GI-RADS Gynecologic Imaging Reporting and Data System

{
  "categories": [
    {
      "outcome_code": "1",
      "original_name": "definitively_benign",
      "estimated_probability_of_malignancy": "0_percent",
      "pattern": "normal_ovaries_identified_and_no_adnexal_mass_seen"
    },
    {
      "outcome_code": "2",
      "original_name": "very_probably_benign",
      "estimated_probability_of_malignancy": "less_than_1_percent",
      "pattern": "functional_origin_such_as_follicle_corpus_luteum_or_hemorrhagic_cyst"
    },
    {
      "outcome_code": "3",
      "original_name": "probably_benign",
      "estimated_probability_of_malignancy": "1_to_4_percent",
      "pattern": "recognizable_persistent_benign_lesion_such_as_endometrioma_teratoma_simple_cyst_hydrosalpinx_paraovarian_cyst_peritoneal_pseudocyst_pedunculated_myoma_or_PID_pattern"
    },
    {
      "outcome_code": "4",
      "original_name": "probably_malignant",
      "estimated_probability_of_malignancy": "5_to_20_percent",
      "suspicious_feature_count": "1_or_2",
      "prerequisite": "not_classified_as_GI_RADS_1_2_or_3"
    },
    {
      "outcome_code": "5",
      "original_name": "very_probably_malignant",
      "estimated_probability_of_malignancy": "greater_than_20_percent",
      "suspicious_feature_count": "3_or_more"
    }
  ],
  "applicability": {
    "use_for": "Structured risk categorization of an adnexal mass on a diagnostic pelvic ultrasound examination, primarily transvaginal ultrasound with Doppler as indicated.",
    "classification_unit": "one_adnexal_mass_with_each_mass_categorized_separately; category_1_is_an_exam_state_with_normal_ovaries_and_no_mass",
    "required_inputs": [
      "exam_adequacy_and_approach",
      "patient_age_and_menopausal_status",
      "side_and_organ_of_origin",
      "maximum_dimensions",
      "unilocular_multilocular_or_solid_morphology",
      "wall_and_septal_morphology",
      "papillary_projections",
      "solid_components",
      "ascites_or_peritoneal_findings",
      "color_or_power_Doppler_location",
      "classic_functional_or_benign_pattern"
    ],
    "outside_scope": [
      "CT_or_MRI_only_classification",
      "screening_without_an_adnexal_ultrasound_finding",
      "histologic_diagnosis",
      "FIGO_staging",
      "individualized_treatment_selection",
      "silent_conversion_to_O_RADS_or_IOTA_ADNEX"
    ]
  },
  "examination_and_adequacy_gate": {
    "expected_technique": "Use transvaginal ultrasound for detailed morphology, adding transabdominal views for a large or high lesion and color or power Doppler when vascularity is required.",
    "category_1_gate": "Assign category 1 only when normal ovaries are identified and no adnexal mass is seen; nonvisualization, incomplete coverage or a technically limited examination is not category 1.",
    "limited_exam_rule": "If the mass cannot be completely characterized or Doppler assessment of a decisive solid component is inadequate, return technically limited or indeterminate and obtain expert ultrasound or another supported pathway rather than force a number."
  },
  "suspicious_features_original": [
    "thick_papillary_projections",
    "thick_septations",
    "solid_areas",
    "ascites",
    "vascularization_within_solid_areas_or_papillary_projections_or_the_central_area_of_a_solid_tumor"
  ],
  "assignment_algorithm": [
    {
      "priority": 1,
      "if": "normal_ovaries_identified_AND_no_adnexal_mass_seen",
      "output_code": "1"
    },
    {
      "priority": 2,
      "if": "classic_functional_origin_finding",
      "output_code": "2"
    },
    {
      "priority": 3,
      "if": "classic_recognizable_benign_persistent_lesion_without_suspicious_features",
      "output_code": "3"
    },
    {
      "priority": 4,
      "if": "not_GI_RADS_1_to_3_AND_exactly_1_or_2_defined_suspicious_features",
      "output_code": "4"
    },
    {
      "priority": 5,
      "if": "at_least_3_defined_suspicious_features",
      "output_code": "5"
    },
    {
      "priority": 6,
      "if": "no_category_is_fully_supported_or_required_feature_is_unassessable",
      "output": "unclassified_and_expert_review_not_an_invented_GI_RADS_number"
    }
  ],
  "risk_band_and_validation_boundary": {
    "estimated_not_patient_specific": "The 0, less-than-1, 1-to-4, 5-to-20 and greater-than-20 percent values are the framework's estimated probability bands, not a personalized posterior probability.",
    "cohort_performance_rule": "Sensitivity, specificity and observed malignancy proportions depend on prevalence, referral setting, examiner expertise and verification method; do not export one validation cohort's rates to an individual.",
    "high_risk_binary_use": "Studies often combine GI-RADS 4 and 5 as high risk, but the individual categories and their distinct estimated bands must remain visible."
