Classify the CT pattern, then reason about etiology and diagnosis. Pattern, diagnosis, prognosis, sampling and treatment are separate agent outputs.
{
"categories": [
{
"outcome_code": "uip",
"distribution": "subpleural_and_basal_predominant_often_heterogeneous",
"required_fibrotic_feature": "honeycombing_with_or_without_peripheral_traction_bronchiectasis_or_bronchiolectasis",
"ground_glass": "may_be_mild_but_not_predominant",
"alternative_features": "absent"
},
{
"outcome_code": "probable-uip",
"distribution": "subpleural_and_basal_predominant_often_heterogeneous",
"required_fibrotic_feature": "reticulation_with_peripheral_traction_bronchiectasis_or_bronchiolectasis",
"honeycombing": "absent",
"ground_glass": "may_be_mild_but_not_predominant",
"alternative_features": "absent"
},
{
"outcome_code": "indeterminate",
"distribution": "often_subpleural_and_basal_but_not_required",
"required_fibrotic_feature": "fibrosis_that_meets_neither_UIP_nor_probable_UIP",
"included_patterns": [
"subtle_reticulation_with_or_without_mild_ground_glass_or_distortion_sometimes_called_early_UIP",
"fibrosis_without_a_pattern_specific_for_any_etiology"
],
"alternative_features": "absent"
},
{
"outcome_code": "alternative",
"trigger": "one_or_more_features_or_distributions_that_suggest_a_diagnosis_other_than_IPF_or_UIP_pattern",
"meaning": "suggestive_of_an_alternative_diagnosis_not_proof_of_one_specific_disease"
}
],
"applicability": {
"use_for": "Morphologic classification of fibrotic interstitial lung disease on one diagnostic-quality adult high-resolution chest CT when UIP is in the differential.",
"classification_unit": "one_HRCT_examination_at_one_timepoint_with_the_background_fibrotic_pattern_separated_from_any_superimposed_acute_process",
"required_inputs": [
"thin_section_inspiratory_HRCT",
"distribution_craniocaudal_and_axial",
"honeycombing",
"traction_bronchiectasis_or_bronchiolectasis",
"reticulation_and_architectural_distortion",
"ground_glass_relative_to_fibrosis",
"mosaic_attenuation_and_air_trapping",
"nodules_cysts_and_consolidation",
"pleural_and_extrapulmonary_clues",
"clinical_exposure_drug_and_connective_tissue_context_when_available",
"comparison_date"
],
"outside_scope": [
"diagnosing_IPF_from_CT_morphology_alone",
"classifying_nonfibrotic_ILD",
"grading_disease_severity_or_progression",
"predicting_individual_survival",
"selecting_antifibrotic_or_immunosuppressive_treatment",
"ordering_or_forbidding_biopsy_without_multidisciplinary_risk_assessment"
]
},
"acquisition_and_adequacy_gate": {
"core_protocol": "Use volumetric thin-section images at full inspiration with high-spatial-frequency reconstruction and multiplanar review. Inadequate inspiration can mimic ground-glass opacity and dependent fibrosis.",
"expiratory_role": "Expiratory imaging is used when air trapping is clinically relevant, especially when fibrotic hypersensitivity pneumonitis is a competing diagnosis.",
"prone_role": "Prone inspiratory images are optional but important when subtle dependent subpleural opacity could be atelectasis or pseudohoneycombing; persistence away from the dependent surface supports true abnormality.",
"technical_failure_rule": "If motion, poor inspiration, thick sections, dependent opacity without prone clarification, or incomplete coverage prevents confident evaluation of a decisive feature, return technically limited or a bounded differential rather than force one of the four patterns."
},
"decisive_feature_definitions": {
"honeycombing": "Clustered, usually subpleural cystic airspaces with well-defined walls and variable but typically similar diameters, in single or multiple layers; distinguish from traction bronchiolectasis, paraseptal emphysema and dependent pseudohoneycombing.",
"traction_bronchiectasis_or_bronchiolectasis": "Irregular airway dilatation caused by surrounding fibrotic retraction; peripheral bronchiolectasis is sufficient for the probable-UIP feature when the distribution and other criteria are met.",
"ground_glass_guard": "Mild ground-glass opacity can accompany UIP or probable UIP. Predominant or extensive ground-glass opacity relative to reticulation suggests an alternative or an acute superimposed process, depending on context.",
"heterogeneity": "Patchwork geographic variation between more and less affected lung supports UIP morphology; asymmetry does not by itself exclude UIP."
