def main(tgz, newid, outdir=None):
""" untars tgz, replaces PatientID PatientName
with new id, and save to outdir (default origdir/newtgz"""
# mv dicom to tmp directory
tgz = os.path.abspath(tgz)
origdir, orignme = os.path.split(tgz)
tmptgz = bg.copy_tmpdir(tgz)
# un-archive
pth = bg.tar_cmd(tmptgz)
newdir, exists = bg.make_dir(pth, dirname='dicomfiles')
startdir = os.getcwd()
os.chdir(pth)
dcms = bg.find_dicoms(pth)
keys = dcms.keys()
keys = [x for x in keys if not newdir in x]
for k in keys:
for dcm in dcms[k]:
plan = dicom.read_file(dcm)
plan.PatientID = newid
plan.PatientName = newid
_, dcm_name = os.path.split(dcm)
newdcm = os.path.join(newdir, dcm_name)
dicom.write_file(newdcm, plan)
# create tar archive of updated dicoms
if outdir is None:
outdir, _ = bg.make_dir(origdir, dirname="newtgz")
newtgz = os.path.join(outdir, orignme)
cmd = 'tar cfvz %s dicomfiles'%(newtgz)
os.system(cmd)
os.chdir(startdir)
print 'removing %s'%pth
os.system('rm -rf %s'%(pth))
print 'wrote ', newtgz
return newtgz
def save_as(self, filename, write_like_original=True):
"""Write the dataset to a file.
:param filename: full path and filename to save the file to
:write_like_original: see dicom.filewriter.write_file for info on this parameter.
"""
dicom.write_file(filename, self, write_like_original)
def main(tgz, newid, outdir=None):
""" untars tgz, replaces PatientID PatientName
with new id, and save to outdir (default origdir/newtgz"""
# mv dicom to tmp directory
tgz = os.path.abspath(tgz)
origdir, orignme = os.path.split(tgz)
tmptgz = bg.copy_tmpdir(tgz)
# un-archive
pth = bg.tar_cmd(tmptgz)
newdir, exists = bg.make_dir(pth, dirname='dicomfiles')
startdir = os.getcwd()
os.chdir(pth)
dcms = bg.find_dicoms(pth)
keys = dcms.keys()
keys = [x for x in keys if not newdir in x]
for k in keys:
for dcm in dcms[k]:
plan = dicom.read_file(dcm)
plan.PatientID = newid
plan.PatientName = newid
_, dcm_name = os.path.split(dcm)
newdcm = os.path.join(newdir, dcm_name)
dicom.write_file(newdcm, plan)
# create tar archive of updated dicoms
if outdir is None:
outdir, _ = bg.make_dir(origdir, dirname="newtgz")
newtgz = os.path.join(outdir, orignme)
cmd = 'tar cfvz %s dicomfiles' % (newtgz)
os.system(cmd)
os.chdir(startdir)
print 'removing %s' % pth
os.system('rm -rf %s' % (pth))
print 'wrote ', newtgz
return newtgz
def save_as(self, filename, write_like_original=True):
"""Write the dataset to a file.
:param filename: full path and filename to save the file to
:write_like_original: see dicom.filewriter.write_file for info on this parameter.
"""
dicom.write_file(filename, self, write_like_original)
def save_as(self, filename, WriteLikeOriginal=True):
"""Write the dataset to a file.
filename -- full path and filename to save the file to
WriteLikeOriginal -- see dicom.filewriter.write_file for info on this parameter.
"""
dicom.write_file(filename, self, WriteLikeOriginal)
def save_as(self, filename, WriteLikeOriginal=True):
"""Write the dataset to a file.
filename -- full path and filename to save the file to
WriteLikeOriginal -- see dicom.filewriter.write_file for info on this parameter.
"""
dicom.write_file(filename, self, WriteLikeOriginal)
def write(self, outdir='.', print_filenames=False):
for modality in self.modalityorder:
for sb in self.seriesbuilders[modality]:
print modality, sb
for ds in sb.build():
dicom.write_file(os.path.join(outdir, ds.filename), ds)
if print_filenames:
print ds.filename
def write(self, outdir='.', print_filenames=False):
for modality in self.modalityorder:
for sb in self.seriesbuilders[modality]:
print modality, sb
for ds in sb.build():
dicom.write_file(os.path.join(outdir, ds.filename), ds)
if print_filenames:
print ds.filename
def _fix_sitk_bug(self, path, metadata):
"""
There is a bug in simple ITK for Z axis in 3D images. This is a fix
:param path:
:param metadata:
:return:
"""
ds = dicom.read_file(path)
ds.SpacingBetweenSlices = str(metadata["voxelsize_mm"][0])[:16]
dicom.write_file(path, ds)
def _fix_sitk_bug(self, path, metadata):
"""
There is a bug in simple ITK for Z axis in 3D images. This is a fix
:param path:
:param metadata:
:return:
"""
ds = dicom.read_file(path)
ds.SpacingBetweenSlices = str(metadata["voxelsize_mm"][0])[:16]
dicom.write_file(path, ds)
def replace_iso_gantry_spots(pdir,
x=None,
y=None,
z=None,
angle=None,
fgs=None,
ibs=None):
rtip = [
f for f in os.listdir(pdir) if 'RTIP' in f and '.dcm' in f.lower()
][0]
data = pd.read_file(pdir + rtip)
num_beams = len(data.IonBeamSequence)
if x and y and z:
for i in range(num_beams):
data.IonBeamSequence[i].IonControlPointSequence[
0].IsocenterPosition[0] = str(x)
data.IonBeamSequence[i].IonControlPointSequence[
0].IsocenterPosition[1] = str(y)
data.IonBeamSequence[i].IonControlPointSequence[
0].IsocenterPosition[2] = str(z)
else:
pass
if angle:
for i in range(num_beams):
data.IonBeamSequence[i].IonControlPointSequence[
0].GantryAngle = str(angle)
else:
pass
#Copy fraction group sequence and ion beam sequence if we are making a new plan
if fgs and ibs:
data.FractionGroupSequence = fgs
data.IonBeamSequence = ibs
else:
pass
#Make sure other items are dicom-compliant (labels/names/geometry/)
data.RTPlanLabel = 'label'
data.RTPlanName = 'name'
data.RTPlanGeometry = 'PATIENT'
data.FractionGroupSequence[0].FractionGroupNumber = '1'
#Check the setup beam
if data.IonBeamSequence[0].IonControlPointSequence[
0].NominalBeamEnergy == '0':
data.IonBeamSequence[0].TreatmentDeliveryType = 'SETUP'
data.IonBeamSequence[0].IonControlPointSequence[
0].PatientSupportAngle = '90'
#Write data to file
pd.write_file(pdir + rtip[0], data)
def write_ds(ds, fn, default_sopclass=None):
ds.file_meta = dicom.dataset.Dataset()
ds.file_meta.TransferSyntaxUID = dicom.UID.ImplicitVRLittleEndian
if default_sopclass == None:
default_sopclass = get_uid("Study Root Query/Retrieve Information Model - FIND")
ds.file_meta.MediaStorageSOPClassUID = getattr(ds, 'SOPClassUID', default_sopclass)
ds.file_meta.MediaStorageSOPInstanceUID = getattr(ds, 'SOPInstanceUID', generate_uid())
ds.is_little_endian = True
ds.is_implicit_VR = True
ds.file_meta.ImplementationClassUID = '2.25.4282708245307149051252828097685724107'
dicom.write_file(fn, ds, WriteLikeOriginal=False)
def updateDicomFile(self, event):
"""
Переопределяет секцию байтов, отвечающих за PN, записывает изменения в текущий файл
и рендерит его сигнатуру
"""
self.pre_save_ds[0x10, 0x10].value = self.patientname
dicom.write_file(self.filename, self.pre_save_ds)
self.status_text.delete(3.0, END)
str = u"Выбран файл: %s" % self.filename
ds = dicom.read_file(self.filename)
self.fillTextField(str, ds)
def export_dynalog_plan(self,plan_name,filename,export_expected=False,leafgap=0.7):
"""
Parameter
-----------------------------------------------------------------------
plan_name : str
Wert des RTPlanLabel-Tags im exportierten Plan. Maximal 13
Zeichen, längere Namen werden abgeschnitten.
UID : str
Wert des SOPInstanceUID-Tags im exportierten Plan. Muss der UID-Syntax
entsprechen.
filename : str
Der Dateiname, unter dem das exportierte RTPLAN Objekt abgelegt
werden soll.
Beschreibung
-----------------------------------------------------------------------
Exportiert den derzeitigen Planzustand als DICOM-Objekt. Basis ist eine
Kopie des DICOM-Files mit dem der Plan initialisiert wurde. StudyInstance,
SeriesInstance und Study UIDs werden ersetzt/geändert. Alle im DynaLog
enthaltenen Werte werden anstelle der Originalparameter exportiert.
Plan muss validiert sein bevor der Export erfolgen kann!
Ausgabe
-----------------------------------------------------------------------
output : DICOM RTPLAN Objekt
Planobjekt mit DynaLog-Werten.
"""
if self.validated == False:
raise PlanMismatchError(self.header["plan_uid"],"validation",
"can't export unvalidated plan.")
exportplan = copy.deepcopy(self.dicom_data)
for num in range(len(self.beams)):
exportplan.BeamSequence[num] = self.beams[num].export_logbeam(export_expected,leafgap)
exportplan.RTPlanLabel = ("dyn_"+plan_name)[:13]
ltime = time.localtime()
study_instance = exportplan.StudyInstanceUID.split(".")
series_instance = exportplan.SeriesInstanceUID.split(".")
instance_id = exportplan.SOPInstanceUID.split(".")
exportplan.StudyInstanceUID = ".".join(study_instance[:-1])+\
"."+"".join([str(t) for t in ltime[3:6]])
exportplan.SeriesInstanceUID = ".".join(series_instance[:-1])+\
"."+"".join([str(t) for t in ltime[:6]])
exportplan.StudyID = "Id"+\
"".join([str(t) for t in ltime[3:6]])
exportplan.SOPInstanceUID = ".".join(instance_id[:-1])+\
"."+"".join([str(t) for t in ltime[:6]])
exportplan.ApprovalStatus = "UNAPPROVED"
dcm.write_file(filename,exportplan)
def _write_file(cls, ds, filename):
""" Write out the Dicom files to an internal location.
Could be extended to use S3 or another file store later on.
