classification_from_label.py

#!/usr/bin/env python
'''
Infer acquisition classification by parsing the description label.
'''

import re


def feature_check(label):
    '''Check the label for a list of features.'''

    feature_list = ['2D', 'AAscout', 'Spin-Echo', 'Gradient-Echo',
                   'EPI', 'WASSR', 'FAIR', 'FAIREST', 'PASL', 'EPISTAR',
                   'PICORE', 'pCASL', 'MPRAGE', 'MP2RAGE', 'FLAIR',
                   'SWI', 'QSM', 'RMS', 'DTI', 'DSI', 'DKI', 'HARDI',
                   'NODDI', 'Water-Reference', 'Transmit-Reference',
                   'SBRef', 'Uniform', 'Singlerep', 'QC', 'TRACE',
                   'FA', 'MIP', 'Navigator', 'Contrast-Agent',
                   'Phase-Contrast', 'TOF', 'VASO', 'iVASO', 'DSC',
                   'DCE', 'Task', 'Resting-State', 'PRESS', 'STEAM',
                   'M0', 'Phase-Reversed', 'Spiral', 'SPGR',
                   'Quantitative', 'Multi-Shell', 'Multi-Echo', 'Multi-Flip',
                   'Multi-Band', 'Steady-State', '3D', 'Compressed-Sensing',
                   'Eddy-Current-Corrected', 'Fieldmap-Corrected',
                   'Gradient-Unwarped', 'Motion-Corrected', 'Physio-Corrected',
                   'Derived', 'In-Plane', 'Phase', 'Magnitude']

    return _find_matches(label, feature_list)


def measurement_check(label):
    '''Check the label for a list of measurements.'''

    measurement_list = ['MRA', 'CEST', 'T1rho', 'SVS', 'CSI', 'EPSI', 'BOLD',
                        'Phoenix','B0', 'B1', 'T1', 'T2', 'T2*', 'PD', 'MT',
                        'Perfusion','Diffusion', 'Susceptibility', 'Fingerprinting']

    return _find_matches(label, measurement_list)


def intent_check(label):
    '''Check the label for a list of intents.'''

    intent_list = [ 'Localizer', 'Shim', 'Calibration', 'Fieldmap', 'Structural',
                    'Functional', 'Screenshot', 'Non-Image', 'Spectroscopy' ]

    return _find_matches(label, intent_list)


def _find_matches(label, list):
    """For a given list find those entries that match a given label."""

    matches = []

    for l in list:
        regex = _compile_regex(l)
        if regex.findall(label):
            matches.append(l)

    return matches


def _compile_regex(string):
    """Generate the regex for label checking"""
    # Escape * for T2*
    if string == 'T2*':
        string = 'T2\*'
        regex = re.compile(r"(\b%s\b)|(_%s_)|(_%s)|(%s_)|(t2star)" % (string,string,string,string), re.IGNORECASE)
    # Prevent T2 from capturing T2*
    elif string == 'T2':
        regex = re.compile(r"(\b%s\b)|(_%s_)|(_%s$)|(%s_)" % (string,string,string,string), re.IGNORECASE)
    else:
        regex = re.compile(r"(\b%s\b)|(_%s_)|(_%s)|(%s_)" % (string,string,string,string), re.IGNORECASE)
    return regex


# Anatomy, T1
def is_anatomy_t1(label):
    regexes = [
        re.compile('t1', re.IGNORECASE),
        re.compile('t1w', re.IGNORECASE),
        re.compile('(?=.*3d anat)(?![inplane])', re.IGNORECASE),
        re.compile('(?=.*3d)(?=.*bravo)(?![inplane])', re.IGNORECASE),
        re.compile('spgr', re.IGNORECASE),
        re.compile('tfl', re.IGNORECASE),
        re.compile('mprage', re.IGNORECASE),
        re.compile('(?=.*mm)(?=.*iso)', re.IGNORECASE),
        re.compile('(?=.*mp)(?=.*rage)', re.IGNORECASE)
    ]
    return regex_search_label(regexes, label)

# Anatomy, T2
def is_anatomy_t2(label):
    regexes = [
        re.compile('t2[^*]*$', re.IGNORECASE)
    ]
    return regex_search_label(regexes, label)

