def test_is_localizer():
input_label = 'LOCALIZER'
assert common_utils.is_localizer(input_label)
input_label = 2
assert not common_utils.is_localizer(input_label)
def classify_CT(df, dcm_metadata, acquisition):
'''
Classifies a CT dicom series
Args:
df (DataFrame): A pandas DataFrame where each row is a dicom image header information
Returns:
dict: The dictionary for the CT classification
'''
log.info("Determining CT Classification...")
header_dicom = dcm_metadata['info']['header']['dicom']
series_description = header_dicom.get('SeriesDescription') or ''
classification_source = {}
classification = {}
info_object = {}
if common_utils.is_localizer(acquisition.label) or common_utils.is_localizer(series_description) or len(df) < 10:
classification['Scan Type'] = ['Localizer']
else:
classification['Scan Type'] = \
common_utils.get_scan_type_classification(
acquisition.label, header_dicom)
if not classification['Scan Type']:
classification['Scan Type'] = \
common_utils.get_scan_type_classification(
series_description, header_dicom)
# # Compute scan coverage
scan_coverage, info_object = \
common_utils.compute_scan_coverage_if_original(header_dicom, df,
info_object)
# # Reconstruction window
reconstruction_window = None
reconstruction_window = common_utils.get_reconstruction_window(
acquisition.label)
if reconstruction_window:
info_object['ReconstructionWindow'] = reconstruction_window
# # Classify Scan orientation
classification['Scan Orientation'] = common_utils.get_scan_orientation(acquisition.label)
if not classification['Scan Orientation']:
classification['Scan Orientation'] = common_utils.get_scan_orientation(
series_description)
# # Classify Anatomy
classification = common_utils.classify_anatomy(
classification, acquisition, series_description, scan_coverage)
# # Classify Contrast
classification['Contrast'] = common_utils.get_contrast_classification(
acquisition.label)
if not classification['Contrast']:
classification['Contrast'] = \
common_utils.get_contrast_classification(
series_description)
# # Set Custom classification to 'Unknown' if any of the CT classification keys is absent or empty:
if 'Scan Orientation' not in classification.keys():
classification['Custom'] = ['Unknown']
elif not classification['Scan Orientation']:
classification['Custom'] = ['Unknown']
else:
# If classification exists then create a tag in classification_source dict
# to indicate the source of Scan Orientation classification
# This classification_source dict is then added to info object
classification_source['Scan Orientation'] = 'Original'
if 'Contrast' not in classification.keys():
classification['Custom'] = ['Unknown']
elif not classification['Contrast']:
classification['Custom'] = ['Unknown']
else:
# If classification exists then create a tag in classification_source dict
# to indicate the source of Contrast classification
# This classification_source dict is then added to info object
classification_source['Contrast'] = 'Original'
if 'Anatomy' not in classification.keys():
classification['Custom'] = ['Unknown']
elif not classification['Anatomy']:
classification['Custom'] = ['Unknown']
else:
# If classification exists then create a tag in classification_source dict
# to indicate the source of Anatomy classification
# This classification_source dict is then added to info object
classification_source['Anatomy'] = 'Original'
# # Scan Coverage
if scan_coverage:
spacing_between_slices = scan_coverage / len(df)
info_object['SpacingBetweenSlices'] = round(spacing_between_slices, 2)
info_object['ClassificationSource'] = classification_source
dcm_metadata['info'].update(info_object)
dcm_metadata['classification'] = classification
return dcm_metadata
def classify_CT(df, dcm_metadata, acquisition):
'''
Classifies a CT dicom series
Args:
df (DataFrame): A pandas DataFrame where each row is a dicom image header information
Returns:
dict: The dictionary for the CT classification
'''
log.info("Determining CT Classification...")