},
"assignment_algorithm": [
{
"priority": 1,
"if": "technical_quality_does_not_allow_decisive_feature_or_distribution_assessment",
"output": "ungradable_or_bounded_pattern_not_a_false_indeterminate"
},
{
"priority": 2,
"if": "a_defined_alternative_distribution_lung_feature_or_extrapulmonary_clue_is_dominant",
"output_code": "alternative"
},
{
"priority": 3,
"if": "subpleural_basal_often_heterogeneous_fibrosis_AND_honeycombing_present_AND_no_alternative_feature",
"output_code": "uip"
},
{
"priority": 4,
"if": "subpleural_basal_often_heterogeneous_reticulation_AND_peripheral_traction_bronchiectasis_or_bronchiolectasis_AND_no_honeycombing_AND_no_alternative_feature",
"output_code": "probable-uip"
},
{
"priority": 5,
"if": "fibrosis_present_but_UIP_and_probable_UIP_not_met_AND_no_specific_alternative_pattern",
"output_code": "indeterminate"
}
],
"alternative_diagnosis_feature_map": {
"distributions": [
"peribronchovascular_predominance",
"perilymphatic_predominance",
"upper_or_mid_lung_predominance",
"subpleural_sparing"
],
"lung_features": [
"cysts_not_explained_by_honeycombing_or_emphysema",
"marked_mosaic_attenuation_or_three_density_pattern",
"extensive_air_trapping",
"predominant_ground_glass_opacity",
"profuse_micronodules",
"centrilobular_nodules",
"other_nodules",
"consolidation"
],
"extrapulmonary_clues": [
"pleural_plaques_consider_asbestos_exposure",
"dilated_esophagus_consider_connective_tissue_disease",
"distal_clavicular_erosions_consider_rheumatoid_arthritis",
"extensive_lymphadenopathy_consider_other_etiologies",
"pleural_effusion_or_thickening_consider_connective_tissue_disease_or_drug_reaction"
],
"examples_not_diagnoses": [
"fibrotic_hypersensitivity_pneumonitis",
"connective_tissue_disease_related_ILD",
"fibrotic_NSIP",
"sarcoidosis",
"asbestosis_or_other_exposure_related_ILD",
"drug_toxicity",
"pleuroparenchymal_fibroelastosis"
]
},
"morphology_etiology_and_diagnosis_boundary": {
"UIP_is_not_IPF": "UIP is a radiologic and histopathologic pattern. It can occur in IPF, fibrotic hypersensitivity pneumonitis, connective-tissue-disease ILD, exposure-related disease and other settings; the word idiopathic requires exclusion of a known cause.",
"required_clinical_exclusions": [
"medication_and_radiation_history",
"occupational_environmental_and_avocational_exposures",
"connective_tissue_disease_assessment",
"familial_pulmonary_fibrosis_context",
"other_known_causes_of_ILD"
],
"MDD_rule": "Integrate pulmonology, radiology and, when tissue exists, pathology in multidisciplinary discussion when the diagnosis is not already secure. Preserve both the HRCT pattern and the final multidisciplinary diagnosis as separate fields.",
"negative_guard": "Probable, indeterminate or alternative HRCT patterns do not exclude histopathologic UIP or IPF; UIP morphology does not exclude a secondary cause."
},
"diagnostic_procedure_boundary_2018_to_2022": {
"UIP_pattern": "The 2018 guideline recommended against surgical, transbronchial and cryobiopsy solely to confirm UIP in a clinically suspected IPF case with HRCT UIP, and suggested against BAL. This applies only after the clinical diagnostic frame and alternatives have been evaluated.",
"probable_UIP_update": "The 2022 update treats probable UIP more like UIP in the diagnostic algorithm: in an appropriate clinical setting, a confident IPF diagnosis can be reached after multidisciplinary discussion without a lung biopsy. This is permission for case-specific diagnosis, not a universal no-biopsy rule.",
"indeterminate_or_alternative": "Further evaluation can include exposure and CTD work-up, BAL, biopsy or disease-specific testing, but no CT category automatically mandates a procedure.",
"cryobiopsy_update": "The 2022 guideline conditionally accepts transbronchial lung cryobiopsy as an alternative to surgical lung biopsy in centers with appropriate expertise for selected undetermined ILD; patient risk, sampling limitations and center capability remain explicit."
