def bidsmapper_plugin(session: Path, bidsmap_new: dict, bidsmap_old: dict,
template: dict, store: dict) -> None:
"""
All the logic to map the DICOM/PAR source fields onto bids labels go into this function
:param session: The full-path name of the subject/session raw data source folder
:param bidsmap_new: The new study bidsmap that we are building
:param bidsmap_old: The previous study bidsmap that has precedence over the template bidsmap
:param template: The template bidsmap with the default heuristics
:param store: The paths of the source- and target-folder
:return:
"""
# Get started
plugin = {
'dcm2niix2bids': bidsmap_new['Options']['plugins']['dcm2niix2bids']
}
datasource = bids.get_datasource(session, plugin)
dataformat = datasource.dataformat
if not dataformat:
return
# Collect the different DICOM/PAR source files for all runs in the session
sourcefiles = []
if dataformat == 'DICOM':
for sourcedir in bidscoin.lsdirs(session):
for n in range(
1
): # Option: Use range(2) to scan two files and catch e.g. magnitude1/2 fieldmap files that are stored in one Series folder (but bidscoiner sees only the first file anyhow and it makes bidsmapper 2x slower :-()
sourcefile = bids.get_dicomfile(sourcedir, n)
if sourcefile.name:
sourcefiles.append(sourcefile)
elif dataformat == 'PAR':
sourcefiles = bids.get_parfiles(session)
else:
LOGGER.exception(f"Unsupported dataformat '{dataformat}'")
# Update the bidsmap with the info from the source files
for sourcefile in sourcefiles:
# Input checks
if not sourcefile.name or (not template[dataformat]
and not bidsmap_old[dataformat]):
LOGGER.error(
f"No {dataformat} source information found in the bidsmap and template for: {sourcefile}"
)
return
# See if we can find a matching run in the old bidsmap
datasource = bids.DataSource(sourcefile, plugin, dataformat)
run, match = bids.get_matching_run(datasource, bidsmap_old)
# If not, see if we can find a matching run in the template
if not match:
run, _ = bids.get_matching_run(datasource, template)
# See if we have collected the run somewhere in our new bidsmap
if not bids.exist_run(bidsmap_new, '', run):
# Communicate with the user if the run was not present in bidsmap_old or in template, i.e. that we found a new sample
if not match:
LOGGER.info(
f"Discovered '{datasource.datatype}' {dataformat} sample: {sourcefile}"
)
# Now work from the provenance store
if store:
targetfile = store['target'] / sourcefile.relative_to(
store['source'])
targetfile.parent.mkdir(parents=True, exist_ok=True)
run['provenance'] = str(shutil.copy2(sourcefile, targetfile))
# Copy the filled-in run over to the new bidsmap
bids.append_run(bidsmap_new, run)
else:
# Communicate with the user if the run was already present in bidsmap_old or in template
LOGGER.debug(
f"Known '{datasource.datatype}' {dataformat} sample: {sourcefile}"
)
def bidscoiner_plugin(session: Path, bidsmap: dict, bidsses: Path) -> None:
"""
The bidscoiner plugin to convert the session DICOM and PAR/REC source-files into BIDS-valid nifti-files in the
corresponding bids session-folder and extract personals (e.g. Age, Sex) from the source header
:param session: The full-path name of the subject/session source folder
:param bidsmap: The full mapping heuristics from the bidsmap YAML-file
:param bidsses: The full-path name of the BIDS output `sub-/ses-` folder
:return: Nothing
"""
# Get the subject identifiers and the BIDS root folder from the bidsses folder
if bidsses.name.startswith('ses-'):
bidsfolder = bidsses.parent.parent
subid = bidsses.parent.name
sesid = bidsses.name
else:
bidsfolder = bidsses.parent
subid = bidsses.name
sesid = ''
# Get started and see what dataformat we have
options = bidsmap['Options']['plugins']['dcm2niix2bids']
datasource = bids.get_datasource(session, {'dcm2niix2bids': options})
dataformat = datasource.dataformat
if not dataformat:
LOGGER.info(f"No {__name__} sourcedata found in: {session}")
return
# Make a list of all the data sources / runs
manufacturer = 'UNKNOWN'
sources = []
if dataformat == 'DICOM':
sources = bidscoin.lsdirs(session)
manufacturer = datasource.attributes('Manufacturer')
elif dataformat == 'PAR':
sources = bids.get_parfiles(session)
manufacturer = 'Philips Medical Systems'
else:
LOGGER.exception(f"Unsupported dataformat '{dataformat}'")
# Read or create a scans_table and tsv-file
scans_tsv = bidsses / f"{subid}{bids.add_prefix('_',sesid)}_scans.tsv"
if scans_tsv.is_file():
scans_table = pd.read_csv(scans_tsv, sep='\t', index_col='filename')
else:
scans_table = pd.DataFrame(columns=['acq_time'], dtype='str')
scans_table.index.name = 'filename'
# Process all the source files or run subfolders
sourcefile = Path()
for source in sources:
# Get a sourcefile
if dataformat == 'DICOM':
sourcefile = bids.get_dicomfile(source)
elif dataformat == 'PAR':
sourcefile = source
if not sourcefile.name:
continue
# Get a matching run from the bidsmap
datasource = bids.DataSource(sourcefile, {'dcm2niix2bids': options},
dataformat)
run, match = bids.get_matching_run(datasource, bidsmap, runtime=True)
# Check if we should ignore this run
if datasource.datatype in bidsmap['Options']['bidscoin'][
'ignoretypes']:
LOGGER.info(f"Leaving out: {source}")
continue
# Check if we already know this run
if not match:
LOGGER.error(
f"Skipping unknown '{datasource.datatype}' run: {sourcefile}\n-> Re-run the bidsmapper and delete {bidsses} to solve this warning"
)
continue
LOGGER.info(f"Processing: {source}")
# Create the BIDS session/datatype output folder
if run['bids']['suffix'] in bids.get_derivatives(datasource.datatype):
outfolder = bidsfolder / 'derivatives' / manufacturer.replace(
' ', '') / subid / sesid / datasource.datatype
else:
outfolder = bidsses / datasource.datatype
outfolder.mkdir(parents=True, exist_ok=True)
# Compose the BIDS filename using the matched run
bidsname = bids.get_bidsname(subid, sesid, run, runtime=True)
runindex = run['bids'].get('run', '')
if runindex.startswith('<<') and runindex.endswith('>>'):
bidsname = bids.increment_runindex(outfolder, bidsname)
jsonfiles = [
(outfolder / bidsname).with_suffix('.json')
] # List -> Collect the associated json-files (for updating them later) -- possibly > 1
# Check if file already exists (-> e.g. when a static runindex is used)
if (outfolder / bidsname).with_suffix('.json').is_file():
LOGGER.warning(
f"{outfolder/bidsname}.* already exists and will be deleted -- check your results carefully!"
