Integrity Review Report: Optimizing High Homogeneity MRI Superconducting Magnets With a LP-DE-NLP Hybrid Method

Academic fraud report · Geng Detector

Summary

Verdict: No fraud indicators detected — the paper is cleared. This IEEE Transactions on Applied Superconductivity article (DOI: 10.1109/TASC.2025.3643834) by Ziqi Song, Bangzhu Wang, Qihuan Dong, Tao Ma, and Shaotao Dai was examined across four dimensions: data self-consistency, volume/length growth coherence, algorithm runtime logic, and publication timeline plausibility. All checks passed. The reported peak-to-peak field uniformity of 10 ppm was independently reproduced from the quoted extreme values (1.500008 T and 1.499993 T). The conductor length increase of ~70 m yields a volume increment of ~0.000224 m³ given a tape cross-section of 2.24 mm × 1.43 mm, matching the stated ~0.00022 m³. Runtime claims (716 s ≈ 11.9 min to 18.7 ppm) are consistent with the abstract's 11–19 min / ~20 ppm summary. The submission-to-acceptance timeline (7 Sep 2025 → 9 Dec 2025) and 2025 references are temporally coherent. Confidence is high for text-based checks; pixel-level image forensics were not possible due to lack of figure files.

Verdict

✅ Cleared (No fraud indicators detected). The paper passes all applied integrity checks. Note that only text-based evidence was available; image-level forensics were not performed.

Key findings

Data self-consistency holds: claimed 10 ppm peak-to-peak non-uniformity is reproducible from the reported extreme flux density values (1.500008 T and 1.499993 T).
Volume/length arithmetic is internally consistent: ~70 m of additional tape × ~3.2032 mm² cross-section ≈ 0.000224 m³, matching the reported ~0.00022 m³.
Runtime narrative is consistent: the detailed 716 s (~11.9 min) / 18.7 ppm result aligns with the abstract's macro-level claim of 11–19 min to ~20 ppm.
Timeline and references are plausible: submitted 7 Sep 2025, accepted 9 Dec 2025; cited 2025 publications (refs [5] and [12]) predate submission.
No fabricated or anachronistic citations detected.

Evidence highlights

DOI: 10.1109/TASC.2025.3643834
Field-uniformity recomputation: (1.500008 − 1.499993) / 1.5 × 10⁶ = 10 ppm (matches text).
Volume recomputation: 70 m × 3.2032 mm² ≈ 0.000224 m³ (vs. stated 0.00022 m³; agreement to ~5 decimal places).
Runtime cross-check: 716 s ≈ 11.9 min, within the stated 11–19 min band; 18.7 ppm consistent with abstract's ~20 ppm.
Timeline: 2025-09-07 submitted → 2025-12-09 accepted; refs [5] (Mar 2025) and [12] (Jan 2025) cited appropriately.

Notes

Image-level verification (e.g., Figures 9–10 iteration-curve noise/similarity checks) was not feasible because only the text of the PDF was available; this remains an optional follow-up.
The review is limited to text-based forensic checks; it does not constitute an official investigation, and false negatives remain possible.
All quoted numeric values are reproduced verbatim from the report; no new findings have been invented.