Learned reconstruction reduces perfusion-measurement bias in shortened cardiac SPECT data
Source check: Source checked. Not independent scientific confirmation.
Finding: A residual network reconstructed full-acquisition-like cardiac SPECT from reduced-count or reduced-projection data derived from 363 patient scans. Internal comparisons found closer perfusion-deficit measurements to the full acquisitions than the uncorrected shortened data. These are image-based clinical measurement results, without independently acquired short-scan validation, established diagnostic accuracy or demonstrated radiation-dose benefit.
Technology: Twenty-layer dilated residual convolutional neural networks for projection-domain SPECT reconstruction
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- The learned residual networks restored projection information before reconstruction, enabling the represented within-cohort test of clinical perfusion-measurement agreement under reduced data.
- Test and result
- A prospectively collected 363-person clinical cohort supplied full-acquisition list-mode data. Shorter-time data used the first half of acquisition events, while fewer-projection data omitted alternate views. Residual networks learned projection restoration; the source describes tenfold cross-validation on the cohort. Reconstructed clinical perfusion measures, including total perfusion deficit, were compared with full acquisitions using commercial quantitative software and agreement analyses. Learned reconstructions reduced measured bias relative to uncorrected reduced data.
- Limitation / still unknown
- Reduced acquisitions were derived from the same full scans, rather than separate real-world short scans. Full-acquisition reconstruction is a measurement reference, not independent disease truth. Detailed fold-specific preprocessing/tuning isolation is not established. Gated functional performance, diagnostic decisions, patient outcomes and actual lower-dose acquisition were not tested. Some headline image-metric labels and dispersion values are inconsistent, so they are not adopted as precise comparative claims.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://www.hug.ch/sites/interhug/files/structures/pinlab/documents/JNC2021_SPECT.pdf
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 21 Nov 2026. Review scope: Read authenticated primary institution-deposited PDF front DOI/title, original clinical acquisition and reduced-data derivation, ResNet/training/cross-validation and quantitative clinical evaluation Methods, perfusion Results and limits. Exact publisher online April 28, 2020 distinguished from December 2021 issue. Supplements were not read. Exact original article front DOI/title and fresh PubMed primary relationship metadata checked..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.