)
for ext in ('.nii.gz', '.nii', '.json', '.tsv', '.tsv.gz', '.bval',
'.bvec'):
(outfolder / bidsname).with_suffix(ext).unlink(missing_ok=True)
# Convert physiological log files (dcm2niix can't handle these)
if run['bids']['suffix'] == 'physio':
if bids.get_dicomfile(
source, 2).name: # TODO: issue warning or support PAR
LOGGER.warning(
f"Found > 1 DICOM file in {source}, using: {sourcefile}")
physiodata = physio.readphysio(sourcefile)
physio.physio2tsv(physiodata, outfolder / bidsname)
# Convert the source-files in the run folder to nifti's in the BIDS-folder
else:
command = '{command} {args} -f "{filename}" -o "{outfolder}" "{source}"'.format(
command=options['command'],
args=options.get('args', ''),
filename=bidsname,
outfolder=outfolder,
source=source)
if not bidscoin.run_command(command):
if not list(outfolder.glob(f"{bidsname}.nii*")): continue
if list(outfolder.glob(f"{bidsname}a.nii*")):
LOGGER.warning(
f"Unexpected variants of {outfolder/bidsname}* were produced by dcm2niix. Possibly this can be remedied by using the dcm2niix -i option (to ignore derived, localizer and 2D images)"
)
# Replace uncropped output image with the cropped one
if '-x y' in options.get('args', ''):
for dcm2niixfile in sorted(
outfolder.glob(bidsname +
'*_Crop_*')): # e.g. *_Crop_1.nii.gz
ext = ''.join(dcm2niixfile.suffixes)
newbidsfile = str(dcm2niixfile).rsplit(ext, 1)[0].rsplit(
'_Crop_', 1)[0] + ext
LOGGER.info(
f"Found dcm2niix _Crop_ postfix, replacing original file\n{dcm2niixfile} ->\n{newbidsfile}"
)
dcm2niixfile.replace(newbidsfile)
# Rename all files that got additional postfixes from dcm2niix. See: https://github.com/rordenlab/dcm2niix/blob/master/FILENAMING.md
dcm2niixpostfixes = ('_c', '_i', '_Eq', '_real', '_imaginary',
'_MoCo', '_t', '_Tilt', '_e', '_ph', '_ADC',
'_fieldmaphz')
dcm2niixfiles = sorted(
set([
dcm2niixfile for dcm2niixpostfix in dcm2niixpostfixes
for dcm2niixfile in outfolder.glob(
f"{bidsname}*{dcm2niixpostfix}*.nii*")
]))
if not jsonfiles[0].is_file(
) and dcm2niixfiles: # Possibly renamed by dcm2niix, e.g. with multi-echo data (but not always for the first echo)
jsonfiles.pop(0)
for dcm2niixfile in dcm2niixfiles:
ext = ''.join(dcm2niixfile.suffixes)
postfixes = str(dcm2niixfile).split(bidsname)[1].rsplit(
ext)[0].split('_')[1:]
newbidsname = dcm2niixfile.name # Strip the additional postfixes and assign them to bids entities in the for-loop below
for postfix in postfixes: # dcm2niix postfixes _c%d, _e%d and _ph (and any combination of these in that order) are for multi-coil data, multi-echo data and phase data
# Patch the echo entity in the newbidsname with the dcm2niix echo info # NB: We can't rely on the bids-entity info here because manufacturers can e.g. put multiple echos in one series / run-folder
if 'echo' in run['bids'] and postfix.startswith('e'):
echonr = f"_{postfix}".replace('_e',
'') # E.g. postfix='e1'
if not echonr:
echonr = '1'
if echonr.isnumeric():
newbidsname = bids.insert_bidskeyval(
newbidsname, 'echo', echonr.lstrip('0')
) # In contrast to other labels, run and echo labels MUST be integers. Those labels MAY include zero padding, but this is NOT RECOMMENDED to maintain their uniqueness
else:
LOGGER.error(
f"Unexpected postix '{postfix}' found in {dcm2niixfile}"
)
newbidsname = bids.get_bidsvalue(
newbidsname, 'dummy', postfix
) # Append the unknown postfix to the acq-label
# Patch the phase entity in the newbidsname with the dcm2niix mag/phase info
elif 'part' in run['bids'] and postfix in (
'ph', 'real', 'imaginary'
): # e.g. part: ['', 'mag', 'phase', 'real', 'imag', 0]
if postfix == 'ph':
newbidsname = bids.insert_bidskeyval(
newbidsname, 'part', 'phase')
if postfix == 'real':
newbidsname = bids.insert_bidskeyval(
newbidsname, 'part', 'real')
if postfix == 'imaginary':
newbidsname = bids.insert_bidskeyval(
newbidsname, 'part', 'imag')
# Patch fieldmap images (NB: datatype=='fmap' is too broad, see the fmap.yaml file)
elif run['bids']['suffix'] in bids.bidsdatatypes['fmap'][0][
'suffixes']: # i.e. in ('magnitude','magnitude1','magnitude2','phase1','phase2','phasediff','fieldmap'). TODO: Make this robust for future BIDS versions
if len(dcm2niixfiles) not in (
1, 2, 3, 4
): # Phase / echo data may be stored in the same data source / run folder
LOGGER.debug(
f"Unknown fieldmap {outfolder/bidsname} for '{postfix}'"
)
newbidsname = newbidsname.replace(
'_magnitude1a', '_magnitude2'
) # First catch this potential weird / rare case
newbidsname = newbidsname.replace(
'_magnitude1_pha', '_phase2'
) # First catch this potential weird / rare case
newbidsname = newbidsname.replace(
'_magnitude1_e1', '_magnitude1'
) # Case 2 = Two phase and magnitude images
newbidsname = newbidsname.replace(
'_magnitude1_e2', '_magnitude2'
) # Case 2: This can happen when the e2 image is stored in the same directory as the e1 image, but with the e2 listed first
newbidsname = newbidsname.replace(
'_magnitude2_e1', '_magnitude1'
) # Case 2: This can happen when the e2 image is stored in the same directory as the e1 image, but with the e2 listed first
newbidsname = newbidsname.replace(
'_magnitude2_e2', '_magnitude2') # Case 2
if len(dcm2niixfiles) in (
2, 3
): # Case 1 = One or two magnitude + one phasediff image
newbidsname = newbidsname.replace(
'_magnitude1_ph', '_phasediff')
newbidsname = newbidsname.replace(
'_magnitude2_ph', '_phasediff')
newbidsname = newbidsname.replace(
'_phasediff_e1', '_phasediff') # Case 1
newbidsname = newbidsname.replace(
'_phasediff_e2', '_phasediff') # Case 1
newbidsname = newbidsname.replace(
'_phasediff_ph', '_phasediff') # Case 1
newbidsname = newbidsname.replace(
'_magnitude1_ph', '_phase1'
) # Case 2: One or two magnitude and phase images in one folder / datasource
newbidsname = newbidsname.replace(
'_magnitude2_ph', '_phase2'
) # Case 2: Two magnitude + two phase images in one folder / datasource
newbidsname = newbidsname.replace(
'_phase1_e1', '_phase1') # Case 2
newbidsname = newbidsname.replace(
'_phase1_e2', '_phase2'
) # Case 2: This can happen when the e2 image is stored in the same directory as the e1 image, but with the e2 listed first
newbidsname = newbidsname.replace(
'_phase2_e1', '_phase1'
) # Case 2: This can happen when the e2 image is stored in the same directory as the e1 image, but with the e2 listed first
newbidsname = newbidsname.replace(
'_phase2_e2', '_phase2') # Case 2
newbidsname = newbidsname.replace(
'_phase1_ph', '_phase1'
) # Case 2: One or two magnitude and phase images in one folder / datasource
newbidsname = newbidsname.replace(
'_phase2_ph', '_phase2'
) # Case 2: Two magnitude + two phase images in one folder / datasource
newbidsname = newbidsname.replace(
'_magnitude_e1', '_magnitude'
) # Case 3 = One magnitude + one fieldmap image
if len(dcm2niixfiles) == 2:
newbidsname = newbidsname.replace(
'_fieldmap_e1', '_magnitude'
) # Case 3: One magnitude + one fieldmap image in one folder / datasource
newbidsname = newbidsname.replace(
'_fieldmap_e1', '_fieldmap') # Case 3
newbidsname = newbidsname.replace(
'_magnitude_ph', '_fieldmap'
) # Case 3: One magnitude + one fieldmap image in one folder / datasource
newbidsname = newbidsname.replace(
'_fieldmap_ph', '_fieldmap') # Case 3
# Append the dcm2niix info to acq-label, may need to be improved / elaborated for future BIDS standards, supporting multi-coil data
else:
newbidsname = bids.get_bidsvalue(
newbidsname, 'dummy', postfix)
# Remove the added postfix from the new bidsname
newbidsname = newbidsname.replace(f"_{postfix}_",
'_') # If it is not last
newbidsname = newbidsname.replace(f"_{postfix}.",
'.') # If it is last
# The ADC images are not BIDS compliant
if postfix == 'ADC':
LOGGER.warning(
f"The {newbidsname} image is most likely not BIDS-compliant -- you can probably delete it safely and update the scants.tsv file"
)
# Save the nifti file with a new name
if runindex.startswith('<<') and runindex.endswith('>>'):
newbidsname = bids.increment_runindex(
outfolder, newbidsname, ''
) # Update the runindex now that the acq-label has changed
newbidsfile = outfolder / newbidsname
LOGGER.info(
f"Found dcm2niix {postfixes} postfixes, renaming\n{dcm2niixfile} ->\n{newbidsfile}"
)
if newbidsfile.is_file():
LOGGER.warning(
f"Overwriting existing {newbidsfile} file -- check your results carefully!"
