Concerns regarding 'Increasing functional connectivity of the anterior cingulate cortex during the course of recovery from Bell's palsy' (Hu et al., NeuroReport, 2015)

Verdict

Highly suspicious. Three independent concerns are flagged, the first of which involves physically implausible MRI parameters that strongly suggest fabrication or careless copy-pasting of methods text. Final determination requires institutional investigation and inspection of raw DICOM headers and data.

Key findings

• Implausible/identical MRI parameters across sequences: The paper reports the same TR/TE (2000 ms / 35 ms) for the 2D structural, EPI_BOLD, and 3D T1-weighted structural scans. A TE of 35 ms is incompatible with T1-weighted structural imaging (typical TE ~2–4 ms) and would produce severe T2*-related signal loss.
• Cross-site, cross-platform data pooling without harmonization: Healthy-control task-fMRI was acquired in Korea on a Philips Achieva 3.0 T, while patient resting-state data were acquired in China on a 1.5 T Siemens scanner. Seed regions derived from one cohort were applied to the other without cross-site calibration, introducing substantial systematic bias.
• Suspicious clustering of correlation values in Figure 3: Nine reported Pearson r values (0.59, 0.66, 0.69, 0.59, 0.56, 0.58, 0.54, 0.54, 0.54) cluster tightly around 0.54–0.69, with three regions showing identical r = 0.54 and two showing r = 0.59. Such tight clustering is atypical for noisy inter-regional fMRI correlations.

Evidence highlights

• DOI: 10.1097/WNR.0000000000000295
• Journal: NeuroReport (2015)
• Source PDF: "Increasing functional connectivity of the anterior cingulate cortex during the course of recovery from Bell's palsy.pdf"
• Location of Finding 1: Materials and methods / Image acquisition (healthy-volunteer 3.0 T scan parameters). Quoted parameter triplet — (a) 2D structural: TR/TE 2000/35 ms; (b) EPI_BOLD: TR/TE 2000/35 ms; (c) T1*-weighted 3D structural: TR/TE 2000/35 ms.
• Location of Finding 2: Materials and methods / Participants and Experimental design. 37 healthy controls from Kyung Hee University, Korea (3.0 T Philips Achieva) vs. 67 patients from a hospital in Anhui, China (1.5 T Siemens).
• Location of Finding 3: Figure 3 (Page 10). Reported r values across nine regions: 0.59, 0.66, 0.69, 0.59, 0.56, 0.58, 0.54, 0.54, 0.54. No pixel-level analysis was possible from the available text.

Notes

• All three findings are based on textual inspection; pixel-level or forensic image analysis was not performed.
• The MRI-parameter concern (Finding 1) is, on its face, a hard inconsistency with MR physics and strongly suggests method-section fabrication or unedited copy-paste.
• The cross-site pooling (Finding 2) is a methodological weakness that, while not necessarily fraudulent, severely undermines the validity of the seed-to-voxel functional connectivity analysis and should be addressed in any corrigendum or retraction consideration.
• The correlation-value clustering (Finding 3) is suggestive but not conclusive; alternative explanations (small ROIs, smoothing, partial-volume effects) cannot be ruled out without raw data.
• Recommended actions: request DICOM headers and raw k-space/data from the authors; verify cross-national ethics approval for secondary use of Korean and Chinese datasets; raise the parameter issue on PubPeer; notify the NeuroReport editorial office.
• This AI-assisted report is for academic discussion only and does not constitute a formal finding of misconduct.