"""
if not filename.endswith('.dcm'):
filename = '%s.dcm' % filename
# Generate a unique filename to avoid collisions.
retries = 0
raw_base_path = current_app.config['DICOM_ABS_UPLOAD_DIR']
while retries < cls.MAX_RETRIES:
prefix = str(uuid.uuid4().hex[0:6])
raw_filename = '%s_%s' % (prefix, filename)
raw_path = os.path.join(raw_base_path, raw_filename)
if not os.path.exists(raw_path):
break
retries += 1
if retries > cls.MAX_RETRIES:
# We didn't get a unique filename?!
raise Exception("Unable to generate a unique filename after "
"%d retries!" % cls.MAX_RETRIES)
# Write the raw file.
with open(raw_path, 'wb+') as f:
dicom.write_file(f, ds)
# Attempt to convert to an image and write out.
img_base_path = current_app.config['IMG_ABS_UPLOAD_DIR']
try:
img = dicom_lib.pil_from_dataset(ds)
except Exception:
# Just move on if we can't get an image.
return {'raw': raw_filename}
img_filename = raw_filename.rsplit('.', 1)[0] + '.jpg'
img_path = os.path.join(img_base_path, img_filename)
with open(img_path, 'wb+') as f:
img.save(f, format=cls.IMG_FORMAT)
# Also save a thumbnail.
img.thumbnail((64, 64))
thumb_filename = raw_filename.rsplit('.', 1)[0] + '.thumb.jpg'
thumb_path = os.path.join(img_base_path, thumb_filename)
with open(thumb_path, 'wb+') as f:
img.save(f, format=cls.IMG_FORMAT)
return {
'raw': raw_filename,
'img': img_filename,
'thumb': thumb_filename,
}
def prepare_data(dicom_dir, new_dicom_foldername, FS_folder, dict_new_metadata, ignore_FS_folder):
print new_dicom_foldername
try:
shutil.copytree(os.path.join(TEST_DATA, "dicoms", dicom_dir), os.path.join(TEST_DATA, "dicoms", new_dicom_foldername))
except OSError as e:
if(e.errno == 17):
print "New DICOM directory already exists, using this."
else:
print "Problem copying files into new directory: %s" % e
raise e
except Exception as e:
raise e
try:
files = os.listdir(os.path.join(TEST_DATA, "dicoms", new_dicom_foldername))
except Exception as e:
print "Problem getting files in new DICOM directory: %s" %e
raise e
for f in files:
d = dicom.read_file(os.path.join(TEST_DATA, "dicoms", new_dicom_foldername, f))
change_meta = {"PatientName" : d[0x10,0x10],
"AcquisitionDate" : d[0x08,0x22],
"PatientID" : d[0x10, 0x20],
"AccessionNumber": d[0x08, 0x50],
"SeriesDescription": d[0x08, 0x103e]
}
for key in dict_new_metadata.keys():
change_meta[key].value = dict_new_metadata[key]
dicom.write_file(os.path.join(TEST_DATA, "dicoms", new_dicom_foldername, f), d)
if(files and not ignore_FS_folder and
("PatientID" in dict_new_metadata.keys() or
"AccessionNumber" in dict_new_metadata.keys() or
"SeriesDescription" in dict_new_metadata.keys())
):
d = dicom.read_file(os.path.join(TEST_DATA, "dicoms", new_dicom_foldername, files[0]))
ptuniqid = create_internal_ptid_uniq(d.PatientID, d.AccessionNumber, d.SeriesDescription)
try:
shutil.copytree(os.path.join(TEST_DATA, "fs_res", FS_folder),
os.path.join(TEST_DATA, "fs_res", ptuniqid))
except OSError, e:
if(e.errno == 17): print "New FS folder already exists."
else: raise IOError("Problem copying content from original FS folder %s" %e)
def save_dicom_data(dicom_file_data,
file_path_to_save,
with_original_meta=True):
"""
Saves dicom data as a dicom file
:param dicom_file_data: Dicom data like pydicom.dataset.FileDataset
:param file_path_to_save: Full path to the file to save as a dicom file
:param with_original_meta: Whether to save new file with original metadata.
If original file does not have metadata writes None
"""
dicom.write_file(file_path_to_save,
dicom_file_data,
write_like_original=with_original_meta)
def write_ds(ds, fn, default_sopclass=None):
ds.file_meta = dicom.dataset.Dataset()
ds.file_meta.TransferSyntaxUID = dicom.UID.ImplicitVRLittleEndian
if default_sopclass == None:
default_sopclass = get_uid(
"Study Root Query/Retrieve Information Model - FIND")
ds.file_meta.MediaStorageSOPClassUID = getattr(ds, 'SOPClassUID',
default_sopclass)
ds.file_meta.MediaStorageSOPInstanceUID = getattr(ds, 'SOPInstanceUID',
generate_uid())
ds.is_little_endian = True
ds.is_implicit_VR = True
ds.file_meta.ImplementationClassUID = '2.25.4282708245307149051252828097685724107'
dicom.write_file(fn, ds, WriteLikeOriginal=False)
def C_STORE_RQ_received(self, presentation_context_id, store_rq, dimse_data):
log.msg("received DIMSE command %s" % store_rq)
assert store_rq.__class__ == dimsemessages.C_STORE_RQ
log.msg("replying to %s" % store_rq)
status = 0
try:
dimse_data.file_meta = dicom.dataset.Dataset()
dimse_data.file_meta.TransferSyntaxUID = dicom.UID.ImplicitVRLittleEndian
dimse_data.file_meta.MediaStorageSOPClassUID = dimse_data.SOPClassUID
dimse_data.file_meta.MediaStorageSOPInstanceUID = generate_uid()
dimse_data.is_little_endian = True
dimse_data.is_implicit_VR = True
dimse_data.file_meta.ImplementationClassUID = '2.25.4282708245307149051252828097685724107'
dicom.write_file("%s_%s.dcm" % (getattr(dimse_data, 'Modality', 'XX'), dimse_data.SOPInstanceUID), dimse_data, WriteLikeOriginal=False)
except Exception, e:
log.err(e)
status = 1
def over_write_file(self, outputDir):
print("over writing ROI file! ")
pix2pat = self.imageInfo['Pix2Pat']
ind2loc = self.imageInfo['Ind2Loc']
for ROI in self.ROI_List:
print('{}: {}'.format(ROI.Number, ROI.Name))
ROInames_in_file = []
for SSROISeq in self.di.StructureSetROISequence:
ROInames_in_file.append(SSROISeq.ROIName.lower())
fileROI_set = set(ROInames_in_file)
PatientROI_set = set(self.ROI_byName.keys())
patient_not_file = list(PatientROI_set.difference(fileROI_set))
print('ROIs in pat but not file', patient_not_file)
for patientName in patient_not_file:
thisROI = self.ROI_byName[patientName]
ROIObsSeq = mkNewROIObs_dataset(thisROI)
self.di.RTROIObservationsSequence.append(ROIObsSeq)
SSROI = mkNewStructureSetROI_dataset(thisROI, self.FrameRef_UID)
self.di.StructureSetROISequence.append(SSROI)
ROIContour = mkNewROIContour_dataset(thisROI)
self.di.ROIContourSequence.append(ROIContour)
for index, SS in enumerate(self.di.StructureSetROISequence):
thisROI = self.ROI_byName[SS.ROIName.lower()]
ContourSequence = mkNewContour_Sequence(thisROI, ind2loc, pix2pat)
self.di.ROIContourSequence[index].ContourSequence = ContourSequence
outFile = self.SSFile
outpath = os.path.join(outputDir, outFile)
print('saving to {}'.format(outpath))
dicom.write_file(outpath, self.di)
def swap(source, t1, t2, force=False, backup=True):
''' Swaps values from two tags in a DICOM file or a directory'''
if backup:
print 'Backup to ', backup_archive
os.system('tar cfz %s %s'%(backup_archive, source))
dcm = collect_dicom(source)
table = [['PatientID', 'first_tag', 'tag1_value', 'second_tag', 'tag2_value', 'filepath']]
print 'WARNING: will swap tags %s and %s in the following files:\n%s\n(%s files)'\
%(t1, t2, '\n'.join(dcm), len(dcm))
if (force or raw_input('Proceed? y/N ') == 'y'):
for i, each in enumerate(dcm):
progress = i/float(len(dcm)) * 100.0
sys.stdout.write("Operation progress: %d%% \r" % (progress) )
sys.stdout.flush()
try:
d = dicom.read_file(each)
first = getattr(d, t1)
second = getattr(d, t2)
pid = d.PatientID
table.append([pid, t1, first, t2, second, each])
setattr(d, t2, first)
setattr(d, t1, second)
dicom.write_file(each, d)
except (dicom.filereader.InvalidDicomError, IOError) as e:
print 'WARNING: file %s raised the following error:\n%s'%(each, e)
except KeyboardInterrupt:
print '<Keyboard Interrupt>'
df = pd.DataFrame(table[1:], columns=table[0])
return df
except Exception as e:
print e
df = pd.DataFrame(table[1:], columns=table[0])
return df
df = pd.DataFrame(table[1:], columns=table[0])
return df
def run(self):
"""
Reads the DICOM file, anonymizes it and write the result.
"""
try:
ds = dicom.read_file(self._dicom_filein)
try:
meta_data = dicom.filereader.read_file_meta_info(
self._dicom_filein)
if meta_data[0x0002, 0x0002].value == "Media Storage Directory Storage":
print "This file is a DICOMDIR:", self._dicom_filein
return
except:
pass
ds.walk(self.anonymize)
except:
print "This file is not a DICOM file:", self._dicom_filein
return
try:
dicom.write_file(self._dicom_fileout, ds)
except:
print "The anonymization fails on", self._dicom_filein
return
def remove(source, tag, value='', force=False, backup=False):
''' Clears the value of a given tag in a DICOM file or directory and replaces
it with a new value (default='')'''
if backup:
print 'Backup to ', backup_archive
os.system('tar cfz %s %s'%(backup_archive, source))
dcm = collect_dicom(source)
table = [['PatientID', 'removed_tag', 'tag_value', 'filepath']]
print 'WARNING: will remove tag %s from the following files:\n%s\n(%s files)'\
%(tag, '\n'.join(dcm), len(dcm))
if (force or raw_input('Proceed? y/N ') == 'y'):
for i, each in enumerate(dcm):
progress = i/float(len(dcm)) * 100.0
sys.stdout.write("Operation progress: %d%% \r" % (progress) )
sys.stdout.flush()
try:
d = dicom.read_file(each)
pid = d.PatientID
tagval = getattr(d, tag)
table.append([pid, tag, tagval, each])
setattr(d, tag, value)
dicom.write_file(each, d)
except (dicom.filereader.InvalidDicomError, IOError) as e:
print 'WARNING: file %s raised the following error:\n%s'%(each, e)
except KeyboardInterrupt:
print '<Keyboard Interrupt>'
df = pd.DataFrame(table[1:], columns=table[0])
return df
except Exception as e:
print e
df = pd.DataFrame(table[1:], columns=table[0])
return df
df = pd.DataFrame(table[1:], columns=table[0])
return df
def sabr_deid(subj_info, scan_df, raw_dir, deid_outdir):
#Join raw dir with subject name (assumes directory structure is ./rawdir/subj_name/...
subj_main_dir = os.path.join(raw_dir, subj_info['subj_name'])
new_id = subj_info['subj_id']
#Get list of sessions within main subj directory, make dir and loop over sessions.
subj_sessions = os.walk(subj_main_dir).next()[1]
subj_sessions.sort()
print('\n***{} has {} session(s)'.format(new_id, len(subj_sessions)))
#Create deidentified main (root) directory for subject
subj_deid_main_dir = os.path.join(deid_outdir, new_id)
try:
os.mkdir(subj_deid_main_dir)
except:
print('\nDirectory {} exists\n'.format(subj_deid_main_dir))
#WARNING! LAZY CODING AHEAD!
if len(subj_sessions) == 0:
raise Exception(
'\n***ERROR! NUMBER OF SESSIONS = 0!***\nPlease check directory structure of {}'
.format(subj_info['subj_name']))
elif len(subj_sessions) == 1:
session = subj_sessions[0]
subj_session_dir = os.path.join(subj_main_dir, session)
subj_deid_session_dir = os.path.join(subj_deid_main_dir, 'ses-01')
try:
os.mkdir(subj_deid_session_dir)
except:
print('\nSession folder {} exists\n'.format(subj_deid_session_dir))
for j, scan_type in enumerate(scan_df['scan_type']):
subj_deid_meta_dir = os.path.join(subj_deid_session_dir, scan_type)
try:
os.mkdir(subj_deid_meta_dir)
except:
print('Meta directory {} exists.'.format(scan_type))
#Match common sequence substring with path in os.walk
for root, dr, files in os.walk(subj_session_dir):
match = scan_df.scan_match[j]
match_regex = fnmatch.translate(match)
found = re.search(match_regex, root)
#print('\n***{}***\n'.format(found))
#If match, start deid process. If not, move onto next folder.
if found != None:
subj_deid_sequence_dir = os.path.join(
subj_deid_meta_dir, scan_df.scan_filename[j])
print('Making directory {}'.format(subj_deid_sequence_dir))
try:
os.mkdir(
subj_deid_sequence_dir
) #Make "housing" directory to keep dicoms of different sequences but same meta-category separate.
except:
print(
'\n***SEQUENCE DIRECTORY ALREADY EXISTS!***\nSkipping.'