)
dcm2niixfile.replace(newbidsfile)
# Rename all associated files (i.e. the json-, bval- and bvec-files)
oldjsonfile = dcm2niixfile.with_suffix('').with_suffix('.json')
newjsonfile = newbidsfile.with_suffix('').with_suffix('.json')
if not oldjsonfile.is_file():
LOGGER.warning(
f"Unexpected file conversion result: {oldjsonfile} not found"
)
else:
if oldjsonfile in jsonfiles:
jsonfiles.remove(oldjsonfile)
if newjsonfile not in jsonfiles:
jsonfiles.append(newjsonfile)
for oldfile in outfolder.glob(
dcm2niixfile.with_suffix('').stem + '.*'):
oldfile.replace(
newjsonfile.with_suffix(''.join(oldfile.suffixes)))
# Copy over the source meta-data
metadata = bids.copymetadata(sourcefile, outfolder / bidsname,
options.get('meta', []))
# Loop over and adapt all the newly produced json sidecar-files and write to the scans.tsv file (NB: assumes every nifti-file comes with a json-file)
for jsonfile in sorted(set(jsonfiles)):
# Load the json meta-data
with jsonfile.open('r') as json_fid:
jsondata = json.load(json_fid)
# Add all the source meta data to the meta-data
for metakey, metaval in metadata.items():
if jsondata.get(metakey) == metaval:
LOGGER.warning(
f"Replacing {metakey} values in {jsonfile}: {jsondata[metakey]} -> {metaval}"
)
jsondata[metakey] = metaval
# Add all the run meta data to the meta-data. NB: the dynamic `IntendedFor` value is handled separately later
for metakey, metaval in run['meta'].items():
if metakey != 'IntendedFor':
metaval = datasource.dynamicvalue(metaval,
cleanup=False,
runtime=True)
try:
metaval = ast.literal_eval(str(metaval))
except (ValueError, SyntaxError):
pass
LOGGER.info(
f"Adding '{metakey}: {metaval}' to: {jsonfile}")
if not metaval:
metaval = None
jsondata[metakey] = metaval
# Remove unused (but added from the template) B0FieldIdentifiers/Sources
if not jsondata.get('B0FieldSource'):
jsondata.pop('B0FieldSource', None)
if not jsondata.get('B0FieldIdentifier'):
jsondata.pop('B0FieldIdentifier', None)
# Save the meta-data to the json sidecar-file
with jsonfile.open('w') as json_fid:
json.dump(jsondata, json_fid, indent=4)
# Parse the acquisition time from the source header or else from the json file (NB: assuming the source file represents the first acquisition)
outputfile = [
file for file in jsonfile.parent.glob(jsonfile.stem + '.*')
if file.suffix in ('.nii', '.gz')
] # Find the corresponding nifti/tsv.gz file (there should be only one, let's not make assumptions about the .gz extension)
if not outputfile:
LOGGER.exception(
f"No data-file found with {jsonfile} when updating {scans_tsv}"
)
elif datasource.datatype not in bidsmap['Options']['bidscoin'][
'bidsignore'] and not run['bids'][
'suffix'] in bids.get_derivatives(datasource.datatype):
acq_time = ''
if dataformat == 'DICOM':
acq_time = f"{datasource.attributes('AcquisitionDate')}T{datasource.attributes('AcquisitionTime')}"
elif dataformat == 'PAR':
acq_time = datasource.attributes('exam_date')
if not acq_time or acq_time == 'T':
acq_time = f"1925-01-01T{jsondata.get('AcquisitionTime','')}"
try:
acq_time = dateutil.parser.parse(acq_time)
if options.get('anon', 'y') in ('y', 'yes'):
acq_time = acq_time.replace(
year=1925, month=1, day=1
) # Privacy protection (see BIDS specification)
acq_time = acq_time.isoformat()
except Exception as jsonerror:
LOGGER.warning(
f"Could not parse the acquisition time from: {sourcefile}\n{jsonerror}"
)
acq_time = 'n/a'
scanpath = outputfile[0].relative_to(bidsses)
scans_table.loc[scanpath.as_posix(), 'acq_time'] = acq_time
# Write the scans_table to disk
LOGGER.info(f"Writing acquisition time data to: {scans_tsv}")
scans_table.sort_values(by=['acq_time', 'filename'], inplace=True)
scans_table.replace('', 'n/a').to_csv(scans_tsv,
sep='\t',
encoding='utf-8',
na_rep='n/a')
# Collect personal data from a source header (PAR/XML does not contain personal info)
personals = {}
if sesid and 'session_id' not in personals:
personals['session_id'] = sesid
personals['age'] = ''
if dataformat == 'DICOM':
age = datasource.attributes(
'PatientAge'
) # A string of characters with one of the following formats: nnnD, nnnW, nnnM, nnnY
if age.endswith('D'): age = float(age.rstrip('D')) / 365.2524
elif age.endswith('W'): age = float(age.rstrip('W')) / 52.1775
elif age.endswith('M'): age = float(age.rstrip('M')) / 12
elif age.endswith('Y'): age = float(age.rstrip('Y'))
if age:
if options.get('anon', 'y') in ('y', 'yes'):
age = int(float(age))
personals['age'] = str(age)
personals['sex'] = datasource.attributes('PatientSex')
personals['size'] = datasource.attributes('PatientSize')
personals['weight'] = datasource.attributes('PatientWeight')
# Store the collected personals in the participants_table
participants_tsv = bidsfolder / 'participants.tsv'
if participants_tsv.is_file():
participants_table = pd.read_csv(participants_tsv, sep='\t', dtype=str)
participants_table.set_index(['participant_id'],
verify_integrity=True,
inplace=True)
else:
participants_table = pd.DataFrame()
participants_table.index.name = 'participant_id'
if subid in participants_table.index and 'session_id' in participants_table.keys(
) and participants_table.loc[subid, 'session_id']:
return # Only take data from the first session -> BIDS specification
for key in personals: # TODO: Check that only values that are consistent over sessions go in the participants.tsv file, otherwise put them in a sessions.tsv file
if key not in participants_table or participants_table[key].isnull(
).get(subid, True) or participants_table[key].get(subid) == 'n/a':
participants_table.loc[subid, key] = personals[key]
# Write the collected data to the participants tsv-file
LOGGER.info(f"Writing {subid} subject data to: {participants_tsv}")
participants_table.replace('', 'n/a').to_csv(participants_tsv,
sep='\t',
encoding='utf-8',
na_rep='n/a')
def coin_data2bids(dataformat: str, session: Path, bidsmap: dict,
bidsfolder: Path, personals: dict, subprefix: str,
sesprefix: str) -> None:
"""
Converts the session source-files into BIDS-valid nifti-files in the corresponding bidsfolder and
extracts personals (e.g. Age, Sex) from the source header
:param dataformat: The format of the raw input data that is to be coined (e.g. 'DICOM' or 'PAR', see bids.get_dataformat)
:param session: The full-path name of the subject/session source file/folder
:param bidsmap: The full mapping heuristics from the bidsmap YAML-file
:param bidsfolder: The full-path name of the BIDS root-folder
:param personals: The dictionary with the personal information
:param subprefix: The prefix common for all source subject-folders
:param sesprefix: The prefix common for all source session-folders
:return: Nothing
"""
# Get valid BIDS subject/session identifiers from the (first) DICOM- or PAR/XML source file
if dataformat == 'DICOM':
sourcefile = Path()
sources = bids.lsdirs(session)
for source in sources:
sourcefile = bids.get_dicomfile(source)
manufacturer = bids.get_dicomfield('Manufacturer', sourcefile)
if sourcefile.name:
break
elif dataformat == 'PAR':
sources = bids.get_parfiles(session)
manufacturer = 'Philips Medical Systems'
if sources:
sourcefile = sources[0]
else:
LOGGER.error(
f"Unsupported data format: {dataformat}\nPlease report this bug")
return
if not sources:
LOGGER.info(f"No data found for: {session}")
return
subid, sesid = bids.get_subid_sesid(sourcefile,
bidsmap[dataformat]['subject'],
bidsmap[dataformat]['session'],
subprefix, sesprefix)
if subid == subprefix:
LOGGER.error(f"No valid subject identifier found for: {session}")
return
# Create the BIDS session-folder and a scans.tsv file
bidsses = bidsfolder / subid / sesid
if bidsses.is_dir():
LOGGER.warning(
f"Existing BIDS output-directory found, which may result in duplicate data (with increased run-index). Make sure {bidsses} was cleaned-up from old data before (re)running the bidscoiner"
)
bidsses.mkdir(parents=True, exist_ok=True)
scans_tsv = bidsses / f"{subid}{bids.add_prefix('_',sesid)}_scans.tsv"
if scans_tsv.is_file():
scans_table = pd.read_csv(scans_tsv, sep='\t', index_col='filename')
else:
scans_table = pd.DataFrame(columns=['acq_time'], dtype='str')
scans_table.index.name = 'filename'
# Process all the source files or run subfolders
for source in sources:
# Get a source-file
if dataformat == 'DICOM':
sourcefile = bids.get_dicomfile(source)
elif dataformat == 'PAR':
sourcefile = source
if not sourcefile.name:
continue
# Get a matching run from the bidsmap
run, datatype, index = bids.get_matching_run(sourcefile, bidsmap,
dataformat)
# Check if we should ignore this run
if datatype == bids.ignoredatatype:
LOGGER.info(f"Leaving out: {source}")
continue
# Check if we already know this run
if index is None:
LOGGER.error(
f"Skipping unknown '{datatype}' run: {sourcefile}\n-> Re-run the bidsmapper and delete {bidsses} to solve this warning"
)
continue
LOGGER.info(f"Processing: {source}")
# Create the BIDS session/datatype output folder
if run['bids']['suffix'] in bids.get_derivatives(datatype):
outfolder = bidsfolder / 'derivatives' / manufacturer.replace(
' ', '') / subid / sesid / datatype
else:
outfolder = bidsses / datatype
outfolder.mkdir(parents=True, exist_ok=True)
# Compose the BIDS filename using the matched run
bidsname = bids.get_bidsname(subid, sesid, run)
runindex = run['bids'].get('run', '')
if runindex.startswith('<<') and runindex.endswith('>>'):
bidsname = bids.increment_runindex(outfolder, bidsname)
jsonfiles = [
(outfolder / bidsname).with_suffix('.json')
] # List -> Collect the associated json-files (for updating them later) -- possibly > 1
# Check if file already exists (-> e.g. when a static runindex is used)
if (outfolder / bidsname).with_suffix('.json').is_file():
LOGGER.warning(
f"{outfolder/bidsname}.* already exists and will be deleted -- check your results carefully!"
