{"paper":{"title":"Dual-Correction Physics-Informed Neural Networks for Hemodynamic Reconstruction from Sparse Data","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A dual-correction physics-informed neural network reconstructs accurate blood flow fields in tortuous intracranial arteries from sparse data.","cross_cats":[],"primary_cat":"physics.med-ph","authors_text":"Hao Wu, Jingtai Song, Qinsheng Zhu, Xianwen Zhang, Xiaodong Xing, Yufeng Tang, Zhiyun Zhang","submitted_at":"2026-05-09T09:32:51Z","abstract_excerpt":"Quantifying hemodynamics in the curved segments of the intracranial internal carotid artery is a core challenge in diagnosing vascular stenosis. Conventional full-field imaging, such as 4D Flow MRI, is costly and difficult to widely promote. Meanwhile, reconstructing full-field fluid information from easily accessible and non-invasive sparse measurement data (such as transcranial Doppler ultrasound/computed tomography angiography) is essentially a highly challenging ill-posed inverse problem. To overcome the severe optimization difficulties and generalization failures of conventional physics-i"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"The proposed DCP-INN model utilizes a diamond-shaped main network to capture low-frequency trends in physical evolution, and employs a parallel wide-deep correction network to compensate for high-frequency residuals resulting from complex geometric shapes... The results demonstrate that the method effectively mitigates optimization challenges and significantly reduces flow field reconstruction error.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the causal decoupling strategy, dual-network architecture, and Taylor-based high-order loss will reliably overcome severe optimization difficulties and generalization failures of standard PINNs specifically in highly tortuous intracranial geometries under extremely sparse data constraints.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"DCP-INN combines a diamond-shaped main network for low-frequency flow trends with a parallel correction network for high-frequency residuals, plus a Taylor-expansion high-order loss, to reconstruct hemodynamics accurately from sparse data in tortuous vessels.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A dual-correction physics-informed neural network reconstructs accurate blood flow fields in tortuous intracranial arteries from sparse data.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"cdc312db3985d27df97fb455d7c16779eb2293400e2239f8c5471fe4d2b584e3"},"source":{"id":"2605.12544","kind":"arxiv","version":1},"verdict":{"id":"9154bf5d-3f48-4e6f-9aec-b8a00dd44bad","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T21:55:30.612821Z","strongest_claim":"The proposed DCP-INN model utilizes a diamond-shaped main network to capture low-frequency trends in physical evolution, and employs a parallel wide-deep correction network to compensate for high-frequency residuals resulting from complex geometric shapes... The results demonstrate that the method effectively mitigates optimization challenges and significantly reduces flow field reconstruction error.","one_line_summary":"DCP-INN combines a diamond-shaped main network for low-frequency flow trends with a parallel correction network for high-frequency residuals, plus a Taylor-expansion high-order loss, to reconstruct hemodynamics accurately from sparse data in tortuous vessels.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the causal decoupling strategy, dual-network architecture, and Taylor-based high-order loss will reliably overcome severe optimization difficulties and generalization failures of standard PINNs specifically in highly tortuous intracranial geometries under extremely sparse data constraints.","pith_extraction_headline":"A dual-correction physics-informed neural network reconstructs accurate blood flow fields in tortuous intracranial arteries from sparse data."},"references":{"count":40,"sample":[{"doi":"","year":2009,"title":"A review of 3D vessel lumen segmentation techniques: models, features and extraction schemes,","work_id":"090c54ef-0b35-4fc2-9c36-55df2665be26","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2018,"title":"Blood vessel segmentation algorithms—Review of methods, datasets and evaluation metrics,","work_id":"b03e9804-2a71-4bdb-9fe5-f215a14f6c9b","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Fluid-structure interaction analysis of pulsatile flow in arterial aneurysms with physics-informed neural networks and computational fluid dynamics,","work_id":"4b4c5beb-30d6-442b-8449-da752978e565","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Automated generation of 0D and 1D reduced-order models of patient-specific blood flow,","work_id":"35a18734-2031-4c52-9c1e-9935c3f4cc51","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2018,"title":"Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations,","work_id":"7f6a4445-c2cb-4455-8d22-f65bc9439b97","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":40,"snapshot_sha256":"df1b7ef9c9c153afafa3d275057d47945a48dffe621f6a0dcf2ab48ea4d954d4","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"77f312a6fc01a6aeffad773b00789ab34969f8157b8881cc89732c4b6909331b"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}