)
continue
#Create list of dicoms in sequence dir rather than use
#files (more control in case any non-dicoms)
anon_files = os.listdir(root)
anon_files = [
x for x in anon_files if 'nii' not in x
] #Remove any previous nii files that may be present < To do - expand to other file types (mgh, analyze, etc)
anon_files.sort()
for anon_file in anon_files:
#Read files in 1 at a time, remove the remove / alter the below tags.
dcm = dicom.read_file(
os.path.join(root, anon_file), force=True
) #Uses force = True incase dicoms haven't had identifier added to header
#Strip aquisition date information
aqusition_date_list = [[0x0008, 0x0020],
[0x0008, 0x0021],
[0x0008, 0x0022],
[0x0008, 0x0023]]
for tag in aqusition_date_list:
try:
dcm[hex(tag[0]), hex(tag[1])].value = ''
except:
print(
'Tag {} {} does not exist in {}. Moving to next tag'
.format(hex(tag[0]), hex(tag[1]),
scan_df.scan_filename[j]))
#Strip aquisition time information
aqusition_time_list = [[0x0008, 0x0030],
[0x0008, 0x0031],
[0x0008, 0x0032],
[0x0008, 0x0033]]
for tag in aqusition_time_list:
try:
dcm[hex(tag[0]), hex(tag[1])].value = ''
except:
print(
'Tag {} {} does not exist in {}. Moving to next tag'
.format(hex(tag[0]), hex(tag[1]),
scan_df.scan_filename[j]))
#Strip physician information
physician_list = [[0x0008, 0x0090], [0x0008, 0x1050]]
for tag in physician_list:
try:
dcm[hex(tag[0]), hex(tag[1])].value = ''
except:
print(
'Tag {} {} does not exist in {}. Moving to next tag'
.format(hex(tag[0]), hex(tag[1]),
scan_df.scan_filename[j]))
#Strip study description
#dcm[0x0008,0x1030].value = ''
#Strip subject name / patient ID
subj_name_list = [[0x0010, 0x0010], [0x0010, 0x0020]]
#PatientName, PatientID
for tag in subj_name_list:
try:
dcm[hex(tag[0]), hex(tag[1])].value = new_id
except:
print(
'Tag {} {} does not exist in {}. Moving to next tag'
.format(hex(tag[0]), hex(tag[1]),
scan_df.scan_filename[j]))
#Strip subject attributes
subj_attrib_list = [[0x0010, 0x0030], [0x0010, 0x1010],
[0x0010, 0x1020], [0x0010, 0x1030]]
#, DoB, Age, PatientHeight, PatientWeight
for tag in subj_attrib_list:
try:
dcm[hex(tag[0]), hex(tag[1])].value = ''
except:
print(
'Tag {} {} does not exist in {}. Moving to next tag'
.format(hex(tag[0]), hex(tag[1]),
scan_df.scan_filename[j]))
#Write anonymised file
dicom.write_file(
os.path.join(subj_deid_sequence_dir, anon_file),
dcm)
elif len(subj_sessions) > 1:
for sn, session in enumerate(subj_sessions):
#MAKE DIRECTORIES BUT ZERO PAD SESSION
subj_deid_session_dir = os.path.join(
subj_deid_main_dir, 'ses-'
'{:02d}'.format(sn + 1))
try:
os.mkdir(subj_deid_session_dir)
except:
print('\nSession folder {} exists\n'.format(
subj_deid_session_dir))
#Session folder for identifiable subject
subj_session_dir = os.path.join(subj_main_dir, session)
#Loop over scan folder types within scan dataframe (anat, task, etc)
for j, scan_type in enumerate(scan_df['scan_type']):
subj_deid_meta_dir = os.path.join(subj_deid_session_dir,
scan_type)
try:
os.mkdir(subj_deid_meta_dir)
except:
print('Meta directory {} exists.'.format(scan_type))
#Match common sequence substring with path in os.walk
for root, dr, files in os.walk(subj_session_dir):
match = scan_df.scan_match[j]
match_regex = fnmatch.translate(match)
found = re.search(match_regex, root)
#If match, start deid process, not not, move onto next folder.
if found != None:
subj_deid_sequence_dir = os.path.join(
subj_deid_meta_dir, scan_df.scan_filename[j])
print('Making directory {}'.format(
subj_deid_sequence_dir))
try:
os.mkdir(
subj_deid_sequence_dir
) #Make "housing" directory to keep dicoms of different sequences but same meta-category separate.
except:
print(
'\n***SEQUENCE DIRECTORY ALREADY EXISTS!***\nSkipping.'
)
continue
#Create list of dicoms in sequence dir rather than use
#files (more control in case any non-dicoms)
anon_files = os.listdir(root)
anon_files = [
x for x in anon_files if 'nii' not in x
] #Remove any previous nii files that may be present < To do - expand to other file types (mgh, analyze, etc)
anon_files.sort()
for anon_file in anon_files:
#Read files in 1 at a time, remove the remove / alter the below tags.
dcm = dicom.read_file(
os.path.join(root, anon_file), force=True
) #Uses force = True incase dicoms haven't had identifier added to header
#Strip aquisition date information
aqusition_date_list = [[0x0008, 0x0020],
[0x0008, 0x0021],
[0x0008, 0x0022],
[0x0008, 0x0023]]
for tag in aqusition_date_list:
try:
dcm[hex(tag[0]), hex(tag[1])].value = ''
except:
print(
'Tag {} {} does not exist in {}. Moving to next tag'
.format(hex(tag[0]), hex(tag[1]),
scan_df.scan_filename[j]))
#Strip aquisition time information
aqusition_time_list = [[0x0008, 0x0030],
[0x0008, 0x0031],
[0x0008, 0x0032],
[0x0008, 0x0033]]
for tag in aqusition_time_list:
try:
dcm[hex(tag[0]), hex(tag[1])].value = ''
except:
print(
'Tag {} {} does not exist in {}. Moving to next tag'
.format(hex(tag[0]), hex(tag[1]),
scan_df.scan_filename[j]))
#Strip physician information
physician_list = [[0x0008, 0x0090],
[0x0008, 0x1050]]
for tag in physician_list:
try:
dcm[hex(tag[0]), hex(tag[1])].value = ''
except:
print(
'Tag {} {} does not exist in {}. Moving to next tag'
.format(hex(tag[0]), hex(tag[1]),
scan_df.scan_filename[j]))
#Strip study description
#dcm[0x0008,0x1030].value = ''
#Strip subject name / patient ID
subj_name_list = [[0x0010, 0x0010],
[0x0010, 0x0020]]
#PatientName, PatientID
for tag in subj_name_list:
try:
dcm[hex(tag[0]),
hex(tag[1])].value = new_id
except:
print(
'Tag {} {} does not exist in {}. Moving to next tag'
.format(hex(tag[0]), hex(tag[1]),
scan_df.scan_filename[j]))
#Strip subject attributes
subj_attrib_list = [[0x0010, 0x0030],
[0x0010, 0x1010],
[0x0010, 0x1020],
[0x0010, 0x1030]]
#, DoB, Age, PatientHeight, PatientWeight
for tag in subj_attrib_list:
try:
dcm[hex(tag[0]), hex(tag[1])].value = ''
except:
print(
'Tag {} {} does not exist in {}. Moving to next tag'
.format(hex(tag[0]), hex(tag[1]),
scan_df.scan_filename[j]))
#Write anonymised file
dicom.write_file(
os.path.join(subj_deid_sequence_dir,
anon_file), dcm)
return (subj_deid_main_dir, subj_sessions, new_id)
import dicom path = "/media/sf_hserver1_images/hugh/patient/" orig = dicom.read_file(path + "orig.ima") mcor = dicom.read_file(path + "mcor_hermes_axis.ima") nf = mcor.NumberOfFrames nf2 = nf ** 2 pixel_list = [mcor.PixelData[2 * a * nf2:2 * (a + 1) * nf2] for a in range(nf)] pixel_list = [pixel_list[p - nf / 2] for p in range(nf)] orig.PixelData = ''.join(pixel_list) orig.SeriesDescription = "DMSA SPECT [E1 Corrected]" orig.SeriesInstanceUID = dicom.UID.generate_uid() orig.SOPInstanceUID = dicom.UID.generate_uid() dicom.write_file(path + "fixed.ima", orig, WriteLikeOriginal=True)
def write_dicom(path, dataset):
"""write a pydicom dataset to dicom file"""
if not os.path.splitext(path)[1] == '.dcm':
path += '.dcm'
dicom.write_file(path, dataset)
def patchDicomDir(self,
inputDirPath,
outputDirPath,
exportDicom=True,
anonymizeDicom=False,
exportUltrasoundToNrrd=False):
"""
Since CTK (rightly) requires certain basic information [1] before it can import
data files that purport to be dicom, this code patches the files in a directory
with some needed fields. Apparently it is possible to export files from the
Philips PMS QLAB system with these fields missing.
Calling this function with a directory path will make a patched copy of each file.
Importing the old files to CTK should still fail, but the new ones should work.
The directory is assumed to have a set of instances that are all from the
same study of the same patient. Also that each instance (file) is an
independent (multiframe) series.
[1] https://github.com/commontk/CTK/blob/16aa09540dcb59c6eafde4d9a88dfee1f0948edc/Libs/DICOM/Core/ctkDICOMDatabase.cpp#L1283-L1287
"""
if slicer.app.majorVersion == 4 and slicer.app.minorVersion <= 10:
import dicom
else:
import pydicom as dicom
if not outputDirPath:
outputDirPath = inputDirPath
self.addLog('DICOM patching started...')
logging.debug('DICOM patch input directory: ' + inputDirPath)
logging.debug('DICOM patch output directory: ' + outputDirPath)
patientIDToRandomIDMap = {}
studyUIDToRandomUIDMap = {}
seriesUIDToRandomUIDMap = {}
numberOfSeriesInStudyMap = {}
# All files without a patient ID will be assigned to the same patient
randomPatientID = self.generateUid()
requiredTags = [
'PatientName', 'PatientID', 'StudyInstanceUID',
'SeriesInstanceUID', 'SeriesNumber'
]
for root, subFolders, files in os.walk(inputDirPath):
# Assume that all files in a directory belongs to the same study
randomStudyUID = self.generateUid()
currentSubDir = os.path.relpath(root, inputDirPath)
rootOutput = os.path.join(outputDirPath, currentSubDir)
for file in files:
filePath = os.path.join(root, file)
self.addLog('Examining %s...' %
os.path.join(currentSubDir, file))
try:
ds = dicom.read_file(filePath)
except (IOError, dicom.filereader.InvalidDicomError):
self.addLog(' Not DICOM file. Skipped.')
continue
if not hasattr(ds, 'SOPClassUID'):
self.addLog(' No SOPClassUID tag found. Skipped.')
continue
if ds.SOPClassUID != '1.2.840.113543.6.6.1.3.10002':
self.addLog(
' Not recognized as Philips Cartesian 4D ultrasound DICOM file. Skipped.'