)
for ext in ('.nii.gz', '.nii', '.json', '.bval', '.bvec',
'tsv.gz'):
(outfolder / bidsname).with_suffix(ext).unlink(missing_ok=True)
# Convert physiological log files (dcm2niix can't handle these)
if run['bids']['suffix'] == 'physio':
if bids.get_dicomfile(source, 2).name:
LOGGER.warning(
f"Found > 1 DICOM file in {source}, using: {sourcefile}")
physiodata = physio.readphysio(sourcefile)
physio.physio2tsv(physiodata, outfolder / bidsname)
# Convert the source-files in the run folder to nifti's in the BIDS-folder
else:
command = '{path}dcm2niix {args} -f "{filename}" -o "{outfolder}" "{source}"'.format(
path=bidsmap['Options']['dcm2niix']['path'],
args=bidsmap['Options']['dcm2niix']['args'],
filename=bidsname,
outfolder=outfolder,
source=source)
if not bids.run_command(command):
continue
# Replace uncropped output image with the cropped one
if '-x y' in bidsmap['Options']['dcm2niix']['args']:
for dcm2niixfile in sorted(
outfolder.glob(bidsname +
'*_Crop_*')): # e.g. *_Crop_1.nii.gz
ext = ''.join(dcm2niixfile.suffixes)
newbidsfile = str(dcm2niixfile).rsplit(ext, 1)[0].rsplit(
'_Crop_', 1)[0] + ext
LOGGER.info(
f"Found dcm2niix _Crop_ postfix, replacing original file\n{dcm2niixfile} ->\n{newbidsfile}"
)
dcm2niixfile.replace(newbidsfile)
# Rename all files that got additional postfixes from dcm2niix. See: https://github.com/rordenlab/dcm2niix/blob/master/FILENAMING.md
dcm2niixpostfixes = ('_c', '_i', '_Eq', '_real', '_imaginary',
'_MoCo', '_t', '_Tilt', '_e', '_ph')
dcm2niixfiles = sorted(
set([
dcm2niixfile for dcm2niixpostfix in dcm2niixpostfixes
for dcm2niixfile in outfolder.glob(
f"{bidsname}*{dcm2niixpostfix}*")
]))
for dcm2niixfile in dcm2niixfiles:
ext = ''.join(dcm2niixfile.suffixes)
postfixes = str(dcm2niixfile).split(bidsname)[1].rsplit(
ext)[0].split('_')[1:]
newbidsname = dcm2niixfile.name # Strip the additional postfixes and assign them to bids entities in the for-loop below
for postfix in postfixes: # dcm2niix postfixes _c%d, _e%d and _ph (and any combination of these in that order) are for multi-coil data, multi-echo data and phase data
# Patch the echo entity in the newbidsname with the dcm2niix echo info # NB: We can't rely on the bids-entity info here because manufacturers can e.g. put multiple echos in one series / run-folder
if postfix[0] == 'e' and bids.get_bidsvalue(
newbidsname, 'echo'
): # NB: Check if postfix[0]=='e' uniquely refers to the right dcm2niixpostfix
echonr = f"_{postfix}" # E.g. echonr='_e1' or echonr='_pha'
for dcm2niixpostfix in dcm2niixpostfixes:
echonr = echonr.replace(
dcm2niixpostfix, ''
) # Strip the dcm2niixpostfix to keep the echonr info. E.g. [echonr='_e1' or echonr='_pha'] -> [echonr='1' or echonr='a']
if echonr.isalpha():
echonr = ord(
echonr
) - 95 # dcm2niix adds an alphabetically ordered character if it outputs more than one image with the same name. Convert character to echo-number: '' -> 1, 'a'->2, etc
elif not echonr:
echonr = 1
newbidsname = bids.get_bidsvalue(
newbidsname, 'echo', str(echonr)
) # In contrast to other labels, run and echo labels MUST be integers. Those labels MAY include zero padding, but this is NOT RECOMMENDED to maintain their uniqueness
# Patch fieldmap images (NB: datatype=='fmap' is too broad, see the fmap.yaml file)
elif run['bids']['suffix'] in ('magnitude', 'magnitude1',
'magnitude2', 'phase1',
'phase2', 'phasediff',
'fieldmap'):
if len(dcm2niixfiles) not in (
0, 2, 4, 6, 8
): # Phase / echo data may be stored in the same data source / run folder
LOGGER.warning(
f"Unknown fieldmap {outfolder/bidsname} for '{postfix}'"
)
newbidsname = newbidsname.replace(
'_fieldmap_ph', '_fieldmap')
newbidsname = newbidsname.replace(
'_magnitude_e1', '_magnitude')
newbidsname = newbidsname.replace(
'_magnitude_ph', '_fieldmap')
newbidsname = newbidsname.replace(
'_magnitude1_e1', '_magnitude1')
newbidsname = newbidsname.replace(
'_magnitude2_e1', '_magnitude1'
) # This can happen when the e2 image is stored in the same directory as the e1 image, but with the e2 listed first
newbidsname = newbidsname.replace(
'_magnitude1_e2', '_magnitude2')
if len(dcm2niixfiles) == 8:
newbidsname = newbidsname.replace(
'_magnitude1_ph', '_phase1'
) # Two magnitude + 2 phase images in one folder / datasource
else:
newbidsname = newbidsname.replace(
'_magnitude1_ph', '_phasediff'
) # One or two magnitude + 1 phasediff image
newbidsname = newbidsname.replace(
'_magnitude1a', '_magnitude2')
newbidsname = newbidsname.replace(
'_magnitude1_pha', '_phase2')
newbidsname = newbidsname.replace(
'_magnitude2_e2', '_magnitude2')
newbidsname = newbidsname.replace(
'_magnitude2_ph', '_phase2')
newbidsname = newbidsname.replace(
'_phase1_e1', '_phase1')
newbidsname = newbidsname.replace(
'_phase2_e1', '_phase1'
) # This can happen when the e2 image is stored in the same directory as the e1 image, but with the e2 listed first
newbidsname = newbidsname.replace(
'_phase1_ph', '_phase1')
newbidsname = newbidsname.replace(
'_phase1_e2', '_phase2')
newbidsname = newbidsname.replace(
'_phase2_e2', '_phase2')
newbidsname = newbidsname.replace(
'_phase2_ph', '_phase2')
# Append the dcm2niix info to acq-label, may need to be improved / elaborated for future BIDS standards, supporting multi-coil data
else:
newbidsname = bids.get_bidsvalue(
newbidsname, 'dummy', postfix)
# Remove the added postfix from the new bidsname
newbidsname = newbidsname.replace(f"_{postfix}_",
'_') # If it is not last
newbidsname = newbidsname.replace(f"_{postfix}.",
'.') # If it is last
# Save the file with a new name
if runindex.startswith('<<') and runindex.endswith('>>'):
newbidsname = bids.increment_runindex(
outfolder, newbidsname, ''
) # Update the runindex now that the acq-label has changed
newbidsfile = outfolder / newbidsname
LOGGER.info(
f"Found dcm2niix {postfixes} postfixes, renaming\n{dcm2niixfile} ->\n{newbidsfile}"
)
if newbidsfile.is_file():
LOGGER.warning(
f"Overwriting existing {newbidsfile} file -- check your results carefully!"
