc.run()
return tracts + '.trk'
def recon_csd(input_dwi: str, do_flip_bvecs: bool, out_folder: str, algo: str = 'tournier'):
""" Perform constrained spherical deconvolution modelling and output the resulting spherical harmonics fODFs.
"""
os.makedirs(out_folder, exist_ok=True)
if do_flip_bvecs:
flipper = CmdInterface(flip_bvecs)
flipper.add_arg('input_dwi', input_dwi, check_input=True)
input_dwi = input_dwi.replace(os.path.dirname(input_dwi), out_folder)
input_dwi = input_dwi.replace('.nii.gz', '_flipped.nii.gz')
flipper.add_arg('output_dwi', input_dwi, check_output=True)
flipper.run()
csd_algo = 'csd'
num_responses = 1
if algo != 'tournier':
num_responses = 3
csd_algo = 'msmt_csd'
dwi2response = CmdInterface('dwi2response')
dwi2response.add_arg(arg=algo)
dwi2response.add_arg(arg=input_dwi, check_input=True)
for i in range(num_responses):
dwi2response.add_arg(arg=out_folder + 'response_' + algo + '_' + str(i) + '_mrtrix.txt', check_output=True)
dwi2response.add_arg(key='-force')
dwi2response.add_arg('-nthreads', 12)
dwi2response.add_arg(key='-fslgrad',
arg=[input_dwi.replace('.nii.gz', '.bvecs'), input_dwi.replace('.nii.gz', '.bvals')],
check_input=True)
dwi2response.run()
dwi2fod = CmdInterface('dwi2fod')
dwi2fod.add_arg(arg=csd_algo)
dwi2fod.add_arg(arg=input_dwi, check_input=True)
for i in range(num_responses):
dwi2fod.add_arg(arg=out_folder + 'response_' + algo + '_' + str(i) + '_mrtrix.txt', check_input=True)
dwi2fod.add_arg(arg=out_folder + 'odf_' + csd_algo + '_' + str(i) + '_mrtrix.nii.gz', check_output=True)
dwi2fod.add_arg(key='-force')
dwi2fod.add_arg('-nthreads', 12)
dwi2fod.add_arg(key='-fslgrad',
arg=[input_dwi.replace('.nii.gz', '.bvecs'), input_dwi.replace('.nii.gz', '.bvals')],
check_input=True)
dwi2fod.run(version_arg='--version')
return out_folder + 'odf_' + csd_algo + '_' + str(0) + '_mrtrix.nii.gz'
sh2peaks = CmdInterface('sh2peaks')
sh2peaks.add_arg(arg=out_folder + 'odf_' + csd_algo + '_0_mrtrix.nii.gz', check_input=True)
sh2peaks.add_arg(arg=out_folder + 'peaks_' + csd_algo + '_0_mrtrix.nii.gz', check_output=True)
sh2peaks.add_arg('-threshold', 0.1)
sh2peaks.run(version_arg='--version')
flipper = CmdInterface('MitkFlipPeaks')
flipper.add_arg('-i', out_folder + 'peaks_' + csd_algo + '_0_mrtrix.nii.gz', check_input=True)
flipper.add_arg('-o', out_folder + 'peaks_' + csd_algo + '_0_flipped_mrtrix.nii.gz', check_output=True)
flipper.add_arg('-z')
flipper.run()
def track_streamline(i: str,
out_folder: str,
algorithm: str,
flip_x: bool = False,
flip_y: bool = False,
flip_z: bool = False,
no_data_interpolation: bool = False,
no_mask_interpolation: bool = False,
compress: float = None,
seeds: int = 1,
seed_image: str = None,
trials_per_seed: int = 10,
max_tracts: int = -1,
tracking_mask: str = None,
stop_image: str = None,
exclusion_image: str = None,
ep_constraint: str = None,
target_image: str = None,
sharpen_odfs: bool = False,
cutoff: float = 0.1,
odf_cutoff: float = 0,
step_size: float = 0.5,
min_tract_length: float = 20,
angular_threshold: float = None,
loop_check: float = None,
prior_image: str = None,
prior_weight: float = 0.5,
restrict_to_prior: bool = False,
new_directions_from_prior: bool = False,
num_samples: int = 0,
sampling_distance: float = 0.25,
use_stop_votes: bool = False,
use_only_forward_samples: bool = False,
tend_f: float = 1,
tend_g: float = 0,
forest: str = None,
use_sh_features: bool = False,
additional_images: str = None):
"""
Perform MITK streamline tractography.