)
continue
self.addLog(' Patching...')
for tag in requiredTags:
if not hasattr(ds, tag):
setattr(ds, tag, '')
# Generate a new SOPInstanceUID to avoid different files having the same SOPInstanceUID
ds.SOPInstanceUID = self.generateUid()
if ds.PatientName == '':
ds.PatientName = "Unspecified Patient"
if ds.PatientID == '':
ds.PatientID = randomPatientID
if ds.StudyInstanceUID == '':
ds.StudyInstanceUID = randomStudyUID
if ds.SeriesInstanceUID == '':
ds.SeriesInstanceUID = self.generateUid()
# Generate series number to make it easier to identify a sequence within a study
if ds.SeriesNumber == '':
if ds.StudyInstanceUID not in numberOfSeriesInStudyMap:
numberOfSeriesInStudyMap[ds.StudyInstanceUID] = 0
numberOfSeriesInStudyMap[
ds.StudyInstanceUID] = numberOfSeriesInStudyMap[
ds.StudyInstanceUID] + 1
ds.SeriesNumber = numberOfSeriesInStudyMap[
ds.StudyInstanceUID]
if anonymizeDicom:
self.addLog(' Anonymizing...')
ds.StudyDate = ''
ds.StudyTime = ''
ds.ContentDate = ''
ds.ContentTime = ''
ds.AccessionNumber = ''
ds.ReferringPhysiciansName = ''
ds.PatientsBirthDate = ''
ds.PatientsSex = ''
ds.StudyID = ''
ds[(
0x3001, 0x1004
)].value = '' # Some ID in a private tag - clear it, just in case
ds.PatientName = "Unspecified Patient"
# replace ids with random values - re-use if we have seen them before
if ds.PatientID not in patientIDToRandomIDMap:
patientIDToRandomIDMap[
ds.PatientID] = self.generateUid()
ds.PatientID = patientIDToRandomIDMap[ds.PatientID]
if ds.StudyInstanceUID not in studyUIDToRandomUIDMap:
studyUIDToRandomUIDMap[
ds.StudyInstanceUID] = self.generateUid()
ds.StudyInstanceUID = studyUIDToRandomUIDMap[
ds.StudyInstanceUID]
if ds.SeriesInstanceUID not in studyUIDToRandomUIDMap:
seriesUIDToRandomUIDMap[
ds.SeriesInstanceUID] = self.generateUid()
ds.SeriesInstanceUID = seriesUIDToRandomUIDMap[
ds.SeriesInstanceUID]
if inputDirPath == outputDirPath:
(name, ext) = os.path.splitext(filePath)
patchedFilePath = name + ('-anon' if anonymizeDicom else
'') + '-patched' + ext
nrrdFilePath = name + '.seq.nrrd'
else:
patchedFilePath = os.path.abspath(
os.path.join(rootOutput, file))
nrrdFilePath = os.path.splitext(
patchedFilePath)[0] + '.seq.nrrd'
if not os.path.exists(rootOutput):
os.makedirs(rootOutput)
self.addLog(' Writing DICOM...')
dicom.write_file(patchedFilePath, ds)
self.addLog(' Created DICOM file: %s' % patchedFilePath)
if exportUltrasoundToNrrd and self.isDicomUltrasoundFile(
patchedFilePath):
self.addLog(' Writing NRRD...')
if self.convertUltrasoundDicomToNrrd(
patchedFilePath, nrrdFilePath):
self.addLog(' Created NRRD file: %s' % nrrdFilePath)
else:
self.addLog(' NRRD file save failed')
if not exportDicom:
os.remove(patchedFilePath)
self.addLog(' Deleted temporary DICOM file')
self.addLog('DICOM patching completed.')
def save_scan(path, dataset):
for i in range(0, len(dataset) - 1):
dicom.write_file(path + "/slice" + i, dataset[i])
return 0
pix[j][i] = 0
if (output_dcm==1):
bin_value = int(round(pix[j][i]*100)) # Convert cm floats into tenths of mm integers to store in dicom image
for k1 in range(0, output_binning):
for k2 in range(0, output_binning):
mammo.pixel_array[j+k1][i+k2] = bin_value
sys.stdout.write('\n'); sys.stdout.flush()
if (output_dcm==1):
# Save final data as dicom: help(dicom.write_file)
print '\n\n Saving final data as DICOM image in integer units of tenths of mm of plastic...'
mammo.PixelData = mammo.pixel_array.tostring() # Save the data array into the actual pixel data string
dicom.write_file(image_file_name+"_bin"+str(output_binning)+".dcm", mammo)
# Generate triangle mesh:
print '\n Generating the output triangle mesh focused to the x-ray source focal spot (num_air_pix='+str(num_air_pix)+')...'
num_binned_pixels = (num_columns/output_binning) * (num_rows/output_binning) - num_air_pix
num_vertices = num_binned_pixels * 8 # 8 vertices per printed column
if (output_base==1):
num_triangles = num_binned_pixels * 8 + 2*(num_rows/output_binning) + 2*(num_columns/output_binning) #!!ReducedTriangles!!
else:
num_triangles = num_binned_pixels * 6 + 2*(num_rows/output_binning) + 2*(num_columns/output_binning)
image_file_name2 = image_file_name+"_RawData_bin"+str(output_binning)+"_focusedZ"+str(source_coord[2])
if (subtract_layer>0.00001):
import dicom, os, shutil, sys, string
from dicom.tag import Tag
# input: directory with the DICOM files
inputDir = sys.argv[1]
project = sys.argv[2]
subject = sys.argv[3]
outputDir = sys.argv[4]
for r,d,files in os.walk(inputDir):
for f in files:
print 'Processing ', f
inputFile = r+'/'+f
dcm = dicom.read_file(inputFile)
outputFile = outputDir+'/'+f
pcTag = Tag(0x10,0x4000)
time = string.split(dcm.StudyTime,'.')[0]
dcm.add_new(pcTag,'LT',"Project: "+project+"; Subject: "+subject+"; Session: "+subject+"_"+dcm.StudyDate+"_"+time+';')
print dcm.PatientComments
f = open(outputFile,'wb')
dicom.write_file(outputFile,dcm)
# shutil.copyfile(inputFile, outputFile)
def patchDicomDir(self,
inputDirPath,
outputDirPath,
generateMissingIds=True,
generateImagePositionFromSliceThickness=True,
anonymize=False):
"""
Since CTK (rightly) requires certain basic information [1] before it can import
data files that purport to be dicom, this code patches the files in a directory
with some needed fields. Apparently it is possible to export files from the
Philips PMS QLAB system with these fields missing.
Calling this function with a directory path will make a patched copy of each file.
Importing the old files to CTK should still fail, but the new ones should work.
The directory is assumed to have a set of instances that are all from the
same study of the same patient. Also that each instance (file) is an
independent (multiframe) series.
[1] https://github.com/commontk/CTK/blob/16aa09540dcb59c6eafde4d9a88dfee1f0948edc/Libs/DICOM/Core/ctkDICOMDatabase.cpp#L1283-L1287
"""
import dicom
if not outputDirPath:
outputDirPath = inputDirPath
self.addLog('DICOM patching started...')
logging.debug('DICOM patch input directory: ' + inputDirPath)
logging.debug('DICOM patch output directory: ' + outputDirPath)
patientIDToRandomIDMap = {}
studyUIDToRandomUIDMap = {}
seriesUIDToRandomUIDMap = {}
numberOfSeriesInStudyMap = {}
# All files without a patient ID will be assigned to the same patient
randomPatientID = dicom.UID.generate_uid(None)
requiredTags = [
'PatientName', 'PatientID', 'StudyInstanceUID',
'SeriesInstanceUID', 'SeriesNumber'
]
for root, subFolders, files in os.walk(inputDirPath):
# Assume that all files in a directory belongs to the same study
randomStudyUID = dicom.UID.generate_uid(None)
# Assume that all files in a directory belongs to the same series
randomSeriesInstanceUID = dicom.UID.generate_uid(None)
currentSubDir = os.path.relpath(root, inputDirPath)
rootOutput = os.path.join(outputDirPath, currentSubDir)
for file in files:
filePath = os.path.join(root, file)
self.addLog('Examining %s...' %
os.path.join(currentSubDir, file))
if file == 'DICOMDIR':
self.addLog(
'DICOMDIR file is ignored (its contents may be inconsistent with the contents of the indexed DICOM files, therefore it is safer not to use it)'
)
continue
try:
ds = dicom.read_file(filePath)
except (IOError, dicom.filereader.InvalidDicomError):
self.addLog(' Not DICOM file. Skipped.')
continue
self.addLog(' Patching...')