)
dcm2niixfile.replace(newbidsfile)
if ext == '.json':
oldjsonfile = (outfolder / bidsname).with_suffix('.json')
if oldjsonfile in jsonfiles and not oldjsonfile.is_file():
jsonfiles.remove(
(outfolder / bidsname).with_suffix('.json'))
jsonfiles.append(newbidsfile)
# Loop over and adapt all the newly produced json files and write to the scans.tsv file (every nifti-file comes with a json-file)
for jsonfile in sorted(set(jsonfiles)):
# Check if dcm2niix behaved as expected
if not jsonfile.is_file():
LOGGER.error(
f"Unexpected file conversion result: {jsonfile} not found")
continue
# Add a dummy b0 bval- and bvec-file for any file without a bval/bvec file (e.g. sbref, b0 scans)
if datatype == 'dwi':
bvecfile = jsonfile.with_suffix('.bvec')
bvalfile = jsonfile.with_suffix('.bval')
if not bvecfile.is_file():
LOGGER.info(f"Adding dummy bvec file: {bvecfile}")
with bvecfile.open('w') as bvec_fid:
bvec_fid.write('0\n0\n0\n')
if not bvalfile.is_file():
LOGGER.info(f"Adding dummy bval file: {bvalfile}")
with bvalfile.open('w') as bval_fid:
bval_fid.write('0\n')
# Add the TaskName to the func json-file
elif datatype == 'func':
with jsonfile.open('r') as json_fid:
data = json.load(json_fid)
if not 'TaskName' in data:
LOGGER.info(f"Adding TaskName to: {jsonfile}")
data['TaskName'] = run['bids']['task']
with jsonfile.open('w') as json_fid:
json.dump(data, json_fid, indent=4)
# Parse the acquisition time from the json file or else from the source header (NB: assuming the source file represents the first acquisition)
niifile = list(
jsonfile.parent.glob(jsonfile.stem + '.nii*')
) # Find the corresponding nifti file (there should be only one, let's not make assumptions about the .gz extension)
if niifile and datatype not in bidsmap['Options']['bidscoin'][
'bidsignore'] and not run['bids'][
'suffix'] in bids.get_derivatives(datatype):
with jsonfile.open('r') as json_fid:
data = json.load(json_fid)
if 'AcquisitionTime' not in data or not data['AcquisitionTime']:
data['AcquisitionTime'] = bids.get_sourcefield(
'AcquisitionTime', sourcefile) # DICOM
if not data['AcquisitionTime']:
data['AcquisitionTime'] = bids.get_sourcefield(
'exam_date', sourcefile) # PAR/XML
try:
acq_time = dateutil.parser.parse(data['AcquisitionTime'])
except:
LOGGER.warning(
f"Could not parse the acquisition time from: '{data['AcquisitionTime']}' in {sourcefile}"
)
acq_time = dateutil.parser.parse('00:00:00')
scanpath = niifile[0].relative_to(bidsses)
scans_table.loc[
scanpath.as_posix(),
'acq_time'] = '1925-01-01T' + acq_time.strftime('%H:%M:%S')
# Write the scans_table to disk
LOGGER.info(f"Writing acquisition time data to: {scans_tsv}")
scans_table.sort_values(by=['acq_time', 'filename'], inplace=True)
scans_table.to_csv(scans_tsv, sep='\t', encoding='utf-8')
# Add IntendedFor and TE1+TE2 meta-data to the fieldmap json-files. This has been postponed untill all datatypes have been processed (i.e. so that all target images are indeed on disk)
if bidsmap[dataformat]['fmap'] is not None:
for fieldmap in bidsmap[dataformat]['fmap']:
bidsname = bids.get_bidsname(subid, sesid, fieldmap)
niifiles = []
intendedfor = fieldmap['bids']['IntendedFor']
# Search for the imaging files that match the IntendedFor search criteria
if intendedfor:
if intendedfor.startswith('<<') and intendedfor.endswith('>>'):
intendedfor = intendedfor[2:-2].split('><')
elif not isinstance(intendedfor, list):
intendedfor = [intendedfor]
for selector in intendedfor:
niifiles.extend(
[
Path(niifile).relative_to(bidsfolder / subid)
for niifile in sorted(
bidsses.rglob(f"*{selector}*.nii*"))
if selector
]
) # Search in all runs using a relative path to the subject folder
else:
intendedfor = []
# Get the set of json-files (account for multiple runs in one data source and dcm2niix postfixes inserted into the acquisition label)
jsonfiles = []
acqlabel = bids.get_bidsvalue(bidsname, 'acq')
patterns = (bidsname.replace('_run-1_', '_run-[0-9]*_').replace(
'_magnitude1',
'_magnitude*').replace('_magnitude2', '_magnitude*').replace(
'_phase1', '_phase*').replace('_phase2', '_phase*'),
bidsname.replace('_run-1_', '_run-[0-9]*_').replace(
'_magnitude1',
'_phase*').replace('_magnitude2', '_phase*'))
for pattern in patterns:
jsonfiles.extend((bidsses / 'fmap').glob(pattern + '.json'))
if acqlabel:
cepattern = bids.get_bidsvalue(pattern, 'acq',
acqlabel + '[CE][0-9]*')
jsonfiles.extend(
list((bidsses / 'fmap').glob(cepattern + '.json')))
# Save the meta-data in the jsonfiles
for jsonfile in sorted(set(jsonfiles)):
# Add the IntendedFor data
with jsonfile.open('r') as json_fid:
data = json.load(json_fid)
if 'IntendedFor' not in data:
if niifiles:
LOGGER.info(f"Adding IntendedFor to: {jsonfile}")
elif intendedfor:
LOGGER.warning(
f"Empty 'IntendedFor' fieldmap value in {jsonfile}: the search for {intendedfor} gave no results"
)
else:
LOGGER.warning(
f"Empty 'IntendedFor' fieldmap value in {jsonfile}: the IntendedFor value of the bidsmap entry was empty"
)
data['IntendedFor'] = [
niifile.as_posix() for niifile in niifiles
] # The path needs to use forward slashes instead of backward slashes
with jsonfile.open('w') as json_fid:
json.dump(data, json_fid, indent=4)
# Extract the echo times from magnitude1 and magnitude2 and add them to the phasediff json-file
if jsonfile.name.endswith('phasediff.json'):
json_magnitude = [None, None]
TE = [None, None]
for n in (0, 1):
json_magnitude[
n] = jsonfile.parent / jsonfile.name.replace(
'_phasediff', f"_magnitude{n+1}")
if not json_magnitude[n].is_file():
LOGGER.error(
f"Could not find expected magnitude{n+1} image associated with: {jsonfile}"
)
else:
with json_magnitude[n].open('r') as json_fid:
data = json.load(json_fid)
TE[n] = data['EchoTime']
if None in TE:
LOGGER.error(
f"Cannot find and add valid EchoTime1={TE[0]} and EchoTime2={TE[1]} data to: {jsonfile}"
)
elif TE[0] > TE[1]:
LOGGER.error(
f"Found invalid EchoTime1={TE[0]} > EchoTime2={TE[1]} for: {jsonfile}"
)
else:
with jsonfile.open('r') as json_fid:
data = json.load(json_fid)
data['EchoTime1'] = TE[0]
data['EchoTime2'] = TE[1]
LOGGER.info(
f"Adding EchoTime1: {TE[0]} and EchoTime2: {TE[1]} to {jsonfile}"
)
with jsonfile.open('w') as json_fid:
json.dump(data, json_fid, indent=4)
# Collect personal data from a source header (PAR/XML does not contain personal info)
if dataformat == 'DICOM' and sourcefile.name:
personals['participant_id'] = subid
if sesid:
if 'session_id' not in personals:
personals['session_id'] = sesid
else:
return # Only take data from the first session -> BIDS specification
age = bids.get_dicomfield(
'PatientAge', sourcefile
) # A string of characters with one of the following formats: nnnD, nnnW, nnnM, nnnY
if age.endswith('D'):
personals['age'] = str(int(float(age.rstrip('D')) / 365.2524))
elif age.endswith('W'):
personals['age'] = str(int(float(age.rstrip('W')) / 52.1775))
elif age.endswith('M'):
personals['age'] = str(int(float(age.rstrip('M')) / 12))
elif age.endswith('Y'):
personals['age'] = str(int(float(age.rstrip('Y'))))
elif age:
personals['age'] = age
personals['sex'] = bids.get_dicomfield('PatientSex', sourcefile)
personals['size'] = bids.get_dicomfield('PatientSize', sourcefile)
personals['weight'] = bids.get_dicomfield('PatientWeight', sourcefile)
def coin_data2bids(dataformat: str, session: Path, bidsmap: dict, bidsfolder: Path, personals: dict, subprefix: str, sesprefix: str) -> None:
"""
Converts the session source-files into BIDS-valid nifti-files in the corresponding bidsfolder and
extracts personals (e.g. Age, Sex) from the source header
:param dataformat: The format of the raw input data that is to be coined (e.g. 'DICOM' or 'PAR', see bids.get_dataformat)
:param session: The full-path name of the subject/session source file/folder
:param bidsmap: The full mapping heuristics from the bidsmap YAML-file
:param bidsfolder: The full-path name of the BIDS root-folder
:param personals: The dictionary with the personal information
:param subprefix: The prefix common for all source subject-folders
