{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:AQFSF5FC3ZUTKB7WOI2K6YZ2FE","short_pith_number":"pith:AQFSF5FC","schema_version":"1.0","canonical_sha256":"040b22f4a2de693507f67234af633a291e671a423e86a29f69b9186b508a17e6","source":{"kind":"arxiv","id":"2502.07516","version":2},"attestation_state":"computed","paper":{"title":"The Devil is in the Prompts: De-Identification Traces Enhance Memorization Risks in Synthetic Chest X-Ray Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CV","cs.LG"],"primary_cat":"eess.IV","authors_text":"Raman Dutt","submitted_at":"2025-02-11T12:36:00Z","abstract_excerpt":"Generative models, particularly text-to-image (T2I) diffusion models, play a crucial role in medical image analysis. However, these models are prone to training data memorization, posing significant risks to patient privacy. Synthetic chest X-ray generation is one of the most common applications in medical image analysis with the MIMIC-CXR dataset serving as the primary data repository for this task. This study presents the first systematic attempt to identify prompts and text tokens in MIMIC-CXR that contribute the most to training data memorization. Our analysis reveals two unexpected findin"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2502.07516","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"eess.IV","submitted_at":"2025-02-11T12:36:00Z","cross_cats_sorted":["cs.AI","cs.CV","cs.LG"],"title_canon_sha256":"aec22e17c348bfc85c35f9d257d3561d3be2bdc69b38c03a0d6cb6e40ad04ec5","abstract_canon_sha256":"4dcad510da5ac930a046db119634c3b2639984c552806ad0a1fd13414cdbea4b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:14:26.186354Z","signature_b64":"p1WM+hDBExeM8taXakyVyeaAmiuNqqN/fMMIaPNKx6DJGP7EfZEvRQvnyCa6VK10ATLGc+gMazf7/pJbsTAiCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"040b22f4a2de693507f67234af633a291e671a423e86a29f69b9186b508a17e6","last_reissued_at":"2026-07-05T10:14:26.185093Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:14:26.185093Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Devil is in the Prompts: De-Identification Traces Enhance Memorization Risks in Synthetic Chest X-Ray Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CV","cs.LG"],"primary_cat":"eess.IV","authors_text":"Raman Dutt","submitted_at":"2025-02-11T12:36:00Z","abstract_excerpt":"Generative models, particularly text-to-image (T2I) diffusion models, play a crucial role in medical image analysis. However, these models are prone to training data memorization, posing significant risks to patient privacy. Synthetic chest X-ray generation is one of the most common applications in medical image analysis with the MIMIC-CXR dataset serving as the primary data repository for this task. This study presents the first systematic attempt to identify prompts and text tokens in MIMIC-CXR that contribute the most to training data memorization. Our analysis reveals two unexpected findin"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.07516","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2502.07516/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2502.07516","created_at":"2026-07-05T10:14:26.185860+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.07516v2","created_at":"2026-07-05T10:14:26.185860+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.07516","created_at":"2026-07-05T10:14:26.185860+00:00"},{"alias_kind":"pith_short_12","alias_value":"AQFSF5FC3ZUT","created_at":"2026-07-05T10:14:26.185860+00:00"},{"alias_kind":"pith_short_16","alias_value":"AQFSF5FC3ZUTKB7W","created_at":"2026-07-05T10:14:26.185860+00:00"},{"alias_kind":"pith_short_8","alias_value":"AQFSF5FC","created_at":"2026-07-05T10:14:26.185860+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2603.13419","citing_title":"Diffusion Models Memorize in Training -- and Generalize in Inference","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2604.07128","citing_title":"A Utility-preserving De-identification Pipeline for Cross-hospital Radiology Data Sharing","ref_index":14,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AQFSF5FC3ZUTKB7WOI2K6YZ2FE","json":"https://pith.science/pith/AQFSF5FC3ZUTKB7WOI2K6YZ2FE.json","graph_json":"https://pith.science/api/pith-number/AQFSF5FC3ZUTKB7WOI2K6YZ2FE/graph.json","events_json":"https://pith.science/api/pith-number/AQFSF5FC3ZUTKB7WOI2K6YZ2FE/events.json","paper":"https://pith.science/paper/AQFSF5FC"},"agent_actions":{"view_html":"https://pith.science/pith/AQFSF5FC3ZUTKB7WOI2K6YZ2FE","download_json":"https://pith.science/pith/AQFSF5FC3ZUTKB7WOI2K6YZ2FE.json","view_paper":"https://pith.science/paper/AQFSF5FC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.07516&json=true","fetch_graph":"https://pith.science/api/pith-number/AQFSF5FC3ZUTKB7WOI2K6YZ2FE/graph.json","fetch_events":"https://pith.science/api/pith-number/AQFSF5FC3ZUTKB7WOI2K6YZ2FE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AQFSF5FC3ZUTKB7WOI2K6YZ2FE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AQFSF5FC3ZUTKB7WOI2K6YZ2FE/action/storage_attestation","attest_author":"https://pith.science/pith/AQFSF5FC3ZUTKB7WOI2K6YZ2FE/action/author_attestation","sign_citation":"https://pith.science/pith/AQFSF5FC3ZUTKB7WOI2K6YZ2FE/action/citation_signature","submit_replication":"https://pith.science/pith/AQFSF5FC3ZUTKB7WOI2K6YZ2FE/action/replication_record"}},"created_at":"2026-07-05T10:14:26.185860+00:00","updated_at":"2026-07-05T10:14:26.185860+00:00"}