######################################################
# Add missing IDs
if generateMissingIds:
for tag in requiredTags:
if not hasattr(ds, tag):
setattr(ds, tag, '')
# Generate a new SOPInstanceUID to avoid different files having the same SOPInstanceUID
ds.SOPInstanceUID = dicom.UID.generate_uid(None)
if ds.PatientName == '':
ds.PatientName = "Unspecified Patient"
if ds.PatientID == '':
ds.PatientID = randomPatientID
if ds.StudyInstanceUID == '':
ds.StudyInstanceUID = randomStudyUID
if ds.SeriesInstanceUID == '':
#ds.SeriesInstanceUID = dicom.UID.generate_uid(None) # this line has to be used if each file is a separate series
ds.SeriesInstanceUID = randomSeriesInstanceUID
# Generate series number to make it easier to identify a sequence within a study
if ds.SeriesNumber == '':
if ds.StudyInstanceUID not in numberOfSeriesInStudyMap:
numberOfSeriesInStudyMap[ds.StudyInstanceUID] = 0
numberOfSeriesInStudyMap[
ds.StudyInstanceUID] = numberOfSeriesInStudyMap[
ds.StudyInstanceUID] + 1
ds.SeriesNumber = numberOfSeriesInStudyMap[
ds.StudyInstanceUID]
######################################################
# Add missing slice spacing info to multiframe files
numberOfFrames = ds.NumberOfFrames if hasattr(
ds, 'NumberOfFrames') else 1
if generateImagePositionFromSliceThickness and numberOfFrames > 1:
# Multi-frame sequence, we may need to add slice positions
# Error in Dolphin 3D CBCT scanners, they store multiple frames but they keep using CTImageStorage as storage class
if ds.SOPClassUID == '1.2.840.10008.5.1.4.1.1.2': # Computed Tomography Image IOD
ds.SOPClassUID = '1.2.840.10008.5.1.4.1.1.2.1' # Enhanced CT Image IOD
sliceStartPosition = ds.ImagePositionPatient if hasattr(
ds, 'ImagePositionPatient') else [0, 0, 0]
sliceAxes = ds.ImageOrientationPatient if hasattr(
ds, 'ImagePositionPatient') else [1, 0, 0, 0, 1, 0]
x = sliceAxes[:3]
y = sliceAxes[3:]
z = [
x[1] * y[2] - x[2] * y[1], x[2] * y[0] - x[0] * y[2],
x[0] * y[1] - x[1] * y[0]
] # cross(x,y)
sliceSpacing = ds.SliceThickness if hasattr(
ds, 'SliceThickness') else 1.0
pixelSpacing = ds.PixelSpacing if hasattr(
ds, 'PixelSpacing') else [1.0, 1.0]
if not (dicom.tag.Tag(0x5200, 0x9229) in ds):
# (5200,9229) SQ (Sequence with undefined length #=1) # u/l, 1 SharedFunctionalGroupsSequence
# (0020,9116) SQ (Sequence with undefined length #=1) # u/l, 1 PlaneOrientationSequence
# (0020,0037) DS [1.00000\0.00000\0.00000\0.00000\1.00000\0.00000] # 48, 6 ImageOrientationPatient
# (0028,9110) SQ (Sequence with undefined length #=1) # u/l, 1 PixelMeasuresSequence
# (0018,0050) DS [3.00000] # 8, 1 SliceThickness
# (0028,0030) DS [0.597656\0.597656] # 18, 2 PixelSpacing
planeOrientationDataSet = dicom.dataset.Dataset()
planeOrientationDataSet.ImageOrientationPatient = sliceAxes
planeOrientationSequence = dicom.sequence.Sequence()
planeOrientationSequence.insert(
dicom.tag.Tag(0x0020, 0x9116),
planeOrientationDataSet)
pixelMeasuresDataSet = dicom.dataset.Dataset()
pixelMeasuresDataSet.SliceThickness = sliceSpacing
pixelMeasuresDataSet.PixelSpacing = pixelSpacing
pixelMeasuresSequence = dicom.sequence.Sequence()
pixelMeasuresSequence.insert(
dicom.tag.Tag(0x0028, 0x9110),
pixelMeasuresDataSet)
sharedFunctionalGroupsDataSet = dicom.dataset.Dataset()
sharedFunctionalGroupsDataSet.PlaneOrientationSequence = planeOrientationSequence
sharedFunctionalGroupsDataSet.PixelMeasuresSequence = pixelMeasuresSequence
sharedFunctionalGroupsSequence = dicom.sequence.Sequence(
)
sharedFunctionalGroupsSequence.insert(
dicom.tag.Tag(0x5200, 0x9229),
sharedFunctionalGroupsDataSet)
ds.SharedFunctionalGroupsSequence = sharedFunctionalGroupsSequence
if not (dicom.tag.Tag(0x5200, 0x9230) in ds):
#(5200,9230) SQ (Sequence with undefined length #=54) # u/l, 1 PerFrameFunctionalGroupsSequence
# (0020,9113) SQ (Sequence with undefined length #=1) # u/l, 1 PlanePositionSequence
# (0020,0032) DS [-94.7012\-312.701\-806.500] # 26, 3 ImagePositionPatient
# (0020,9113) SQ (Sequence with undefined length #=1) # u/l, 1 PlanePositionSequence
# (0020,0032) DS [-94.7012\-312.701\-809.500] # 26, 3 ImagePositionPatient
# ...
perFrameFunctionalGroupsSequence = dicom.sequence.Sequence(
)
for frameIndex in range(numberOfFrames):
planePositionDataSet = dicom.dataset.Dataset()
slicePosition = [
sliceStartPosition[0] +
frameIndex * z[0] * sliceSpacing,
sliceStartPosition[1] +
frameIndex * z[1] * sliceSpacing,
sliceStartPosition[2] +
frameIndex * z[2] * sliceSpacing
]
planePositionDataSet.ImagePositionPatient = slicePosition
planePositionSequence = dicom.sequence.Sequence()
planePositionSequence.insert(
dicom.tag.Tag(0x0020, 0x9113),
planePositionDataSet)
perFrameFunctionalGroupsDataSet = dicom.dataset.Dataset(
)
perFrameFunctionalGroupsDataSet.PlanePositionSequence = planePositionSequence
perFrameFunctionalGroupsSequence.insert(
dicom.tag.Tag(0x5200, 0x9230),
perFrameFunctionalGroupsDataSet)
ds.PerFrameFunctionalGroupsSequence = perFrameFunctionalGroupsSequence
######################################################
# Anonymize
if anonymize:
self.addLog(' Anonymizing...')
ds.StudyDate = ''
ds.StudyTime = ''
ds.ContentDate = ''
ds.ContentTime = ''
ds.AccessionNumber = ''
ds.ReferringPhysiciansName = ''
ds.PatientsBirthDate = ''
ds.PatientsSex = ''
ds.StudyID = ''
ds.PatientName = "Unspecified Patient"
# replace ids with random values - re-use if we have seen them before
if ds.PatientID not in patientIDToRandomIDMap:
patientIDToRandomIDMap[
ds.PatientID] = dicom.UID.generate_uid(None)
ds.PatientID = patientIDToRandomIDMap[ds.PatientID]
if ds.StudyInstanceUID not in studyUIDToRandomUIDMap:
studyUIDToRandomUIDMap[
ds.StudyInstanceUID] = dicom.UID.generate_uid(None)
ds.StudyInstanceUID = studyUIDToRandomUIDMap[
ds.StudyInstanceUID]
if ds.SeriesInstanceUID not in studyUIDToRandomUIDMap:
seriesUIDToRandomUIDMap[
ds.SeriesInstanceUID] = dicom.UID.generate_uid(
None)
ds.SeriesInstanceUID = seriesUIDToRandomUIDMap[
ds.SeriesInstanceUID]
######################################################
# Write
if inputDirPath == outputDirPath:
(name, ext) = os.path.splitext(filePath)
patchedFilePath = name + ('-anon' if anonymize else
'') + '-patched' + ext
else:
patchedFilePath = os.path.abspath(
os.path.join(rootOutput, file))
if not os.path.exists(rootOutput):
os.makedirs(rootOutput)
self.addLog(' Writing DICOM...')
dicom.write_file(patchedFilePath, ds)
self.addLog(' Created DICOM file: %s' % patchedFilePath)
self.addLog('DICOM patching completed.')
def export(self, filename):
dcm.write_file(filename, self.plan)
# di = np.array(0.01*(d/d_cal5cm)*MU*nFrac/dosescaling,np.int) di = np.array((d / d_ch) * nFrac * MU * D_ch_MU / dosescaling, np.int) dcmdose.PixelData = struct.pack(format, *(di.tolist())) print 'check dose volume sizes' print 'Pixel Range:', x[0], x[-1], y[0], y[-1], z[0], z[-1] print 'dx, dy, dz:', dx, dy, dz print 'ImageOrigien:', dcmdose.ImagePositionPatient print 'sizes:', xsize, ysize, zsize, np.size(x), np.size(y), np.size(z) print 'Max D/Ne:', d.max( ), 'Abs Dmax:', d.max() * MU * nFrac * D_ch_MU / d_ch, '<--' print 'PixelSpacing:', dcmdose.PixelSpacing, 'MaxDose:', di.max() print length, np.size(d), np.size(e) print 'New Image Position:', dcmdose.ImagePositionPatient print 'OffsetVector Length:', np.size(dcmdose.GridFrameOffsetVector) print dcmdose.SOPInstanceUID = get_dicomuid() dcmdose.StudyUID = get_dicomuid() dcmdose.SeriesUID = get_dicomuid() dcmdose.FrameUID = get_dicomuid() dcmdose.SyncUID = get_dicomuid() dcmdose.SrUID = get_dicomuid() dcmdose.StudyInstanceUID = get_dicomuid() dcmdose.SeriesInstanceUID = get_dicomuid() dicom.write_file(outfile, dcmdose) print 'New DCM RD file created: ', outfile #print xsize, ysize, zsize, length, length*dcmdose.BitsAllocated/8, max(di)
def export(self,filename):
dcm.write_file(filename,self.plan)
# di = np.array(d*pdmax/d.max(),np.int)
# di = np.array(0.01*(d/d_cal5cm)*MU*nFrac/dosescaling,np.int)
di = np.array((d/d_ch)*nFrac*MU*D_ch_MU/dosescaling,np.int)
dcmdose.PixelData = struct.pack(format,*(di.tolist()))
print 'check dose volume sizes'
print 'Pixel Range:', x[0],x[-1],y[0],y[-1],z[0],z[-1]
print 'dx, dy, dz:', dx, dy, dz
print 'ImageOrigien:',dcmdose.ImagePositionPatient
print 'sizes:',xsize, ysize, zsize, np.size(x),np.size(y),np.size(z)
print 'Max D/Ne:',d.max(), 'Abs Dmax:',d.max()*MU*nFrac*D_ch_MU/d_ch, '<--'
print 'PixelSpacing:',dcmdose.PixelSpacing, 'MaxDose:',di.max()
print length, np.size(d), np.size(e)
print 'New Image Position:', dcmdose.ImagePositionPatient
print 'OffsetVector Length:', np.size(dcmdose.GridFrameOffsetVector)
print
dcmdose.SOPInstanceUID = get_dicomuid()
dcmdose.StudyUID = get_dicomuid()
dcmdose.SeriesUID = get_dicomuid()
dcmdose.FrameUID = get_dicomuid()
dcmdose.SyncUID = get_dicomuid()
dcmdose.SrUID = get_dicomuid()
dcmdose.StudyInstanceUID = get_dicomuid()
dcmdose.SeriesInstanceUID = get_dicomuid()
dicom.write_file(outfile, dcmdose)
print 'New DCM RD file created: ',outfile
#print xsize, ysize, zsize, length, length*dcmdose.BitsAllocated/8, max(di)
parser.add_option('--in_dir',\
help='Input dicom directory. All dicom images inside \
this directory will be anonymized. It is assumed that \
all files inside this directory are dicom files.',\
dest='in_dir', metavar='<string>', default=None)
parser.add_option('--out_dir',\
help='Output dicom directory. If none specified, the input \
directory be used as the output directory, overwriting \
the dicom images that were read from the directory.',\
dest='out_dir', metavar='<string>', default=None)
(options, args) = parser.parse_args()
if options.in_im is not None:
assert isfile(options.in_im), 'File does not exist'
ds = dicom.read_file(options.in_im)
anonymize_dicom(ds)
if options.out_im is not None:
dicom.write_file(options.out_im, ds)
if options.in_dir is not None:
assert isdir(options.in_dir), 'Directory does not exist'
files = [f for f in listdir(options.in_dir) if \
isfile(join(options.in_dir, f))]
for f in files:
ds = dicom.read_file(join(options.in_dir, f))
anonymize_dicom(ds)
if options.out_dir is not None:
assert isdir(options.out_dir), 'Directory does not exist'
dicom.write_file(join(options.out_dir, f), ds)
def encrypt_dicom_name(dcm):
# ( dicomname, (bool_encrypt, bool_digitcheck))
dcmname = dcm[0]
opts = dcm[1]
bool_encrypt = opts[0]
bool_digitcheck = opts[1]
bool_write = False
try:
dcminf = dicom.read_file(dcmname)
except:
# If not readable, simply exit
return
# If it is a dicom, scramble all information.
for field in fields_to_anon:
# Check to make sure dicom field exists.
if hasattr(dcminf,field):
name = getattr(dcminf,field)
# If it's not a string, skip any anonymization for the field
if not isinstance(name, basestring):
continue
else:
name = name.encode('ascii','ignore')
if bool_encrypt:
# encrypt the dicomfield
if bool_digitcheck:
bool_hasdigits = re.findall('\d+',name)
else:
bool_hasdigits = False
# Ignore if the name has the words "anonymous" or "volunteer" in it
if ( name.lower().find("anonymous") >= 0 ) | ( name.lower().find("volunteer") >= 0 ):
continue
# The additional "_JNO" ending is a safety to prevent items
# from being re-encrypted. It is assumed that if the name
# has any numbers, the patient field has already been anonymized.
if (name[-4:] != "_JNO") & (not bool_hasdigits):
anon_name = encrypt_string(name,KEY) + "_JNO"
setattr(dcminf,field,anon_name)
bool_write = True
else:
# If the name has "_JNO" as the ending, it has been encrypted
# and needs to be unencrypted.
if name[-4:] == "_JNO":
anon_name = unencrypt_string(name[:-4],KEY)
setattr(dcminf,field,anon_name)
bool_write = True
if bool_write:
dicom.write_file(dcmname,dcminf)
else:
pix[j][i] = 0
if (output_dcm==1):
bin_value = int(round(pix[j][i]*100)) # Convert cm floats into tenths of mm integers to store in dicom image
for k1 in range(0, output_binning):
for k2 in range(0, output_binning):
mammo.pixel_array[j+k1][i+k2] = bin_value
if (output_dcm==1):
# Save final data as dicom: help(dicom.write_file)
print '\n\n Saving final data as DICOM image in integer units of tenths of mm of plastic...'
mammo.PixelData = mammo.pixel_array.tostring() # Save the data array into the actual pixel data string
#dicom.write_file(image_file_name+"_plastic_bin"+str(output_binning)+".dcm", mammo)
dicom.write_file(image_file_name+"_STL_bin"+str(output_binning)+".dcm", mammo) # !!STL!!