# Aanatomy, Inplane
def is_anatomy_inplane(label):
    regexes = [
        re.compile('inplane', re.IGNORECASE)
    ]
    return regex_search_label(regexes, label)

# Anatomy, other
def is_anatomy(label):
    regexes = [
        re.compile('(?=.*IR)(?=.*EPI)', re.IGNORECASE),
        re.compile('flair', re.IGNORECASE)
    ]
    return regex_search_label(regexes, label)

# Diffusion
def is_diffusion(label):
    regexes = [
        re.compile('dti', re.IGNORECASE),
        re.compile('dwi', re.IGNORECASE),
        re.compile('diff_', re.IGNORECASE),
        re.compile('diffusion', re.IGNORECASE),
        re.compile('(?=.*diff)(?=.*dir)', re.IGNORECASE),
        re.compile('hardi', re.IGNORECASE)
        ]
    return regex_search_label(regexes, label)

# Diffusion - Derived
def is_diffusion_derived(label):
    regexes = [
        re.compile('_ADC$', re.IGNORECASE),
        re.compile('_TRACEW$', re.IGNORECASE),
        re.compile('_ColFA$', re.IGNORECASE),
        re.compile('_FA$', re.IGNORECASE),
        re.compile('_EXP$', re.IGNORECASE)
        ]
    return regex_search_label(regexes, label)

# Functional
def is_functional(label):
    regexes = [
        re.compile('functional', re.IGNORECASE),
        re.compile('fmri', re.IGNORECASE),
        re.compile('func', re.IGNORECASE),
        re.compile('bold', re.IGNORECASE),
        re.compile('resting', re.IGNORECASE),
        re.compile('(?=.*rest)(?=.*state)', re.IGNORECASE),
        # NON-STANDARD
        re.compile('(?=.*ret)(?=.*bars)', re.IGNORECASE),
        re.compile('(?=.*ret)(?=.*wedges)', re.IGNORECASE),
        re.compile('(?=.*ret)(?=.*rings)', re.IGNORECASE),
        re.compile('(?=.*ret)(?=.*check)', re.IGNORECASE),
        re.compile('go-no-go', re.IGNORECASE),
        re.compile('words', re.IGNORECASE),
        re.compile('checkers', re.IGNORECASE),
        re.compile('retinotopy', re.IGNORECASE),
        re.compile('faces', re.IGNORECASE),
        re.compile('rings', re.IGNORECASE),
        re.compile('wedges', re.IGNORECASE),
        re.compile('emoreg', re.IGNORECASE),
        re.compile('conscious', re.IGNORECASE),
        re.compile('^REST$'),
        re.compile('ep2d', re.IGNORECASE),
        re.compile('task', re.IGNORECASE),
        re.compile('rest', re.IGNORECASE),
        re.compile('fBIRN', re.IGNORECASE),
        re.compile('^Curiosity', re.IGNORECASE),
        re.compile('^DD_', re.IGNORECASE),
        re.compile('^Poke', re.IGNORECASE),
        re.compile('^Effort', re.IGNORECASE),
        re.compile('emotion|conflict', re.IGNORECASE)
        ]
    return regex_search_label(regexes, label)

# Functional, Derived
def is_functional_derived(label):
    regexes = [
        re.compile('mocoseries', re.IGNORECASE),
        re.compile('GLM$', re.IGNORECASE),
        re.compile('t-map', re.IGNORECASE),
        re.compile('design', re.IGNORECASE),
        re.compile('StartFMRI', re.IGNORECASE)
        ]
    return regex_search_label(regexes, label)

# Localizer
def is_localizer(label):
    regexes = [
        re.compile('localizer', re.IGNORECASE),
        re.compile('localiser', re.IGNORECASE),
        re.compile('survey', re.IGNORECASE),
        re.compile('loc\.', re.IGNORECASE),
        re.compile(r'\bscout\b', re.IGNORECASE),
        re.compile('(?=.*plane)(?=.*loc)', re.IGNORECASE),
        re.compile('(?=.*plane)(?=.*survey)', re.IGNORECASE),
        re.compile('3-plane', re.IGNORECASE),
        re.compile('^loc*', re.IGNORECASE),
        re.compile('Scout', re.IGNORECASE),
        re.compile('AdjGre', re.IGNORECASE)
        ]
    return regex_search_label(regexes, label)