i: input image (multiple possible for 'DetTensor' algorithm)
out_folder: output folder
algorithm: which algorithm to use (Peaks
flip_x: multiply x-coordinate of direction proposal by -1 (optional)
flip_y: multiply y-coordinate of direction proposal by -1 (optional)
flip_z: multiply z-coordinate of direction proposal by -1 (optional)
no_data_interpolation: don't interpolate input image values (optional)
no_mask_interpolation: don't interpolate mask image values (optional)
compress: compress output fibers using the given error threshold (in mm) (optional)
seeds: number of seed points per voxel (optional)
seed_image: mask image defining seed voxels (optional)
trials_per_seed: try each seed N times until a valid streamline is obtained (only for probabilistic tractography) (optional)
max_tracts: tractography is stopped if the reconstructed number of tracts is exceeded (optional)
tracking_mask: streamlines leaving the mask will stop immediately (optional)
stop_image: streamlines entering the mask will stop immediately (optional)
exclusion_image: streamlines entering the mask will be discarded (optional)
ep_constraint: determines which fibers are accepted based on their endpoint location - options are NONE
target_image: effact depends on the chosen endpoint constraint (option ep_constraint) (optional)
sharpen_odfs: if you are using dODF images as input
cutoff: set the FA
odf_cutoff: threshold on the ODF magnitude. this is useful in case of CSD fODF tractography. (optional)
step_size: step size (in voxels) (optional)
min_tract_length: minimum fiber length (in mm) (optional)
angular_threshold: angular threshold between two successive steps
loop_check: threshold on angular stdev over the last 4 voxel lengths (optional)
prior_image: tractography prior in thr for of a peak image (optional)
prior_weight: weighting factor between prior and data. (optional)
restrict_to_prior: restrict tractography to regions where the prior is valid. (optional)
new_directions_from_prior: the prior can create directions where there are none in the data. (optional)
num_samples: number of neighborhood samples that are use to determine the next progression direction (optional)
sampling_distance: distance of neighborhood sampling points (in voxels) (optional)
use_stop_votes: use stop votes (optional)
use_only_forward_samples: use only forward samples (optional)
tend_f: weighting factor between first eigenvector (f=1 equals FACT tracking) and input vector dependent direction (f=0). (optional)
tend_g: weighting factor between input vector (g=0) and tensor deflection (g=1 equals TEND tracking) (optional)
forest: input random forest (HDF5 file) (optional)
use_sh_features: use SH features (optional)
additional_images: specify a list of float images that hold additional information (FA
"""
os.makedirs(out_folder, exist_ok=True)
tracts = out_folder + os.path.basename(i).split('.')[0] + '_' + algorithm + '_mitk.trk'
runner = CmdInterface('MitkStreamlineTractography')
runner.add_arg(key='-i', arg=i, check_input=True)
runner.add_arg(key='-o', arg=tracts, check_output=True)
runner.add_arg(key='--algorithm', arg=algorithm)
if flip_x:
runner.add_arg(key='--flip_x')
if flip_y:
runner.add_arg(key='--flip_y')
if flip_z:
runner.add_arg(key='--flip_z')
if no_data_interpolation:
runner.add_arg(key='--no_data_interpolation')
if no_mask_interpolation:
runner.add_arg(key='--no_mask_interpolation')
if compress is not None:
runner.add_arg(key='--compress', arg=compress)
if seeds is not None:
runner.add_arg(key='--seeds', arg=seeds)
if seed_image is not None:
runner.add_arg(key='--seed_image', arg=seed_image)
if trials_per_seed is not None:
runner.add_arg(key='--trials_per_seed', arg=trials_per_seed)
if max_tracts is not None:
runner.add_arg(key='--max_tracts', arg=max_tracts)
if tracking_mask is not None:
runner.add_arg(key='--tracking_mask', arg=tracking_mask)
if stop_image is not None:
runner.add_arg(key='--stop_image', arg=stop_image)
if exclusion_image is not None:
runner.add_arg(key='--exclusion_image', arg=exclusion_image)
if ep_constraint is not None:
runner.add_arg(key='--ep_constraint', arg=ep_constraint)
if target_image is not None:
runner.add_arg(key='--target_image', arg=target_image)
if sharpen_odfs:
runner.add_arg(key='--sharpen_odfs')
if cutoff is not None:
runner.add_arg(key='--cutoff', arg=cutoff)
if odf_cutoff is not None:
runner.add_arg(key='--odf_cutoff', arg=odf_cutoff)
if step_size is not None:
runner.add_arg(key='--step_size', arg=step_size)
if min_tract_length is not None:
runner.add_arg(key='--min_tract_length', arg=min_tract_length)
if angular_threshold is not None:
runner.add_arg(key='--angular_threshold', arg=angular_threshold)
if loop_check is not None:
runner.add_arg(key='--loop_check', arg=loop_check)
if prior_image is not None:
runner.add_arg(key='--prior_image', arg=prior_image)
if prior_weight is not None:
runner.add_arg(key='--prior_weight', arg=prior_weight)
if restrict_to_prior:
runner.add_arg(key='--restrict_to_prior')
if new_directions_from_prior:
runner.add_arg(key='--new_directions_from_prior')
if num_samples is not None:
runner.add_arg(key='--num_samples', arg=num_samples)
if sampling_distance is not None:
runner.add_arg(key='--sampling_distance', arg=sampling_distance)
if use_stop_votes:
runner.add_arg(key='--use_stop_votes')
if use_only_forward_samples:
runner.add_arg(key='--use_only_forward_samples')
if tend_f is not None:
runner.add_arg(key='--tend_f', arg=tend_f)
if tend_g is not None:
runner.add_arg(key='--tend_g', arg=tend_g)
if forest is not None:
runner.add_arg(key='--forest', arg=forest)
if use_sh_features:
runner.add_arg(key='--use_sh_features')
if additional_images is not None:
runner.add_arg(key='--additional_images', arg=additional_images)
runner.run()
return tracts
from cmdint import CmdInterface
from pathlib import Path
""" Example how to call one of MITKs command line tools
"""
# Optional. We set this to change the default logfile name to mitk_example for this example script and to overwrite a
# potentially existing old logfile instead of appending the new command line call.
CmdInterface.set_static_logfile(file='mitk_example.json', delete_existing=True)
CmdInterface.set_throw_on_error(do_throw=False)
# Convert a nrrd image file to a nifti file. Also log input and output file hashes and check if the corresponding
# files are present before or after the command execution respectively by setting check_input and check_output.
converter = CmdInterface('MitkFileFormatConverter')
converter.add_arg('-i',
str(Path.home()) + '/mitk/mitkdata/brain.nrrd',
check_input=True)
converter.add_arg('-o',
str(Path.home()) + '/mitk/mitkdata/brain.nii.gz',
check_output=True)
converter.run()
converter.anonymize_log(CmdInterface.get_static_logfile())
# If you are using a downloaded prebuilt version of MITK, call CmdInterface.set_use_installer(True) or
# directly call MitkFileFormatConverter.sh instead of MitkFileFormatConverter. set_use_installer(True) simply appends
# '.sh' to your command, which is just convenience if you frequently want to switch between installer and non installer
# versions of MITK and don't want to change all your command line calls every time. On Windows you can adjust this to
# append '.bat': CmdInterface.set_use_installer(True, command_suffix='.bat')