{"paper":{"title":"A General B\\'ezier Tree Encoding Counterfactual Framework for Retinal-Vessel-Mediated Disease Analysis","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Encoding retinal vessels as interconnected cubic-Bézier segments allows isolated geometric interventions that shift disease classifier predictions in a dose-dependent manner.","cross_cats":["cs.CV","cs.LG"],"primary_cat":"eess.IV","authors_text":"Ethan Elio Meidinger, Lin Gu, Ruogu Fang, Tan Su","submitted_at":"2026-05-13T05:10:20Z","abstract_excerpt":"The geometry of the retinal vessel is a key biomarker of vascular diseases, yet clinical evidence remains primarily observational. Existing generative counterfactuals intervene only at the image-level disease label, failing to isolate explicit anatomical structure. To address this limitation, we propose the B\\'ezier Tree Encoding Counterfactual Framework (BTECF). By abstracting vascular networks into interconnected cubic-B\\'ezier segments, BTECF establishes a disease-agnostic representation in which structural topology is explicitly preserved and atomically perturbable. Coupling this encoding "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Isolated counterfactual interventions produce dose-responsive shifts in classifier predictions; a matched pixel-drop control attenuates this response by an order of magnitude or more, ruling out out-of-distribution generation artifacts.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That abstracting vascular networks into interconnected cubic-Bézier segments preserves all disease-relevant anatomical structure and enables truly atomic perturbations without introducing biases or artifacts that affect downstream classifiers.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"BTECF encodes retinal vessels as Bézier trees to enable targeted, parameter-level counterfactual interventions on vessel geometry for causal analysis of vascular diseases.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Encoding retinal vessels as interconnected cubic-Bézier segments allows isolated geometric interventions that shift disease classifier predictions in a dose-dependent manner.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"71bfec377808c8ce21ef35bdf533991ff3275192b437544b7e15b48da5cfe4b6"},"source":{"id":"2605.13015","kind":"arxiv","version":1},"verdict":{"id":"368928b2-e0fb-47f3-a691-ef55f4044008","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T18:57:42.817723Z","strongest_claim":"Isolated counterfactual interventions produce dose-responsive shifts in classifier predictions; a matched pixel-drop control attenuates this response by an order of magnitude or more, ruling out out-of-distribution generation artifacts.","one_line_summary":"BTECF encodes retinal vessels as Bézier trees to enable targeted, parameter-level counterfactual interventions on vessel geometry for causal analysis of vascular diseases.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That abstracting vascular networks into interconnected cubic-Bézier segments preserves all disease-relevant anatomical structure and enables truly atomic perturbations without introducing biases or artifacts that affect downstream classifiers.","pith_extraction_headline":"Encoding retinal vessels as interconnected cubic-Bézier segments allows isolated geometric interventions that shift disease classifier predictions in a dose-dependent manner."},"references":{"count":62,"sample":[{"doi":"","year":1994,"title":"Vascular endothelial growth factor in ocular fluid of patients with diabetic retinopathy and other retinal disorders.New England Journal of Medicine, 331(22):1480–1487, 1994","work_id":"f52b3a10-a61b-46b3-a887-cd6ca45d42d7","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1056/nejmra1005073","year":2012,"title":"Antonetti, Ronald Klein, and Thomas W","work_id":"041e2ea9-a1a3-47e5-876f-273aae52e6fc","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Louis Arnould, Fabrice Meriaudeau, Charles Guenancia, Clement Germanese, Cécile Delcourt, Ryo Kawasaki, Carol Y Cheung, Catherine Creuzot-Garcher, and Andrzej Grzybowski. Using artificial intelligence","work_id":"e989d767-65eb-4434-bbe8-9829aa08426a","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2021,"title":"Retinal vascular caliber association with nonperfusion and diabetic retinopathy severity depends on vascular caliber measurement location.Ophthalmology Retina, 5(6):571–579, 2021","work_id":"041a532c-e8ef-49cc-8803-22c56a54c942","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1972,"title":"Numerical control-mathematics and applications.Translated by AR Forrest, 1972","work_id":"9d788a2e-0597-4db8-ba78-01a07c963f12","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":62,"snapshot_sha256":"64dff0ba22fd74325f1efbe5cdb347b50d2ce389b2bf5251502323842d5e31ba","internal_anchors":3},"formal_canon":{"evidence_count":2,"snapshot_sha256":"f9232324bc134b88c429288959636de5f434f92c3bd068bec988d72cf2c94b76"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}