# Generate triangle mesh:
print '\n Generating the output triangle mesh focused to the x-ray source focal spot (num_air_pix='+str(num_air_pix)+')...'
num_binned_pixels = (num_columns/output_binning) * (num_rows/output_binning) - num_air_pix
num_vertices = num_binned_pixels * 8 # 8 vertices per printed column
num_triangles = num_binned_pixels * 8 + 2*(num_rows/output_binning) + 2*(num_columns/output_binning) #!!ReducedTriangles!!
print ' Writing ',num_triangles,' triangles in output file: '+image_file_name+"_bin"+str(output_binning)+'.ply'
image_file_name2 = image_file_name+"_STL_bin"+str(output_binning)+"_focusedZ"+str(source_coord[2]) # !!STL!!
if (subtract_layer>0.00001):
image_file_name2 = image_file_name2+"_-"+str(subtract_layer)+"cm"
if (output_mm==1):
def BrowseFile(self, fname):
"""Allows the user to select a filename"""
self._savedFileName = fname
self._reader._reader = None
full_name = os.path.abspath(fname)
_dir, fname = os.path.split(full_name)
self._num_list = []
self._file_list = [full_name]
self._slice_locations = []
# is file a DICOM file?
ds = {}
examiner = MicroViewIO.wxMicroViewDICOMExaminer(_dir)
self.bIsDICOM = examiner.IsDICOMFile(full_name)
self.match_tags = {}
if self.bIsDICOM:
try:
ds = dicom.read_file(full_name,
stop_before_pixels=True,
force=True)
if 'PhotometricInterpretation' not in ds:
self.bIsDICOM = False
if 'AcquisitionNumber' in ds:
self.bHasAcquisitionNumber = True
tags = ['StudyID', 'SeriesNumber', 'AcquisitionNumber']
if ds.Modality == 'MR':
tags.append('EchoNumbers')
for tag in tags:
if tag in ds:
self.match_tags[tag] = ds.get(tag)
except:
pass
with wx.BusyCursor():
self.update_file_list(full_name, ds, examiner)
# update DICOMDIR file
if self.bIsDICOM:
dicomdir_filename = os.path.join(_dir, 'DICOMDIR')
dicomdir = examiner.GetDICOMDIR().get_dicomdir()
if dicomdir:
try:
dicom.write_file(dicomdir_filename, dicomdir)
logging.info(
"Successfully wrote {}.".format(dicomdir_filename))
except:
pass
self.guess_image_pixel_size(full_name)
self.updateGUIState()
self.updateButtonState()
# if files are DICOM-format, fill in a reasonable guess at a name
if self.bIsDICOM:
name = ''
if 'StudyID' in ds:
name += '{}_'.format(ds.StudyID)
if 'SeriesDescription' in ds:
name += '{}_'.format(ds.SeriesDescription.strip()).replace(
' ', '_')
if 'PatientName' in ds:
name += '{}'.format(ds.PatientName.strip()).replace(' ', '_')
if not name:
name = "DICOM Import"
self.m_textCtrlImageTitle.SetValue(name)
import dicom, os, shutil, sys, string from dicom.tag import Tag # for each DICOM file in inputDir, set the content of the tagName # to be the same as in the first file in that dir if len(sys.argv)<4: print 'Usage: ',sys.argv[0],' inputDir tagName outputDir' exit inputDir = sys.argv[1] inputTag = sys.argv[2] outputDir = sys.argv[3] files = os.listdir(inputDir) f = files[0] dcm = dicom.read_file(os.path.join(inputDir,f)) dicom.write_file(os.path.join(outputDir,f),dcm) firstVal = dcm.data_element(inputTag).value for f in files[1:]: dcm = dicom.read_file(os.path.join(inputDir,f)) dcm.data_element(inputTag).value = firstVal dicom.write_file(os.path.join(outputDir,f),dcm)
def patchDicomDir(self, inputDirPath, outputDirPath, exportDicom = True, anonymizeDicom = False, exportUltrasoundToNrrd = False):
"""
Since CTK (rightly) requires certain basic information [1] before it can import
data files that purport to be dicom, this code patches the files in a directory
with some needed fields. Apparently it is possible to export files from the
Philips PMS QLAB system with these fields missing.
Calling this function with a directory path will make a patched copy of each file.
Importing the old files to CTK should still fail, but the new ones should work.
The directory is assumed to have a set of instances that are all from the
same study of the same patient. Also that each instance (file) is an
independent (multiframe) series.
[1] https://github.com/commontk/CTK/blob/16aa09540dcb59c6eafde4d9a88dfee1f0948edc/Libs/DICOM/Core/ctkDICOMDatabase.cpp#L1283-L1287
"""
import dicom
if not outputDirPath:
outputDirPath = inputDirPath
self.addLog('DICOM patching started...')
logging.debug('DICOM patch input directory: '+inputDirPath)
logging.debug('DICOM patch output directory: '+outputDirPath)
patientIDToRandomIDMap = {}
studyUIDToRandomUIDMap = {}
seriesUIDToRandomUIDMap = {}
numberOfSeriesInStudyMap = {}
# All files without a patient ID will be assigned to the same patient
randomPatientID = dicom.UID.generate_uid(None)
requiredTags = ['PatientName', 'PatientID', 'StudyInstanceUID', 'SeriesInstanceUID', 'SeriesNumber']
for root, subFolders, files in os.walk(inputDirPath):
# Assume that all files in a directory belongs to the same study
randomStudyUID = dicom.UID.generate_uid(None)
currentSubDir = os.path.relpath(root, inputDirPath)
rootOutput = os.path.join(outputDirPath, currentSubDir)
for file in files:
filePath = os.path.join(root,file)
self.addLog('Examining %s...' % os.path.join(currentSubDir,file))
try:
ds = dicom.read_file(filePath)
except (IOError, dicom.filereader.InvalidDicomError):
self.addLog(' Not DICOM file. Skipped.')
continue
if ds.SOPClassUID != '1.2.840.113543.6.6.1.3.10002':
self.addLog(' Not Philips 4D ultrasound DICOM file. Skipped.')
continue
self.addLog(' Patching...')
for tag in requiredTags:
if not hasattr(ds,tag):
setattr(ds,tag,'')
# Generate a new SOPInstanceUID to avoid different files having the same SOPInstanceUID
ds.SOPInstanceUID = dicom.UID.generate_uid(None)
if ds.PatientName == '':
ds.PatientName = "Unspecified Patient"
if ds.PatientID == '':
ds.PatientID = randomPatientID
if ds.StudyInstanceUID == '':
ds.StudyInstanceUID = randomStudyUID
if ds.SeriesInstanceUID == '':
ds.SeriesInstanceUID = dicom.UID.generate_uid(None)
# Generate series number to make it easier to identify a sequence within a study
if ds.SeriesNumber == '':
if ds.StudyInstanceUID not in numberOfSeriesInStudyMap:
numberOfSeriesInStudyMap[ds.StudyInstanceUID] = 0
numberOfSeriesInStudyMap[ds.StudyInstanceUID] = numberOfSeriesInStudyMap[ds.StudyInstanceUID] + 1
ds.SeriesNumber = numberOfSeriesInStudyMap[ds.StudyInstanceUID]
if anonymizeDicom:
self.addLog(' Anonymizing...')
ds.StudyDate = ''
ds.StudyTime = ''
ds.ContentDate = ''
ds.ContentTime = ''
ds.AccessionNumber = ''
ds.ReferringPhysiciansName = ''
ds.PatientsBirthDate = ''
ds.PatientsSex = ''
ds.StudyID = ''
ds[(0x3001,0x1004)].value = '' # Some ID in a private tag - clear it, just in case
ds.PatientName = "Unspecified Patient"
# replace ids with random values - re-use if we have seen them before
if ds.PatientID not in patientIDToRandomIDMap:
patientIDToRandomIDMap[ds.PatientID] = dicom.UID.generate_uid(None)
ds.PatientID = patientIDToRandomIDMap[ds.PatientID]
if ds.StudyInstanceUID not in studyUIDToRandomUIDMap:
studyUIDToRandomUIDMap[ds.StudyInstanceUID] = dicom.UID.generate_uid(None)
ds.StudyInstanceUID = studyUIDToRandomUIDMap[ds.StudyInstanceUID]
if ds.SeriesInstanceUID not in studyUIDToRandomUIDMap:
seriesUIDToRandomUIDMap[ds.SeriesInstanceUID] = dicom.UID.generate_uid(None)
ds.SeriesInstanceUID = seriesUIDToRandomUIDMap[ds.SeriesInstanceUID]
if inputDirPath==outputDirPath:
(name, ext) = os.path.splitext(filePath)
patchedFilePath = name + ('-anon' if anonymizeDicom else '') + '-patched' + ext
nrrdFilePath = name + '.seq.nrrd'
else:
patchedFilePath = os.path.abspath(os.path.join(rootOutput,file))
nrrdFilePath = os.path.splitext(patchedFilePath)[0]+'.nrrd'
if not os.path.exists(rootOutput):
os.makedirs(rootOutput)
self.addLog(' Writing DICOM...')
dicom.write_file(patchedFilePath, ds)
self.addLog(' Created DICOM file: %s' % patchedFilePath)
if exportUltrasoundToNrrd and self.isDicomUltrasoundFile(patchedFilePath):
self.addLog(' Writing NRRD...')
if self.convertUltrasoundDicomToNrrd(patchedFilePath, nrrdFilePath):
self.addLog(' Created NRRD file: %s' % nrrdFilePath)
else:
self.addLog(' NRRD file save failed')
if not exportDicom:
os.remove(patchedFilePath)
self.addLog(' Deleted temporary DICOM file')
self.addLog('DICOM patching completed.')