# Shim
def is_shim(label):
    regexes = [
        re.compile('(?=.*HO)(?=.*shim)', re.IGNORECASE), # Contians 'ho' and 'shim'
        re.compile(r'\bHOS\b', re.IGNORECASE),
        re.compile('_HOS_', re.IGNORECASE),
        re.compile('.*shim', re.IGNORECASE)
        ]
    return regex_search_label(regexes, label)

# Fieldmap
def is_fieldmap(label):
    regexes = [
        re.compile('(?=.*field)(?=.*map)', re.IGNORECASE),
        re.compile('(?=.*bias)(?=.*ch)', re.IGNORECASE),
        re.compile('field', re.IGNORECASE),
        re.compile('fmap', re.IGNORECASE),
        re.compile('topup', re.IGNORECASE),
        re.compile('DISTORTION', re.IGNORECASE),
        re.compile('se[-_][aprl]{2}$', re.IGNORECASE)
        ]
    return regex_search_label(regexes, label)

# Calibration
def is_calibration(label):
    regexes = [
        re.compile('(?=.*asset)(?=.*cal)', re.IGNORECASE),
        re.compile('^asset$', re.IGNORECASE),
        re.compile('calibration', re.IGNORECASE)
        ]
    return regex_search_label(regexes, label)

# Coil Survey
def is_coil_survey(label):
    regexes = [
        re.compile('(?=.*coil)(?=.*survey)', re.IGNORECASE)
        ]
    return regex_search_label(regexes, label)

# Perfusion: Arterial Spin Labeling
def is_perfusion(label):
    regexes = [
        re.compile('asl', re.IGNORECASE),
        re.compile('(?=.*blood)(?=.*flow)', re.IGNORECASE),
        re.compile('(?=.*art)(?=.*spin)', re.IGNORECASE),
        re.compile('tof', re.IGNORECASE),
        re.compile('perfusion', re.IGNORECASE),
        re.compile('angio', re.IGNORECASE),
        ]
    return regex_search_label(regexes, label)

# Proton Density
def is_proton_density(label):
    regexes = [
        re.compile('^PD$'),
        re.compile('(?=.*proton)(?=.*density)', re.IGNORECASE),
        re.compile('pd_'),
        re.compile('_pd')
        ]
    return regex_search_label(regexes, label)

# Phase Map
def is_phase_map(label):
    regexes = [
        re.compile('(?=.*phase)(?=.*map)', re.IGNORECASE),
        re.compile('^phase$', re.IGNORECASE)
        ]
    return regex_search_label(regexes, label)

# Screen Save / Screenshot
def is_screenshot(label):
    regexes = [
        re.compile('(?=.*screen)(?=.*save)', re.IGNORECASE),
        re.compile('.*screenshot', re.IGNORECASE),
        re.compile('.*screensave', re.IGNORECASE)
        ]
    return regex_search_label(regexes, label)



# Utility:  Check a list of regexes for truthyness
def regex_search_label(regexes, label):
    if any(regex.search(label) for regex in regexes):
            return True
    else:
            return False

# Spectroscopy
def is_spectroscopy(label):
    regexes = [
        re.compile('mrs', re.IGNORECASE),
        re.compile('svs', re.IGNORECASE),
        re.compile('gaba', re.IGNORECASE),
        re.compile('csi', re.IGNORECASE),
        re.compile('nfl', re.IGNORECASE),
        re.compile('mega', re.IGNORECASE),
        re.compile('press', re.IGNORECASE),
        re.compile('spect', re.IGNORECASE)
        ]
    return regex_search_label(regexes, label)

# Susceptability
def is_susceptability(label):
    regexes = [
        re.compile('swi', re.IGNORECASE),
        re.compile('mag_images', re.IGNORECASE),
        re.compile('pha_images', re.IGNORECASE),
        re.compile('mip_images', re.IGNORECASE)
        ]
    return regex_search_label(regexes, label)


# Call all functions to determine new label
def infer_classification(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_t2(label):
            classification['Intent'] = ['Structural']
            classification['Measurement'] = ['T2']
        elif is_anatomy_t1(label):
            classification['Intent'] = ['Structural']
            classification['Measurement'] = ['T1']
        elif is_anatomy(label):
            classification['Intent'] = ['Structural']
        elif 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_susceptability(label):
            classification['Measurement'] = ['Susceptability']
        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

    return classification