single_header_object = dcm_metadata['info']['header']['dicom']
series_description = single_header_object.get('SeriesDescription') or ''
classifications = {}
info_object = {}
if common_utils.is_localizer(
acquisition.label) or common_utils.is_localizer(
series_description) or len(df) < 10:
classifications['Scan Type'] = ['Localizer']
else:
classifications['Scan Type'] = get_scan_type_classification(
acquisition.label, single_header_object)
if not classifications['Scan Type']:
classifications['Scan Type'] = get_scan_type_classification(
series_description, single_header_object)
scan_coverage = None
if single_header_object['ImageType'][0] == 'ORIGINAL':
scan_coverage = common_utils.compute_scan_coverage(df)
if scan_coverage:
info_object['ScanCoverage'] = scan_coverage
# # Reconstruction window
reconstruction_window = None
reconstruction_window = get_reconstruction_window(acquisition.label)
if reconstruction_window:
info_object['ReconstructionWindow'] = reconstruction_window
# # Scan orientation
scan_orientation = None
scan_orientation = get_scan_orientation(acquisition.label)
if scan_orientation:
info_object['ScanOrientation'] = scan_orientation
else:
scan_orientation = get_scan_orientation(series_description)
if scan_orientation:
info_object['ScanOrientation'] = scan_orientation
# # Anatomy
classifications['Anatomy'] = get_anatomy_from_label(acquisition.label)
if not classifications['Anatomy']:
classifications['Anatomy'] = get_anatomy_from_label(
series_description)
if not classifications['Anatomy']:
classifications['Anatomy'] = get_anatomy_from_scan_coverage(
scan_coverage)
# # Contrast
classifications['Contrast'] = get_contrast_classification(
acquisition.label)
if not classifications['Contrast']:
classifications['Contrast'] = get_contrast_classification(
series_description)
if scan_coverage:
spacing_between_slices = scan_coverage / len(df)
info_object['SpacingBetweenSlices'] = round(
spacing_between_slices, 2)
dcm_metadata['info'].update(info_object)
dcm_metadata['classification'] = classifications
return dcm_metadata
def infer_classification(label):
"""
Get classification based on acquisition label
"""
if not label:
return {}
else:
classification = {}
if is_anatomy_inplane(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['T1']
classification['Features'] = ['In-Plane']
elif is_fieldmap(label):
classification['Intent'] = ['Fieldmap']
classification['Measurement'] = ['B0']
elif is_diffusion_derived(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['Diffusion']
classification['Features'] = ['Derived']
elif is_diffusion(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['Diffusion']
elif is_functional_derived(label):
classification['Intent'] = ['Functional']
classification['Features'] = ['Derived']
elif is_functional(label):
classification['Intent'] = ['Functional']
classification['Measurement'] = ['T2*']
elif is_anatomy_t1(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['T1']
elif is_anatomy_t2star(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['T2*']
elif is_anatomy_t2(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['T2']
elif is_anatomy(label):
classification['Intent'] = ['Structural']
elif is_swi(label) :
classification['Intent'] = ['Structural']
classification['Measurement'] = ['Susceptibility']
elif common_utils.is_localizer(label):
classification['Intent'] = ['Localizer']
classification['Measurement'] = ['T2']
elif is_shim(label):
classification['Intent'] = ['Shim']
elif is_calibration(label):
classification['Intent'] = ['Calibration']
elif is_coil_survey(label):
classification['Intent'] = ['Calibration']
classification['Measurement'] = ['B1']
elif is_proton_density(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['PD']
elif is_perfusion(label):
classification['Measurement'] = ['Perfusion']
elif is_spectroscopy(label):
classification['Intent'] = ['Spectroscopy']
elif is_phase_map(label):
classification['Custom'] = ['Phase Map']
elif is_screenshot(label):
classification['Intent'] = ['Screenshot']
else:
print(label.strip('\n') + ' --->>>> unknown')
# Add Contrast-agent
if is_contrast(label):
class_features = classification.get('Features', [])
class_features.append('Contrast-Agent')
classification['Features'] = class_features
# Add features to classification
features = feature_check(label)
if features:
class_features = classification.get('Features', [])
[class_features.append(x) for x in features if x not in class_features]
classification['Features'] = class_features
# Add measurements to classification
measurements = measurement_check(label)
if measurements:
class_measurement = classification.get('Measurement', [])
[class_measurement.append(x) for x in measurements if x not in class_measurement]
classification['Measurement'] = class_measurement
# Add intents to classification
intents = intent_check(label)
if intents:
class_intent = classification.get('Intent', [])
[class_intent.append(x) for x in intents if x not in class_intent]
classification['Intent'] = class_intent
return classification
def infer_classification(label):
'''
Get classification based on acquisition label
'''
if not label:
return {}
else:
classification = {}
if is_anatomy_inplane(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['T1']
classification['Features'] = ['In-Plane']
elif is_fieldmap(label):
classification['Intent'] = ['Fieldmap']
classification['Measurement'] = ['B0']
elif is_diffusion_derived(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['Diffusion']
classification['Features'] = ['Derived']
elif is_diffusion(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['Diffusion']
elif is_functional_derived(label):
classification['Intent'] = ['Functional']
classification['Features'] = ['Derived']
elif is_functional(label):
classification['Intent'] = ['Functional']
classification['Measurement'] = ['T2*']
elif is_anatomy_t1(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['T1']
elif is_anatomy_t2(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['T2']
elif is_anatomy(label):
classification['Intent'] = ['Structural']
elif common_utils.is_localizer(label):
classification['Intent'] = ['Localizer']