def anonymize(dir_input, dir_out_base, siteCodeDict, lut_dcm_type, lut_dcm_hdr,
lut_rda_type, lut_twix_type):
# anonymize dicom
if re.match('[0-9]{3}-*', dir_input.split('/')[-1]):
[subjectID, sessionID] = get_patientID(dir_input, siteCodeDict)
dir_input_clean = dir_input.replace('-', '_')
series_num = int(
dir_input_clean.split('/')[-1].split('_')[0].lstrip("0"))
new_series_description = get_scan_type(dir_input_clean, lut_dcm_type)
# series number is stripped of leading zeroes
# pad with zeroes here so directories are listed in proper order
dir_out_full = '%s/%s/%s/%03d-%s/DICOM' % (dir_out_base, subjectID,
sessionID, series_num,
new_series_description)
if new_series_description is None:
print('ERROR - Unable to determine scan type for: %s' % dir_input)
else:
shutil.copytree(dir_input, dir_out_full)
os.system('chmod +w -R %s' % (dir_out_full))
list_dcm_files = os.listdir(dir_input)
for fname_scan in list_dcm_files:
ds = pydicom.read_file('%s/%s' % (dir_out_full, fname_scan))
ds.PatientName = subjectID
ds.PatientID = sessionID
ds.SeriesDescription = new_series_description
ds.SeriesNumber = series_num
# instead of this, do for loop (eg. for each key in lut_dcm_hdr, do ...)
for key in lut_dcm_hdr:
setattr(ds, key, lut_dcm_hdr[key])
pydicom.write_file('%s/%s' % (dir_out_full, fname_scan), ds)
# anonymize rda
if dir_input.split('/')[-1] == 'MRS':
[subjectID, sessionID] = get_patientID(dir_input, siteCodeDict)
scan_types = os.listdir(dir_input)
for scan_type in scan_types:
if scan_type.endswith('.rda'):
series_num = int(get_rda_field("SeriesNumber", scan_type))
scan_name = get_rda_field("SeriesDescription", scan_type)
new_series_description = get_scan_type(scan_name, lut_mrs_type)
dir_out_full = '%s/%s/%s/%03d-%s/RDA' % (
dir_out_base, subjectID, sessionID, series_num,
new_series_description)
if new_series_description is None:
print('ERROR - Unable to determine scan type for: %s/%s' %
(dir_input, scan_type))
else:
if not os.path.exists(dir_out_full):
os.makedirs(dir_out_full)
# TO DO: remove series_num from output rda filename, as per Steve Arnott's request
anonymize_rda(dir_input + '/' + scan_type,
dir_out_full + '/' + scan_type)
# anonymize twix
if dir_input.split('/')[-1] == 'twix':
[subjectID, sessionID] = get_patientID(dir_input, siteCodeDict)
scan_types = os.listdir(dir_input)
for scan_type in scan_types:
if scan_type.endswith('.dat'):
new_series_description = get_scan_type(scan_type,
lut_twix_type)
if new_series_description is None:
print('ERROR - Unable to determine scan type for: %s/%s' %
(dir_input, scan_type))
else:
dir_out_partial = '%s/%s/%s' % (dir_out_base, subjectID,
sessionID)
dir_to_match = os.listdir(dir_out_partial)
dir_match = [
x for x in dir_to_match
if '%s' % (new_series_description) in x
]
if not len(dir_match) == 1:
print(
'ERROR - Unable to determine matching directory for: %s/%s'
% (dir_input, scan_type))
else:
dir_out_full = '%s/%s/TWIX' % (dir_out_partial,
dir_match[0])
if not os.path.exists(dir_out_full):
os.makedirs(dir_out_full)
anonymize_twix(dir_input + '/' + scan_type,
dir_out_full + '/' + scan_type)
def patchDicomDir(self, inputDirPath, outputDirPath):
"""
Since CTK (rightly) requires certain basic information [1] before it can import
data files that purport to be dicom, this code patches the files in a directory
with some needed fields.
Calling this function with a directory path will make a patched copy of each file.
Importing the old files to CTK should still fail, but the new ones should work.
The directory is assumed to have a set of instances that are all from the
same study of the same patient. Also that each instance (file) is an
independent (multiframe) series.
[1] https://github.com/commontk/CTK/blob/16aa09540dcb59c6eafde4d9a88dfee1f0948edc/Libs/DICOM/Core/ctkDICOMDatabase.cpp#L1283-L1287
"""
import dicom
self.addLog('DICOM patching started...')
logging.debug('DICOM patch input directory: '+inputDirPath)
logging.debug('DICOM patch output directory: '+outputDirPath)
for rule in self.patchingRules:
rule.logCallback = self.addLog
rule.processStart(inputDirPath, outputDirPath)
for root, subFolders, files in os.walk(inputDirPath):
currentSubDir = os.path.relpath(root, inputDirPath)
rootOutput = os.path.join(outputDirPath, currentSubDir)
# Notify rules that processing of a new subdirectory started
for rule in self.patchingRules:
rule.processDirectory(currentSubDir)
for file in files:
filePath = os.path.join(root,file)
self.addLog('Examining %s...' % os.path.join(currentSubDir,file))
skipFileRequestingRule = None
for rule in self.patchingRules:
if rule.skipFile(currentSubDir):
skipFileRequestingRule = rule
break
if skipFileRequestingRule:
self.addLog(' Rule '+rule.__class__.__name__+' requested to skip this file.')
continue
try:
ds = dicom.read_file(filePath)
except (IOError, dicom.filereader.InvalidDicomError):
self.addLog(' Not DICOM file. Skipped.')
continue
self.addLog(' Patching...')
for rule in self.patchingRules:
rule.processDataSet(ds)
patchedFilePath = os.path.abspath(os.path.join(rootOutput,file))
for rule in self.patchingRules:
patchedFilePath = rule.generateOutputFilePath(ds, patchedFilePath)
######################################################
# Write
dirName = os.path.dirname(patchedFilePath)
if not os.path.exists(dirName):
os.makedirs(dirName)
self.addLog(' Writing DICOM...')
dicom.write_file(patchedFilePath, ds)
self.addLog(' Created DICOM file: %s' % patchedFilePath)
self.addLog('DICOM patching completed. Patched files are written to:\n{0}'.format(outputDirPath))
parser.add_option('--in_dir',\
help='Input dicom directory. All dicom images inside \
this directory will be anonymized. It is assumed that \
all files inside this directory are dicom files.' ,\
dest='in_dir', metavar='<string>', default=None)
parser.add_option('--out_dir',\
help='Output dicom directory. If none specified, the input \
directory be used as the output directory, overwriting \
the dicom images that were read from the directory.' ,\
dest='out_dir', metavar='<string>', default=None)
(options, args) = parser.parse_args()
if options.in_im is not None:
assert isfile(options.in_im), 'File does not exist'
ds = dicom.read_file(options.in_im)
anonymize_dicom(ds)
if options.out_im is not None:
dicom.write_file(options.out_im, ds)
if options.in_dir is not None:
assert isdir(options.in_dir), 'Directory does not exist'
files = [f for f in listdir(options.in_dir) if \
isfile(join(options.in_dir, f))]
for f in files:
ds = dicom.read_file(join(options.in_dir, f))
anonymize_dicom(ds)
if options.out_dir is not None:
assert isdir(options.out_dir), 'Directory does not exist'
dicom.write_file(join(options.out_dir, f), ds)
def patchDicomDir(self, inputDirPath, outputDirPath, generateMissingIds = True, generateImagePositionFromSliceThickness = True, anonymize = False):
"""
Since CTK (rightly) requires certain basic information [1] before it can import
data files that purport to be dicom, this code patches the files in a directory
with some needed fields. Apparently it is possible to export files from the
Philips PMS QLAB system with these fields missing.
Calling this function with a directory path will make a patched copy of each file.
Importing the old files to CTK should still fail, but the new ones should work.
The directory is assumed to have a set of instances that are all from the
same study of the same patient. Also that each instance (file) is an
independent (multiframe) series.
[1] https://github.com/commontk/CTK/blob/16aa09540dcb59c6eafde4d9a88dfee1f0948edc/Libs/DICOM/Core/ctkDICOMDatabase.cpp#L1283-L1287
"""
import dicom
if not outputDirPath:
outputDirPath = inputDirPath
self.addLog('DICOM patching started...')
logging.debug('DICOM patch input directory: '+inputDirPath)
logging.debug('DICOM patch output directory: '+outputDirPath)
patientIDToRandomIDMap = {}
studyUIDToRandomUIDMap = {}
seriesUIDToRandomUIDMap = {}
numberOfSeriesInStudyMap = {}
# All files without a patient ID will be assigned to the same patient
randomPatientID = dicom.UID.generate_uid(None)
requiredTags = ['PatientName', 'PatientID', 'StudyInstanceUID', 'SeriesInstanceUID', 'SeriesNumber']
for root, subFolders, files in os.walk(inputDirPath):
# Assume that all files in a directory belongs to the same study
randomStudyUID = dicom.UID.generate_uid(None)
# Assume that all files in a directory belongs to the same series
randomSeriesInstanceUID = dicom.UID.generate_uid(None)
currentSubDir = os.path.relpath(root, inputDirPath)
rootOutput = os.path.join(outputDirPath, currentSubDir)
for file in files:
filePath = os.path.join(root,file)
self.addLog('Examining %s...' % os.path.join(currentSubDir,file))
if file=='DICOMDIR':
self.addLog('DICOMDIR file is ignored (its contents may be inconsistent with the contents of the indexed DICOM files, therefore it is safer not to use it)')
continue
try:
ds = dicom.read_file(filePath)
except (IOError, dicom.filereader.InvalidDicomError):
self.addLog(' Not DICOM file. Skipped.')
continue
self.addLog(' Patching...')