  },
  "management_boundary": {
    "original_2011_suggestions": {
      "1": "No mass-specific follow-up was suggested after a complete normal examination.",
      "2": "Follow-up ultrasound was suggested for a presumed functional lesion.",
      "3": "Laparoscopic surgery was suggested in the validation-era protocol for a presumed persistent benign lesion.",
      "4": "Referral to a gynecologic oncologist was suggested.",
      "5": "Referral to a gynecologic oncologist was suggested."
    },
    "current_context": "The original suggestions are not a universal contemporary management algorithm. Current decisions integrate symptoms, menopausal status, lesion size and growth, fertility goals, biomarkers when appropriate, local guidance and expert ultrasound; the 2021 multisociety consensus favors expert subjective assessment or validated IOTA models for triage.",
    "no_automatic_action": "No GI-RADS number alone orders surveillance interval, CT, MRI, tumor markers, biopsy, surgery or oncologic treatment."
  },
  "system_collision_boundary": {
    "O_RADS_US": "O-RADS US has different category definitions, risk bands and management. In particular, identically numbered categories are not interchangeable; name the system explicitly and never crosswalk by numeral.",
    "IOTA_ADNEX_and_Simple_Rules": "IOTA models use their own descriptors and probabilities. A calculated ADNEX probability or Simple Rules result may inform current triage but does not mathematically convert into a GI-RADS grade.",
    "biomarkers": "CA-125, HE4 and ROMA are separate clinical inputs and do not silently change the imaging category."
  },
  "agent_output_contract": [
    "Return examination adequacy, transvaginal and transabdominal components, Doppler adequacy, menopausal status and comparison date.",
    "Return one record per mass with side, organ of origin, dimensions and complete morphology.",
    "Return each defined suspicious feature as present, absent or unassessable before the category.",
    "Return the named GI-RADS system, category, original risk band and exact decisive pattern or feature count.",
    "Keep imaging category, individualized malignancy probability, pathology, stage and management as separate fields.",
    "If current management is requested, identify the current institutional pathway rather than treating the 2011 suggestion as a standing order."
  ],
  "missing_input_behavior": [
    "If neither ovary is adequately seen, do not call GI-RADS 1.",
    "If the lesion is large or incompletely visualized transvaginally, request complete transabdominal or other supported characterization before counting features.",
    "If a solid or papillary component lacks adequate Doppler assessment, preserve that uncertainty and do not lower the category from absent vascularity.",
    "If a lesion does not match a classic benign pattern and the suspicious-feature count is unresolved, return unclassified or a bounded differential rather than invent category 3 or 4.",
    "If the requested system is O-RADS or IOTA, do not answer with GI-RADS despite similar numbering or organ scope."
  ],
  "supporting_sources": [
    {
      "role": "original_proposal",
      "citation": "Amor et al. J Ultrasound Med. 2009;28:285-291",
      "doi": "10.7863/jum.2009.28.3.285"
    },
    {
      "role": "prospective_multicenter_validation",
      "citation": "Amor et al. Ultrasound Obstet Gynecol. 2011;38:450-455",
      "doi": "10.1002/uog.9012"
    },
    {
      "role": "current_multisociety_diagnostic_context",
      "citation": "Timmerman et al. Ultrasound Obstet Gynecol. 2021;58:148-168",
      "doi": "10.1002/uog.23635"
    }
  ]
}