:param sesprefix: The prefix common for all source session-folders
:return: Nothing
"""
# Get valid BIDS subject/session identifiers from the (first) DICOM- or PAR/XML source file
if dataformat=='DICOM':
sourcefile = Path()
sources = bids.lsdirs(session)
for source in sources:
sourcefile = bids.get_dicomfile(source)
if sourcefile.name:
break
elif dataformat=='PAR':
sources = bids.get_parfiles(session)
if sources:
sourcefile = sources[0]
else:
LOGGER.error(f"Unsupported data format: {dataformat}\nPlease report this bug")
return
if not sources:
LOGGER.info(f"No data found for: {session}")
return
subid, sesid = bids.get_subid_sesid(sourcefile,
bidsmap[dataformat]['subject'],
bidsmap[dataformat]['session'],
subprefix, sesprefix)
if subid == subprefix:
LOGGER.error(f"No valid subject identifier found for: {session}")
return
# Create the BIDS session-folder and a scans.tsv file
bidsses = bidsfolder/subid/sesid
if bidsses.is_dir():
LOGGER.warning(f"Existing BIDS output-directory found, which may result in duplicate data (with increased run-index). Make sure {bidsses} was cleaned-up from old data before (re)running the bidscoiner")
bidsses.mkdir(parents=True, exist_ok=True)
scans_tsv = bidsses/f"{subid}{bids.add_prefix('_',sesid)}_scans.tsv"
if scans_tsv.is_file():
scans_table = pd.read_csv(scans_tsv, sep='\t', index_col='filename')
else:
scans_table = pd.DataFrame(columns=['acq_time'], dtype='str')
scans_table.index.name = 'filename'
# Process all the source files or run subfolders
for source in sources:
# Get a source-file
if dataformat=='DICOM':
sourcefile = bids.get_dicomfile(source)
elif dataformat=='PAR':
sourcefile = source
if not sourcefile.name:
continue
# Get a matching run from the bidsmap
run, modality, index = bids.get_matching_run(sourcefile, bidsmap, dataformat)
# Check if we should ignore this run
if modality == bids.ignoremodality:
LOGGER.info(f"Leaving out: {source}")
continue
# Check if we already know this run
if index is None:
LOGGER.error(f"Skipping unknown '{modality}' run: {sourcefile}\n-> Re-run the bidsmapper and delete {bidsses} to solve this warning")
continue
LOGGER.info(f"Processing: {source}")
# Create the BIDS session/modality folder
bidsmodality = bidsses/modality
bidsmodality.mkdir(parents=True, exist_ok=True)
# Compose the BIDS filename using the matched run
bidsname = bids.get_bidsname(subid, sesid, modality, run)
runindex = run['bids']['run']
if runindex.startswith('<<') and runindex.endswith('>>'):
bidsname = bids.increment_runindex(bidsmodality, bidsname)
# Check if file already exists (-> e.g. when a static runindex is used). TODO: Future dcm2niix versions may contain a `-w 1` option: https://github.com/rordenlab/dcm2niix/issues/276
if (bidsmodality/bidsname).with_suffix('.json').is_file():
LOGGER.warning(f"{bidsmodality/bidsname}.* already exists and will be deleted -- check your results carefully!")
for ext in ('.nii.gz', '.nii', '.json', '.bval', '.bvec'):
(bidsmodality/bidsname).with_suffix(ext).unlink(missing_ok=True)
# Convert the source-files in the run folder to nifti's in the BIDS-folder
command = '{path}dcm2niix {args} -f "{filename}" -o "{outfolder}" "{source}"'.format(
path = bidsmap['Options']['dcm2niix']['path'],
args = bidsmap['Options']['dcm2niix']['args'],
filename = bidsname,
outfolder = bidsmodality,
source = source)
if not bids.run_command(command):
continue
# Replace uncropped output image with the cropped one
if '-x y' in bidsmap['Options']['dcm2niix']['args']:
for filename in sorted(bidsmodality.glob(bidsname + '*_Crop_*')): # e.g. *_Crop_1.nii.gz
ext = ''.join(filename.suffixes)
newfilename = str(filename).rsplit(ext,1)[0].rsplit('_Crop_',1)[0] + ext
LOGGER.info(f"Found dcm2niix _Crop_ suffix, replacing original file\n{filename} ->\n{newfilename}")
filename.replace(newfilename)
# Rename all files ending with _c%d, _e%d and _ph (and any combination of these) that are added by dcm2niix for multi-coil data, multi-echo data and phase data
# See: https://github.com/rordenlab/dcm2niix/blob/master/FILENAMING.md
jsonfiles = [] # Collect the associated json-files (for updating them later) -- possibly > 1
for dcm2niisuffix in ('_c', '_e', '_ph', '_i', '_Eq', '_real', '_imaginary', '_MoCo', '_t', '_Tilt'):
for filename in sorted(bidsmodality.glob(f"{bidsname}*{dcm2niisuffix}*")):
ext = ''.join(filename.suffixes)
basepath, index = str(filename).rsplit(ext)[0].rsplit(dcm2niisuffix,1) # basepath = the name without the added stuff (i.e. bidsmodality/bidsname), index = added dcm2niix index (e.g. _c1 -> index=1)
basesuffix = basepath.rsplit('_',1)[1] # The BIDS suffix, e.g. basepath = *_magnitude1 -> basesuffix=magnitude1
index = index.split('_')[0].zfill(2) # Zero padd as specified in the BIDS-standard (assuming two digits is sufficient); strip following suffices (fieldmaps produce *_e2_ph files)
# Phase data may be stored in the magnitude data source (e.g. Philips fieldmaps)
if 'ph' in filename.name.rsplit(ext)[0].split('_'):
basepath = basepath.replace('_magnitude', '_phase')
# This is a special hack: dcm2niix does not always add a _c/_e suffix for the first(?) coil/echo image -> add it when we encounter a **_e2/_c2 file
# https://github.com/rordenlab/dcm2niix/issues/381
if dcm2niisuffix in ('_c','_e') and int(index)==2 and basesuffix not in ['magnitude1', 'phase1']: # For fieldmaps: *_magnitude1_e[index] -> *_magnitude[index] (This is handled below)
filename_ce = Path(basepath + ext) # The file without the _c1/_e1 suffix
if dcm2niisuffix=='_e' and bids.get_bidsvalue(basepath, 'echo'):
newbasepath_ce = Path(bids.get_bidsvalue(basepath, 'echo', '1'))
else:
newbasepath_ce = Path(bids.get_bidsvalue(basepath, 'dummy', dcm2niisuffix.upper()[1:] + '1'.zfill(len(index)))) # --> append to acq-label, may need to be elaborated for future BIDS standards, supporting multi-coil data
newfilename_ce = newbasepath_ce.with_suffix(ext) # The file as it should have been
if filename_ce.is_file():
if filename_ce != newfilename_ce:
LOGGER.warning(f"Found no dcm2niix {dcm2niisuffix} suffix for image instance 1, renaming\n{filename_ce} ->\n{newfilename_ce}\nConsider upgrading dcm2niix: https://github.com/rordenlab/dcm2niix/issues/381")
if newfilename_ce.is_file():
LOGGER.warning(f"Overwriting existing {newfilename_ce} file -- check your results carefully!")
filename_ce.replace(newfilename_ce)
if ext == '.json':
jsonfiles.append(newbasepath_ce.with_suffix('.json'))
# Patch the basepath with the dcm2niix suffix info (we can't rely on the basepath info here because Siemens can e.g. put multiple echos in one series / run-folder)
if dcm2niisuffix=='_e' and bids.get_bidsvalue(basepath, 'echo') and index:
basepath = bids.get_bidsvalue(basepath, 'echo', str(int(index))) # In contrast to other labels, run and echo labels MUST be integers. Those labels MAY include zero padding, but this is NOT RECOMMENDED to maintain their uniqueness
elif dcm2niisuffix=='_e' and basesuffix in ('magnitude1','magnitude2','phase1','phase2') and index: # i.e. modality == 'fmap'
basepath = basepath[0:-1] + str(int(index)) # basepath: *_magnitude1_e[index] -> *_magnitude[index] and *_phase1_e[index]_ph -> *_phase[index]
elif dcm2niisuffix=='_e' and basesuffix=='phasediff' and index: # i.e. modality == 'fmap'
pass
else:
basepath = bids.get_bidsvalue(basepath, 'dummy', dcm2niisuffix.upper()[1:] + index) # --> append to acq-label, may need to be elaborated for future BIDS standards, supporting multi-coil data
# Save the file with a new name
newbidsname = str(Path(basepath).name)
if runindex.startswith('<<') and runindex.endswith('>>'):
newbidsname = bids.increment_runindex(bidsmodality, newbidsname, ext) # Update the runindex now that the acq-label has changed
newfilename = (bidsmodality/newbidsname).with_suffix(ext)
LOGGER.info(f"Found dcm2niix {dcm2niisuffix} suffix, renaming\n{filename} ->\n{newfilename}")
if newfilename.is_file():
LOGGER.warning(f"Overwriting existing {newfilename} file -- check your results carefully!")