######################################################
# Add missing IDs
if generateMissingIds:
for tag in requiredTags:
if not hasattr(ds,tag):
setattr(ds,tag,'')
# Generate a new SOPInstanceUID to avoid different files having the same SOPInstanceUID
ds.SOPInstanceUID = dicom.UID.generate_uid(None)
if ds.PatientName == '':
ds.PatientName = "Unspecified Patient"
if ds.PatientID == '':
ds.PatientID = randomPatientID
if ds.StudyInstanceUID == '':
ds.StudyInstanceUID = randomStudyUID
if ds.SeriesInstanceUID == '':
#ds.SeriesInstanceUID = dicom.UID.generate_uid(None) # this line has to be used if each file is a separate series
ds.SeriesInstanceUID = randomSeriesInstanceUID
# Generate series number to make it easier to identify a sequence within a study
if ds.SeriesNumber == '':
if ds.StudyInstanceUID not in numberOfSeriesInStudyMap:
numberOfSeriesInStudyMap[ds.StudyInstanceUID] = 0
numberOfSeriesInStudyMap[ds.StudyInstanceUID] = numberOfSeriesInStudyMap[ds.StudyInstanceUID] + 1
ds.SeriesNumber = numberOfSeriesInStudyMap[ds.StudyInstanceUID]
######################################################
# Add missing slice spacing info to multiframe files
numberOfFrames = ds.NumberOfFrames if hasattr(ds,'NumberOfFrames') else 1
if generateImagePositionFromSliceThickness and numberOfFrames>1:
# Multi-frame sequence, we may need to add slice positions
# Error in Dolphin 3D CBCT scanners, they store multiple frames but they keep using CTImageStorage as storage class
if ds.SOPClassUID == '1.2.840.10008.5.1.4.1.1.2': # Computed Tomography Image IOD
ds.SOPClassUID = '1.2.840.10008.5.1.4.1.1.2.1' # Enhanced CT Image IOD
sliceStartPosition = ds.ImagePositionPatient if hasattr(ds,'ImagePositionPatient') else [0,0,0]
sliceAxes = ds.ImageOrientationPatient if hasattr(ds,'ImagePositionPatient') else [1,0,0,0,1,0]
x = sliceAxes[:3]
y = sliceAxes[3:]
z = [x[1] * y[2] - x[2] * y[1], x[2] * y[0] - x[0] * y[2], x[0] * y[1] - x[1] * y[0]] # cross(x,y)
sliceSpacing = ds.SliceThickness if hasattr(ds,'SliceThickness') else 1.0
pixelSpacing = ds.PixelSpacing if hasattr(ds,'PixelSpacing') else [1.0, 1.0]
if not (dicom.tag.Tag(0x5200,0x9229) in ds):
# (5200,9229) SQ (Sequence with undefined length #=1) # u/l, 1 SharedFunctionalGroupsSequence
# (0020,9116) SQ (Sequence with undefined length #=1) # u/l, 1 PlaneOrientationSequence
# (0020,0037) DS [1.00000\0.00000\0.00000\0.00000\1.00000\0.00000] # 48, 6 ImageOrientationPatient
# (0028,9110) SQ (Sequence with undefined length #=1) # u/l, 1 PixelMeasuresSequence
# (0018,0050) DS [3.00000] # 8, 1 SliceThickness
# (0028,0030) DS [0.597656\0.597656] # 18, 2 PixelSpacing
planeOrientationDataSet = dicom.dataset.Dataset()
planeOrientationDataSet.ImageOrientationPatient = sliceAxes
planeOrientationSequence = dicom.sequence.Sequence()
planeOrientationSequence.insert(dicom.tag.Tag(0x0020,0x9116),planeOrientationDataSet)
pixelMeasuresDataSet = dicom.dataset.Dataset()
pixelMeasuresDataSet.SliceThickness = sliceSpacing
pixelMeasuresDataSet.PixelSpacing = pixelSpacing
pixelMeasuresSequence = dicom.sequence.Sequence()
pixelMeasuresSequence.insert(dicom.tag.Tag(0x0028,0x9110),pixelMeasuresDataSet)
sharedFunctionalGroupsDataSet = dicom.dataset.Dataset()
sharedFunctionalGroupsDataSet.PlaneOrientationSequence = planeOrientationSequence
sharedFunctionalGroupsDataSet.PixelMeasuresSequence = pixelMeasuresSequence
sharedFunctionalGroupsSequence = dicom.sequence.Sequence()
sharedFunctionalGroupsSequence.insert(dicom.tag.Tag(0x5200,0x9229),sharedFunctionalGroupsDataSet)
ds.SharedFunctionalGroupsSequence = sharedFunctionalGroupsSequence
if not (dicom.tag.Tag(0x5200,0x9230) in ds):
#(5200,9230) SQ (Sequence with undefined length #=54) # u/l, 1 PerFrameFunctionalGroupsSequence
# (0020,9113) SQ (Sequence with undefined length #=1) # u/l, 1 PlanePositionSequence
# (0020,0032) DS [-94.7012\-312.701\-806.500] # 26, 3 ImagePositionPatient
# (0020,9113) SQ (Sequence with undefined length #=1) # u/l, 1 PlanePositionSequence
# (0020,0032) DS [-94.7012\-312.701\-809.500] # 26, 3 ImagePositionPatient
# ...
perFrameFunctionalGroupsSequence = dicom.sequence.Sequence()
for frameIndex in range(numberOfFrames):
planePositionDataSet = dicom.dataset.Dataset()
slicePosition = [sliceStartPosition[0]+frameIndex*z[0]*sliceSpacing, sliceStartPosition[1]+frameIndex*z[1]*sliceSpacing, sliceStartPosition[2]+frameIndex*z[2]*sliceSpacing]
planePositionDataSet.ImagePositionPatient = slicePosition
planePositionSequence = dicom.sequence.Sequence()
planePositionSequence.insert(dicom.tag.Tag(0x0020,0x9113),planePositionDataSet)
perFrameFunctionalGroupsDataSet = dicom.dataset.Dataset()
perFrameFunctionalGroupsDataSet.PlanePositionSequence = planePositionSequence
perFrameFunctionalGroupsSequence.insert(dicom.tag.Tag(0x5200,0x9230),perFrameFunctionalGroupsDataSet)
ds.PerFrameFunctionalGroupsSequence = perFrameFunctionalGroupsSequence
######################################################
# Anonymize
if anonymize:
self.addLog(' Anonymizing...')
ds.StudyDate = ''
ds.StudyTime = ''
ds.ContentDate = ''
ds.ContentTime = ''
ds.AccessionNumber = ''
ds.ReferringPhysiciansName = ''
ds.PatientsBirthDate = ''
ds.PatientsSex = ''
ds.StudyID = ''
ds.PatientName = "Unspecified Patient"
# replace ids with random values - re-use if we have seen them before
if ds.PatientID not in patientIDToRandomIDMap:
patientIDToRandomIDMap[ds.PatientID] = dicom.UID.generate_uid(None)
ds.PatientID = patientIDToRandomIDMap[ds.PatientID]
if ds.StudyInstanceUID not in studyUIDToRandomUIDMap:
studyUIDToRandomUIDMap[ds.StudyInstanceUID] = dicom.UID.generate_uid(None)
ds.StudyInstanceUID = studyUIDToRandomUIDMap[ds.StudyInstanceUID]
if ds.SeriesInstanceUID not in studyUIDToRandomUIDMap:
seriesUIDToRandomUIDMap[ds.SeriesInstanceUID] = dicom.UID.generate_uid(None)
ds.SeriesInstanceUID = seriesUIDToRandomUIDMap[ds.SeriesInstanceUID]
######################################################
# Write
if inputDirPath==outputDirPath:
(name, ext) = os.path.splitext(filePath)
patchedFilePath = name + ('-anon' if anonymize else '') + '-patched' + ext
else:
patchedFilePath = os.path.abspath(os.path.join(rootOutput,file))
if not os.path.exists(rootOutput):
os.makedirs(rootOutput)
self.addLog(' Writing DICOM...')
dicom.write_file(patchedFilePath, ds)
self.addLog(' Created DICOM file: %s' % patchedFilePath)
self.addLog('DICOM patching completed.')
def patchDicomDir(self, inputDirPath, outputDirPath):
"""
Since CTK (rightly) requires certain basic information [1] before it can import
data files that purport to be dicom, this code patches the files in a directory
with some needed fields.
Calling this function with a directory path will make a patched copy of each file.
Importing the old files to CTK should still fail, but the new ones should work.
The directory is assumed to have a set of instances that are all from the
same study of the same patient. Also that each instance (file) is an
independent (multiframe) series.
[1] https://github.com/commontk/CTK/blob/16aa09540dcb59c6eafde4d9a88dfee1f0948edc/Libs/DICOM/Core/ctkDICOMDatabase.cpp#L1283-L1287
"""
import dicom
self.addLog('DICOM patching started...')
logging.debug('DICOM patch input directory: '+inputDirPath)
logging.debug('DICOM patch output directory: '+outputDirPath)
for rule in self.patchingRules:
rule.logCallback = self.addLog
rule.processStart(inputDirPath, outputDirPath)
for root, subFolders, files in os.walk(inputDirPath):
currentSubDir = os.path.relpath(root, inputDirPath)
rootOutput = os.path.join(outputDirPath, currentSubDir)
# Notify rules that processing of a new subdirectory started
for rule in self.patchingRules:
rule.processDirectory(currentSubDir)
for file in files:
filePath = os.path.join(root,file)
self.addLog('Examining %s...' % os.path.join(currentSubDir,file))
skipFileRequestingRule = None
for rule in self.patchingRules:
if rule.skipFile(currentSubDir):
skipFileRequestingRule = rule
break
if skipFileRequestingRule:
self.addLog(' Rule '+rule.__class__.__name__+' requested to skip this file.')
continue
try:
ds = dicom.read_file(filePath)
except (IOError, dicom.filereader.InvalidDicomError):
self.addLog(' Not DICOM file. Skipped.')
continue
self.addLog(' Patching...')
for rule in self.patchingRules:
rule.processDataSet(ds)
patchedFilePath = os.path.abspath(os.path.join(rootOutput,file))
for rule in self.patchingRules:
patchedFilePath = rule.generateOutputFilePath(ds, patchedFilePath)
######################################################
# Write
dirName = os.path.dirname(patchedFilePath)
if not os.path.exists(dirName):
os.makedirs(dirName)
self.addLog(' Writing DICOM...')
dicom.write_file(patchedFilePath, ds)
self.addLog(' Created DICOM file: %s' % patchedFilePath)
self.addLog('DICOM patching completed. Patched files are written to:\n{0}'.format(outputDirPath))
def BrowseFile(self, fname):
"""Allows the user to select a filename"""
self._savedFileName = fname
self._reader._reader = None
full_name = os.path.abspath(fname)
_dir, fname = os.path.split(full_name)
self._num_list = []
self._file_list = [full_name]
self._slice_locations = []
# is file a DICOM file?
ds = {}
examiner = MicroViewIO.wxMicroViewDICOMExaminer(_dir)
self.bIsDICOM = examiner.IsDICOMFile(full_name)
self.match_tags = {}
if self.bIsDICOM:
try:
ds = dicom.read_file(
full_name, stop_before_pixels=True, force=True)
if 'PhotometricInterpretation' not in ds:
self.bIsDICOM = False
if 'AcquisitionNumber' in ds:
self.bHasAcquisitionNumber = True
tags = ['StudyID', 'SeriesNumber', 'AcquisitionNumber']
if ds.Modality == 'MR':
tags.append('EchoNumbers')
for tag in tags:
if tag in ds:
self.match_tags[tag] = ds.get(tag)
except:
pass
with wx.BusyCursor():
self.update_file_list(full_name, ds, examiner)
# update DICOMDIR file
if self.bIsDICOM:
dicomdir_filename = os.path.join(_dir, 'DICOMDIR')
dicomdir = examiner.GetDICOMDIR().get_dicomdir()
if dicomdir:
try:
dicom.write_file(dicomdir_filename, dicomdir)
logging.info(
"Successfully wrote {}.".format(dicomdir_filename))
except:
pass
self.guess_image_pixel_size(full_name)
self.updateGUIState()
self.updateButtonState()
# if files are DICOM-format, fill in a reasonable guess at a name
if self.bIsDICOM:
name = ''
if 'StudyID' in ds:
name += '{}_'.format(ds.StudyID)
if 'SeriesDescription' in ds:
name += '{}_'.format(ds.SeriesDescription.strip()
).replace(' ', '_')
if 'PatientName' in ds:
name += '{}'.format(ds.PatientName.strip()).replace(' ', '_')
if not name:
name = "DICOM Import"
self.m_textCtrlImageTitle.SetValue(name)
anon_name = args.anon_name
else:
anon_name = 'anon'
if args.anon_id:
anon_id = args.anon_id
else:
anon_id = '00000000'
FOR_uid = get_dicomuid()
for dcmfile in dcm_list:
dcmfilename = dcmfile.strip()
dcm = dicom.read_file(dcmfilename)
dcm.FrameOfReferenceUID = FOR_uid
dcm.PatientsName = anon_name
dcm.PatientID = anon_id
dcm.PatientsSex = 'O'
dcm.SOPInstanceUID = get_dicomuid()
dcm.StudyUID = get_dicomuid()
dcm.SeriesUID = get_dicomuid()
dcm.FrameUID = get_dicomuid()
dcm.SyncUID = get_dicomuid()
dcm.SrUID = get_dicomuid()
dcm.StudyInstanceUID = get_dicomuid()
dcm.SeriesInstanceUID = get_dicomuid()
outfile = 'anon_'+dcmfilename
dicom.write_file(outfile, dcm)