Loading
Life sciences · Preprint
arXiv · September 9, 2026
Early or partial results. Treat as a signal, not a conclusion.
This preprint describes a machine-learning framework for synthesizing volumetric CT images from 2D chest X-rays using a multi-stage domain adaptation and refinement pipeline. Performance is reported only on the LIDC-IDRI public dataset using image-similarity metrics (PSNR and SSIM), with no clinical validation, independent testing, or comparison to standard clinical CT reconstruction or radiologist assessment.
Uncontrolled computational method development and validation on public dataset. LIDC-IDRI public dataset (paired DRR-CT data); no real patient cohort details provided.. Intervention: Multi-Pass Multi-View Blended Learning framework: Stage 1 unsupervised CXR-to-DRR domain adaptation, Stage 2a supervised DRR-to-CT transformation, Stage 2b unsupervised multi-view slice refinement, Stage 2c progressive transfer learning.. Compared with: Prior methods (unspecified by name in abstract); no comparison to standard clinical CT reconstruction pipelines or human radiologist performance..
Proposed method achieves up to 14% improvement in PSNR over prior methods on LIDC-IDRI dataset Proposed method achieves up to 7.6% improvement in SSIM over prior methods on LIDC-IDRI dataset Framework combines unsupervised domain adaptation, supervised transformation, multi-view refinement, and progressive transfer learning in two stages
Safety was not reported in the material analysed. Check the source before drawing any conclusion about harm.
This computational development has not been validated clinically and uses only image-quality metrics rather than diagnostic accuracy or clinical utility. Clinicians should not rely on this method for clinical CT reconstruction until independent peer review, clinical validation, and comparison to standard reconstruction methods are published.
This is a preprint describing an unvalidated computational method for CT synthesis from chest X-rays, tested only on a public dataset with surrogate image-quality metrics, without clinical validation or comparison to clinical reconstruction standards.
As stated by the source record.
Quoted from the source exactly as published.
This computational development has not been validated clinically and uses only image-quality metrics rather than diagnostic accuracy or clinical utility. Clinicians should not rely on this method for clinical CT reconstruction until independent peer review, clinical validation, and comparison to standard reconstruction methods are published.
Graded across the dimensions that decide whether you should act, each from what the source actually supports. There is no single score, and where a dimension was not assessed it says so.
Reconstructing volumetric Computed Tomography (CT) from a single 2D chest radiograph (CXR) is an ill-posed inverse problem, further complicated by the scarcity of paired CXR-CT training data. Prior approaches address this by training on Digitally Reconstructed Radiographs (DRRs), which are synthetic projections derived from CT volumes. However, the domain gap between DRRs and real CXRs limits generalization, often resulting in coarse or anatomically inconsistent reconstructions when applied to clinical images. To address this challenging problem, this study introduces a Multi-Pass Multi-View Blended Learning framework for synthesizing high-fidelity volumetric CT directly from real chest X-ray (CXR) images. The proposed approach progressively decomposes the synthesis task into two distinct, complementary learning stages. Stage 1 is an unsupervised CXR-to-DRR Domain Adaptation, while Stage 2 includes three passes, namely, (a) supervised DRR-to-CT Transformation, (b) unsupervised Multi-View Slice Refinement, followed by (c) Progressive Transfer Learning (PTL). With such a blended learning paradigm, the proposed approach mitigates the synthetic-to-real domain gap while enhancing both the structural integrity and anatomical detail of the final output. On the LIDC-IDRI dataset, where paired DRR-CT ground truth is available for quantitative evaluation, the proposed method improves upon prior methods by up to 14% in PSNR and 7.6% in SSIM. The framework successfully generates structurally consistent and anatomically realistic high-fidelity CT volumes from real CXRs, marking a significant advancement toward clinical viability of CT reconstruction from standard radiographic images.
Taken from the source record, never inferred. Follow any of these and new work involving them reaches your briefing.