filename.replace(newfilename)
if ext == '.json':
jsonfiles.append((bidsmodality/newbidsname).with_suffix('.json'))
# Loop over and adapt all the newly produced json files and write to the scans.tsv file (every nifti-file comes with a json-file)
if not jsonfiles:
jsonfiles = [(bidsmodality/bidsname).with_suffix('.json')]
for jsonfile in set(jsonfiles):
# Check if dcm2niix behaved as expected
if not jsonfile.is_file():
LOGGER.error(f"Unexpected file conversion result: {jsonfile} not found")
continue
# Add a dummy b0 bval- and bvec-file for any file without a bval/bvec file (e.g. sbref, b0 scans)
if modality == 'dwi':
bvecfile = jsonfile.with_suffix('.bvec')
bvalfile = jsonfile.with_suffix('.bval')
if not bvecfile.is_file():
LOGGER.info(f"Adding dummy bvec file: {bvecfile}")
with bvecfile.open('w') as bvec_fid:
bvec_fid.write('0\n0\n0\n')
if not bvalfile.is_file():
LOGGER.info(f"Adding dummy bval file: {bvalfile}")
with bvalfile.open('w') as bval_fid:
bval_fid.write('0\n')
# Add the TaskName to the func json-file
elif modality == 'func':
with jsonfile.open('r') as json_fid:
data = json.load(json_fid)
if not 'TaskName' in data:
LOGGER.info(f"Adding TaskName to: {jsonfile}")
data['TaskName'] = run['bids']['task']
with jsonfile.open('w') as json_fid:
json.dump(data, json_fid, indent=4)
# Parse the acquisition time from the json file or else from the source header (NB: assuming the source file represents the first acquisition)
if bidsmodality.name not in bidsmap['Options']['bidscoin']['bidsignore']:
with jsonfile.open('r') as json_fid:
data = json.load(json_fid)
if 'AcquisitionTime' not in data or not data['AcquisitionTime']:
data['AcquisitionTime'] = bids.get_sourcefield('AcquisitionTime', sourcefile) # DICOM
if not data['AcquisitionTime']:
data['AcquisitionTime'] = bids.get_sourcefield('exam_date', sourcefile) # PAR/XML
try:
acq_time = dateutil.parser.parse(data['AcquisitionTime'])
except:
LOGGER.warning(f"Could not parse the acquisition time from: '{data['AcquisitionTime']}' in {sourcefile}")
acq_time = dateutil.parser.parse('00:00:00')
scanpath = list(jsonfile.parent.glob(jsonfile.stem + '.nii*'))[0].relative_to(bidsses) # Find the corresponding nifti file (there should be only one, let's not make assumptions about the .gz extension)
scans_table.loc[scanpath.as_posix(), 'acq_time'] = '1925-01-01T' + acq_time.strftime('%H:%M:%S')
# Write the scans_table to disk
LOGGER.info(f"Writing acquisition time data to: {scans_tsv}")
scans_table.sort_values(by=['acq_time','filename'], inplace=True)
scans_table.to_csv(scans_tsv, sep='\t', encoding='utf-8')
# Add IntendedFor and TE1+TE2 meta-data the fieldmap json-files. This has been postponed untill all modalities have been processed (i.e. so that all target images are indeed on disk)
if bidsmap[dataformat]['fmap'] is not None:
for fieldmap in bidsmap[dataformat]['fmap']:
bidsname = bids.get_bidsname(subid, sesid, 'fmap', fieldmap)
niifiles = []
intendedfor = fieldmap['bids']['IntendedFor']
# Search for the imaging files that match the IntendedFor search criteria
if intendedfor:
if intendedfor.startswith('<<') and intendedfor.endswith('>>'):
intendedfor = intendedfor[2:-2].split('><')
elif not isinstance(intendedfor, list):
intendedfor = [intendedfor]
for selector in intendedfor:
niifiles.extend([Path(niifile).relative_to(bidsfolder/subid)
for niifile in sorted(bidsses.rglob(f"*{selector}*.nii*")) if selector]) # Search in all runs using a relative path to the subject folder
else:
intendedfor = []
# Get the set of json-files (account for multiple runs in one data source and dcm2niix suffixes inserted into the acquisition label)
jsonfiles = []
acqlabel = bids.get_bidsvalue(bidsname, 'acq')
patterns = (bidsname.replace('_run-1_', '_run-[0-9]*_').
replace('_magnitude1', '_magnitude*').
replace('_magnitude2', '_magnitude*').
replace('_phase1', '_phase*').
replace('_phase2', '_phase*'),
bidsname.replace('_run-1_', '_run-[0-9]*_').
replace('_magnitude1', '_phase*').
replace('_magnitude2', '_phase*'))
for pattern in patterns:
jsonfiles.extend((bidsses/'fmap').glob(pattern + '.json'))
if acqlabel:
cepattern = bids.get_bidsvalue(pattern, 'acq', acqlabel + '[CE][0-9]*')
jsonfiles.extend(list((bidsses/'fmap').glob(cepattern + '.json')))
# Save the meta-data in the jsonfiles
for jsonfile in set(jsonfiles):
# Add the IntendedFor data
with jsonfile.open('r') as json_fid:
data = json.load(json_fid)
if 'IntendedFor' not in data:
if niifiles:
LOGGER.info(f"Adding IntendedFor to: {jsonfile}")
elif intendedfor:
LOGGER.warning(f"Empty 'IntendedFor' fieldmap value in {jsonfile}: the search for {intendedfor} gave no results")
else:
LOGGER.warning(f"Empty 'IntendedFor' fieldmap value in {jsonfile}: the IntendedFor value of the bidsmap entry was empty")
data['IntendedFor'] = [niifile.as_posix() for niifile in niifiles] # The path needs to use forward slashes instead of backward slashes
with jsonfile.open('w') as json_fid:
json.dump(data, json_fid, indent=4)
# Extract the echo times from magnitude1 and magnitude2 and add them to the phasediff json-file
if jsonfile.name.endswith('phasediff.json'):
json_magnitude = [None, None]
TE = [None, None]
for n in (0,1):
json_magnitude[n] = jsonfile.parent/jsonfile.name.replace('_phasediff', f"_magnitude{n+1}")
if not json_magnitude[n].is_file():
LOGGER.error(f"Could not find expected magnitude{n+1} image associated with: {jsonfile}")
else:
with json_magnitude[n].open('r') as json_fid:
data = json.load(json_fid)
TE[n] = data['EchoTime']
if None in TE:
LOGGER.error(f"Cannot find and add valid EchoTime1={TE[0]} and EchoTime2={TE[1]} data to: {jsonfile}")
elif TE[0] > TE[1]:
LOGGER.error(f"Found invalid EchoTime1={TE[0]} > EchoTime2={TE[1]} for: {jsonfile}")
else:
with jsonfile.open('r') as json_fid:
data = json.load(json_fid)
data['EchoTime1'] = TE[0]
data['EchoTime2'] = TE[1]
LOGGER.info(f"Adding EchoTime1: {TE[0]} and EchoTime2: {TE[1]} to {jsonfile}")
with jsonfile.open('w') as json_fid:
json.dump(data, json_fid, indent=4)
# Collect personal data from a source header (PAR/XML does not contain personal info)
if dataformat=='DICOM' and sourcefile.name:
personals['participant_id'] = subid
if sesid:
if 'session_id' not in personals:
personals['session_id'] = sesid
else:
return # Only from the first session -> BIDS specification
age = bids.get_dicomfield('PatientAge', sourcefile) # A string of characters with one of the following formats: nnnD, nnnW, nnnM, nnnY
if age.endswith('D'):
personals['age'] = str(int(float(age.rstrip('D'))/365.2524))
elif age.endswith('W'):
personals['age'] = str(int(float(age.rstrip('W'))/52.1775))
elif age.endswith('M'):
personals['age'] = str(int(float(age.rstrip('M'))/12))
elif age.endswith('Y'):
personals['age'] = str(int(float(age.rstrip('Y'))))
elif age:
personals['age'] = age
personals['sex'] = bids.get_dicomfield('PatientSex', sourcefile)
personals['size'] = bids.get_dicomfield('PatientSize', sourcefile)
personals['weight'] = bids.get_dicomfield('PatientWeight', sourcefile)
def bidsmapper(rawfolder: str,
bidsfolder: str,
bidsmapfile: str,
templatefile: str,
subprefix: str = 'sub-',
sesprefix: str = 'ses-',
store: bool = False,
interactive: bool = True) -> None:
"""
Main function that processes all the subjects and session in the sourcefolder
and that generates a maximally filled-in bidsmap.yaml file in bidsfolder/code/bidscoin.
Folders in sourcefolder are assumed to contain a single dataset.
