{"paper":{"title":"Progressive $\\mathcal{J}$-Invariant Self-supervised Learning for Low-Dose CT Denoising","license":"http://creativecommons.org/licenses/by/4.0/","headline":"A progressive J-invariant self-supervised method achieves low-dose CT denoising performance comparable to supervised approaches without needing paired normal-dose images.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Junwen Guo, Yichao Liu, YueYang Teng, Zongru Shao","submitted_at":"2026-01-20T17:35:02Z","abstract_excerpt":"Self-supervised learning has been increasingly investigated for low-dose computed tomography (LDCT) image denoising, as it alleviates the dependence on paired normal-dose CT (NDCT) data, which are often difficult to collect. However, many existing self-supervised blind-spot denoising methods suffer from training inefficiencies and suboptimal performance due to restricted receptive fields. To mitigate this issue, we propose a novel Progressive $\\mathcal{J}$-invariant Learning that maximizes the use of $\\mathcal{J}$-invariant to enhance LDCT denoising performance. We introduce a step-wise blind-"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Extensive experiments on the Mayo LDCT dataset demonstrate that the proposed method consistently outperforms existing self-supervised approaches and achieves performance comparable to, or better than, several representative supervised denoising methods.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The assumption that the step-wise blind-spot mechanism with progressive conditional independence enforcement, combined with controlled Gaussian and Poisson noise injection, will reliably improve denoising without introducing new artifacts or bias on real clinical data.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A progressive J-invariant self-supervised learning framework for low-dose CT denoising outperforms prior self-supervised methods and matches some supervised ones on the Mayo dataset.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A progressive J-invariant self-supervised method achieves low-dose CT denoising performance comparable to supervised approaches without needing paired normal-dose images.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"378ac5c87ca5bea3cd86b735941c51d56a5774be3092dac97f99b2d55e8db5c0"},"source":{"id":"2601.14180","kind":"arxiv","version":4},"verdict":{"id":"40113890-c124-40c7-8d77-514395d879f8","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T12:24:26.435904Z","strongest_claim":"Extensive experiments on the Mayo LDCT dataset demonstrate that the proposed method consistently outperforms existing self-supervised approaches and achieves performance comparable to, or better than, several representative supervised denoising methods.","one_line_summary":"A progressive J-invariant self-supervised learning framework for low-dose CT denoising outperforms prior self-supervised methods and matches some supervised ones on the Mayo dataset.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The assumption that the step-wise blind-spot mechanism with progressive conditional independence enforcement, combined with controlled Gaussian and Poisson noise injection, will reliably improve denoising without introducing new artifacts or bias on real clinical data.","pith_extraction_headline":"A progressive J-invariant self-supervised method achieves low-dose CT denoising performance comparable to supervised approaches without needing paired normal-dose images."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2601.14180/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":2,"snapshot_sha256":"46e2817f7010e3ee955f2a6708f03d121d41690a5d180109357e5e4c1d54474f"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}