},
"acute_and_longitudinal_context": {
"acute_overlay": "New bilateral ground-glass opacity or consolidation on a fibrotic UIP background can represent acute exacerbation but infection, edema, aspiration, hemorrhage and drug toxicity must be evaluated; classify the background separately from the acute overlay.",
"progression": "Do not infer progressive pulmonary fibrosis from one category. Compare extent, traction, honeycombing and new fibrotic features over time and integrate symptoms and physiology under the applicable PPF definition.",
"prognosis": "Pattern category is not a calibrated personal prognosis. Fibrotic extent and change, FVC, DLCO, oxygen need, comorbidities, acute exacerbations and the final ILD diagnosis carry separate prognostic information."
},
"management_boundary": {
"no_pattern_only_treatment": "No HRCT pattern independently starts nintedanib, pirfenidone, immunosuppression, oxygen, transplant referral or surveillance, and no category alone withholds them.",
"appropriate_use": "Use the category to communicate diagnostic confidence, focus the etiologic work-up and inform multidisciplinary decisions about whether additional sampling would change care.",
"current_terminology_context": "The 2025 ERS/ATS interstitial-pneumonia statement expands pattern classification beyond idiopathic disease and emphasizes diagnostic confidence; it does not convert the four IPF HRCT categories into etiologic diagnoses."
},
"agent_output_contract": [
"Return technical adequacy, inspiratory/expiratory/prone acquisitions and comparison date.",
"Return each decisive finding: distribution, honeycombing, traction bronchiectasis or bronchiolectasis, reticulation, ground-glass burden, mosaic or air trapping, nodules, cysts, consolidation and extrapulmonary clues.",
"Return exactly one supported HRCT pattern or explicitly ungradable/uncertain, with the decisive positive and absent features.",
"Keep HRCT pattern, suspected etiology, multidisciplinary diagnosis, disease extent/progression and management as separate fields.",
"State whether an acute superimposed process is present and whether an alternative diagnosis is suggested.",
"Never turn UIP into automatic IPF, probable UIP into automatic biopsy, or any category into treatment or prognosis."
],
"missing_input_behavior": [
"If honeycombing versus traction bronchiolectasis is unresolved, return UIP versus probable UIP with the disputed feature and do not invent certainty.",
"If subtle dependent basal opacity lacks adequate inspiration or prone confirmation, return technically limited rather than early or indeterminate UIP.",
"If exposure, drug or connective-tissue history is unavailable, classify morphology but do not label it idiopathic.",
"If expiratory imaging is absent, do not claim absence of air trapping when hypersensitivity pneumonitis remains plausible.",
"If no prior HRCT or physiology is available, do not state stability, progression or prognosis."
],
"supporting_sources": [
{
"role": "four_pattern_primary_guideline",
"citation": "Raghu et al. AJRCCM. 2018;198:e44-e68",
"doi": "10.1164/rccm.201807-1255ST"
},
{
"role": "diagnostic_and_cryobiopsy_update",
"citation": "Raghu et al. AJRCCM. 2022;205:e18-e47",
"doi": "10.1164/rccm.202202-0399ST",
"pmcid": "PMC9851481"
},
{
"role": "current_multidisciplinary_pattern_context",
"citation": "Ryerson et al. Eur Respir J. 2025;66:2500158",
"doi": "10.1183/13993003.00158-2025",
"pmid": "40774805"
}
],
"source_locator": "Raghu et al. 2018 official full-length guideline, DOI 10.1164/rccm.201807-1255ST, Table 3 acquisition and Table 4 HRCT patterns; Raghu et al. 2022, PMC9851481, updated diagnostic algorithm and TBLC recommendation; Ryerson et al. 2025, DOI 10.1183/13993003.00158-2025, multidisciplinary pattern and confidence context."
}