:param rawfolder: The root folder-name of the sub/ses/data/file tree containing the source data files
:param bidsfolder: The name of the BIDS root folder
:param bidsmapfile: The name of the bidsmap YAML-file
:param templatefile: The name of the bidsmap template YAML-file
:param subprefix: The prefix common for all source subject-folders
:param sesprefix: The prefix common for all source session-folders
:param store: If True, the provenance samples will be stored
:param interactive: If True, the user will be asked for help if an unknown run is encountered
:return:bidsmapfile: The name of the mapped bidsmap YAML-file
"""
# Input checking
rawfolder = Path(rawfolder).resolve()
bidsfolder = Path(bidsfolder).resolve()
bidsmapfile = Path(bidsmapfile)
templatefile = Path(templatefile)
bidscoinfolder = bidsfolder / 'code' / 'bidscoin'
# Start logging
bids.setup_logging(bidscoinfolder / 'bidsmapper.log')
LOGGER.info('')
LOGGER.info('-------------- START BIDSmapper ------------')
LOGGER.info(
f">>> bidsmapper sourcefolder={rawfolder} bidsfolder={bidsfolder} bidsmap={bidsmapfile} "
f" template={templatefile} subprefix={subprefix} sesprefix={sesprefix} store={store} interactive={interactive}"
)
# Get the heuristics for filling the new bidsmap
bidsmap_old, _ = bids.load_bidsmap(bidsmapfile, bidscoinfolder)
template, _ = bids.load_bidsmap(templatefile, bidscoinfolder)
# Create the new bidsmap as a copy / bidsmap skeleton with no modality entries (i.e. bidsmap with empty lists)
if bidsmap_old:
bidsmap_new = copy.deepcopy(bidsmap_old)
else:
bidsmap_new = copy.deepcopy(template)
for logic in ('DICOM', 'PAR', 'P7', 'Nifti', 'FileSystem'):
for modality in bids.bidsmodalities + (bids.unknownmodality,
bids.ignoremodality):
if bidsmap_new[logic] and modality in bidsmap_new[logic]:
bidsmap_new[logic][modality] = None
# Start with an empty skeleton if we didn't have an old bidsmap
if not bidsmap_old:
bidsmap_old = copy.deepcopy(bidsmap_new)
# Start the Qt-application
gui = interactive
if gui:
app = QApplication(sys.argv)
app.setApplicationName('BIDS editor')
mainwin = bidseditor.MainWindow()
gui = bidseditor.Ui_MainWindow()
gui.interactive = interactive
gui.subprefix = subprefix
gui.sesprefix = sesprefix
if gui.interactive == 2:
QMessageBox.information(
mainwin, 'BIDS mapping workflow',
f"The bidsmapper will now scan {bidsfolder} and whenever "
f"it detects a new type of scan it will ask you to identify it.\n\n"
f"It is important that you choose the correct BIDS modality "
f"(e.g. 'anat', 'dwi' or 'func') and suffix (e.g. 'bold' or 'sbref').\n\n"
f"At the end you will be shown an overview of all the "
f"different scan types and BIDScoin options (as in the "
f"bidseditor) that you can then (re)edit to your needs")
# Loop over all subjects and sessions and built up the bidsmap entries
dataformat = ''
subjects = bids.lsdirs(rawfolder, subprefix + '*')
if not subjects:
LOGGER.warning(f'No subjects found in: {rawfolder/subprefix}*')
gui = None
for n, subject in enumerate(subjects, 1):
sessions = bids.lsdirs(subject, sesprefix + '*')
if not sessions:
sessions = [subject]
for session in sessions:
# Unpack the data in a temporary folder if it is tarballed/zipped and/or contains a DICOMDIR file
session, unpacked = bids.unpack(session, subprefix, sesprefix, '*')
if unpacked:
store = dict(source=unpacked,
target=bidscoinfolder / 'provenance')
elif store:
store = dict(source=rawfolder,
target=bidscoinfolder / 'provenance')
else:
store = dict()
# Loop of the different DICOM runs (series) and collect source files
sourcefiles = []
dataformat = bids.get_dataformat(session)
if not dataformat:
LOGGER.info(
f"Skipping: {session} (subject {n}/{len(subjects)})")
continue
LOGGER.info(f"Parsing: {session} (subject {n}/{len(subjects)})")
if dataformat == 'DICOM':
for sourcedir in bids.lsdirs(session):
sourcefile = bids.get_dicomfile(sourcedir)
if sourcefile.name:
sourcefiles.append(sourcefile)
if dataformat == 'PAR':
sourcefiles = bids.get_parfiles(session)
if dataformat == 'P7':
sourcefiles = bids.get_p7file(session)
# Update the bidsmap with the info from the source files
for sourcefile in sourcefiles:
bidsmap_new = build_bidsmap(dataformat, sourcefile,
bidsmap_new, bidsmap_old, template,
store, gui)
# Update / append the nifti mapping
if dataformat == 'Nifti':
bidsmap_new = build_niftimap(session, bidsmap_new, bidsmap_old)
# Update / append the file-system mapping
if dataformat == 'FileSystem':
bidsmap_new = build_filesystemmap(session, bidsmap_new,
bidsmap_old)
# Update / append the plugin mapping
if bidsmap_old['PlugIns']:
bidsmap_new = build_pluginmap(session, bidsmap_new,
bidsmap_old)
# Clean-up the temporary unpacked data
if unpacked:
shutil.rmtree(session)
if not dataformat:
LOGGER.warning('Could not determine the dataformat of the source data')
# (Re)launch the bidseditor UI_MainWindow
bidsmapfile = bidscoinfolder / 'bidsmap.yaml'
if gui:
if not dataformat:
QMessageBox.information(
mainwin, 'BIDS mapping workflow',
'Could not determine the dataformat of the source data.\n'
'You can try running the bidseditor tool yourself')
else:
QMessageBox.information(
mainwin, 'BIDS mapping workflow',
f"The bidsmapper has finished scanning {rawfolder}\n\n"
f"Please carefully check all the different BIDS output names "
f"and BIDScoin options and (re)edit them to your needs.\n\n"
f"You can always redo this step later by re-running the "
f"bidsmapper or by just running the bidseditor tool")
LOGGER.info('Opening the bidseditor')
gui.setupUi(mainwin,
bidsfolder,
bidsmapfile,
bidsmap_new,
copy.deepcopy(bidsmap_new),
template,
dataformat,
subprefix=subprefix,
sesprefix=sesprefix)
mainwin.show()
app.exec()
else:
# Save the bidsmap in the bidscoinfolder
bids.save_bidsmap(bidsmapfile, bidsmap_new)
LOGGER.info('-------------- FINISHED! -------------------')
LOGGER.info('')
bids.reporterrors()
def scanparticipant(dataformat: str, session: Path, personals: dict,
subid: str, sesid: str) -> bool:
"""
Converts the session source-files into BIDS-valid nifti-files in the corresponding bidsfolder and
extracts personals (e.g. Age, Sex) from the source header
:param session: The full-path name of the subject/session source file/folder
:param personals: The dictionary with the personal information
:param subid: The subject-id from the bids-folder
:param sesid: The session-id from the bids-folder
:return: True if successful
"""
# Get valid BIDS subject/session identifiers from the (first) DICOM- or PAR/XML source file
sourcefile = Path()
if dataformat == 'DICOM':
sources = bids.lsdirs(session)
for source in sources:
sourcefile = bids.get_dicomfile(source)
if sourcefile.name:
break
elif dataformat == 'PAR':
sources = bids.get_parfiles(session)
if sources:
sourcefile = sources[0]
else:
LOGGER.error(
f"Unsupported data format: {dataformat}\nPlease report this bug")
return False
if not sources:
LOGGER.info(f"No data found for: {session}")
return False
# Collect personal data from a source header (PAR/XML does not contain personal info)
if dataformat == 'DICOM' and sourcefile.name:
personals['participant_id'] = subid
if sesid:
if 'session_id' not in personals:
personals['session_id'] = sesid
else:
return False # Only from the first session -> BIDS specification
age = bids.get_dicomfield(
'PatientAge', sourcefile
) # A string of characters with one of the following formats: nnnD, nnnW, nnnM, nnnY
if age.endswith('D'):
personals['age'] = str(int(float(age.rstrip('D')) / 365.2524))
elif age.endswith('W'):
personals['age'] = str(int(float(age.rstrip('W')) / 52.1775))
elif age.endswith('M'):
personals['age'] = str(int(float(age.rstrip('M')) / 12))
elif age.endswith('Y'):
personals['age'] = str(int(float(age.rstrip('Y'))))
elif age:
personals['age'] = age
personals['sex'] = bids.get_dicomfield('PatientSex', sourcefile)
personals['size'] = bids.get_dicomfield('PatientSize', sourcefile)
personals['weight'] = bids.get_dicomfield('PatientWeight', sourcefile)
return True