classification['Measurement'] = ['T2']
elif is_shim(label):
classification['Intent'] = ['Shim']
elif is_calibration(label):
classification['Intent'] = ['Calibration']
elif is_coil_survey(label):
classification['Intent'] = ['Calibration']
classification['Measurement'] = ['B1']
elif is_proton_density(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['PD']
elif is_perfusion(label):
classification['Measurement'] = ['Perfusion']
elif is_spectroscopy(label):
classification['Intent'] = ['Spectroscopy']
elif is_phase_map(label):
classification['Custom'] = ['Phase Map']
elif is_screenshot(label):
classification['Intent'] = ['Screenshot']
else:
print(label.strip('\n') + ' --->>>> unknown')
return classification
def infer_classification(label):
"""
Get classification based on acquisition label
"""
if not label:
return {}
else:
classification = {}
if is_anatomy_inplane(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['T1']
classification['Features'] = ['In-Plane']
elif is_fieldmap(label):
classification['Intent'] = ['Fieldmap']
classification['Measurement'] = ['B0']
elif is_diffusion_derived(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['Diffusion']
classification['Features'] = ['Derived']
elif is_diffusion(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['Diffusion']
elif is_functional_derived(label):
classification['Intent'] = ['Functional']
classification['Features'] = ['Derived']
elif is_functional(label):
classification['Intent'] = ['Functional']
classification['Measurement'] = ['T2*']
elif is_anatomy_t1(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['T1']
elif is_anatomy_t2(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['T2']
elif is_anatomy(label):
classification['Intent'] = ['Structural']
elif is_swi(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['Susceptibility']
elif common_utils.is_localizer(label):
classification['Intent'] = ['Localizer']
classification['Measurement'] = ['T2']
elif is_shim(label):
classification['Intent'] = ['Shim']
elif is_calibration(label):
classification['Intent'] = ['Calibration']
elif is_coil_survey(label):
classification['Intent'] = ['Calibration']
classification['Measurement'] = ['B1']
elif is_proton_density(label):
classification['Intent'] = ['Structural']
classification['Measurement'] = ['PD']
elif is_perfusion(label):
classification['Measurement'] = ['Perfusion']
elif is_spectroscopy(label):
classification['Intent'] = ['Spectroscopy']
elif is_phase_map(label):
classification['Custom'] = ['Phase Map']
elif is_screenshot(label):
classification['Intent'] = ['Screenshot']
else:
print(label.strip('\n') + ' --->>>> unknown')
# Add features to classification
features = feature_check(label)
if features:
class_features = classification.get('Features', [])
[
class_features.append(x) for x in features
if x not in class_features
]
classification['Features'] = class_features
# Add measurements to classification
measurements = measurement_check(label)
if measurements:
class_measurement = classification.get('Measurement', [])
[
class_measurement.append(x) for x in measurements
if x not in class_measurement
]
classification['Measurement'] = class_measurement
# Add intents to classification
intents = intent_check(label)
if intents:
class_intent = classification.get('Intent', [])
[class_intent.append(x) for x in intents if x not in class_intent]
classification['Intent'] = class_intent
# Code for Contrast/ No-Contrast Classification ( New Feature added on May 13 2021)
# Any label that satisfies the Contrast or No Contrast condition it would be tagged as Contrast/No Contrast
# If a label satisfies both Contrast and No Contrast condition, Contrast would be negated and tagged as No Contrast
# If a label doesn't satisfy both the conditions, the Classification would be left empty
if is_contrast(label):
classification['Contrast'] = ['Contrast']
if is_not_contrast(label):
classification['Contrast'] = ['No Contrast']
return classification
def classify_CT(df, dcm_metadata, acquisition):
'''
Classifies a CT dicom series
Args:
df (DataFrame): A pandas DataFrame where each row is a dicom image header information
Returns:
dict: The dictionary for the CT classification
'''
log.info("Determining CT Classification...")
header_dicom = dcm_metadata['info']['header']['dicom']
series_description = header_dicom.get('SeriesDescription') or ''
classification = {}
info_object = {}
if common_utils.is_localizer(
acquisition.label) or common_utils.is_localizer(
series_description) or len(df) < 10:
classification['Scan Type'] = ['Localizer']
else:
classification['Scan Type'] = \
common_utils.get_scan_type_classification(
acquisition.label, header_dicom)
if not classification['Scan Type']:
classification['Scan Type'] = \
common_utils.get_scan_type_classification(
series_description, header_dicom)
# Compute scan coverage
scan_coverage, info_object = \
common_utils.compute_scan_coverage_if_original(header_dicom, df,
info_object)
# # Reconstruction window
reconstruction_window = None
reconstruction_window = common_utils.get_reconstruction_window(
acquisition.label)
if reconstruction_window:
info_object['ReconstructionWindow'] = reconstruction_window
# # Scan orientation
scan_orientation = None
scan_orientation = common_utils.get_scan_orientation(acquisition.label)
if scan_orientation:
info_object['ScanOrientation'] = scan_orientation
else:
scan_orientation = common_utils.get_scan_orientation(
series_description)
if scan_orientation:
info_object['ScanOrientation'] = scan_orientation
# Classify Anatomy
classification = common_utils.classify_anatomy(classification,
acquisition,
series_description,
scan_coverage)
# # Contrast
classification['Contrast'] = common_utils.get_contrast_classification(
acquisition.label)
if not classification['Contrast']:
classification['Contrast'] = \
common_utils.get_contrast_classification(
series_description)
if scan_coverage:
spacing_between_slices = scan_coverage / len(df)
info_object['SpacingBetweenSlices'] = round(
spacing_between_slices, 2)
dcm_metadata['info'].update(info_object)
dcm_metadata['classification'] = classification
return dcm_metadata