Improved injury detection through harmonizing multi-site neuroimaging data after experimental TBI: a Translational Outcomes Project in Neurotrauma consortium study
Bibliographic record
Abstract
Multi-site neuroimaging studies have become increasingly common in order to generate larger samples of reproducible data to answer questions associated with smaller effect sizes. The data harmonization model NeuroCombat has been shown to remove site effects introduced by differences in site-related technical variance while maintaining group differences, yet its effect on improving statistical power in pre-clinical models of CNS disease is unclear. The present study examined fractional anisotropy data computed from diffusion weighted imaging data at 3 and 30 days post-controlled cortical impact injury from 184 adult rats across four sites as part of the Translational-Outcome-Project-in-Neurotrauma (TOP-NT) Consortium. Findings supported prior clinical reports that NeuroCombat fails to remove site effects in data containing a high proportion-of-outliers (>5%) and skewness, which introduced significant variation in non-outlier sites. After removal of one outlier site and harmonization using a pooled sham population, the data displayed an increase in effect size and group level effects (p < 0.01) in both univariate and voxel-level volumes of pathology. This was characterized by movement toward similar distributions in voxel measurements (Kolmogorov–Smirnov p < <0.001 to >0.01) and statistical power increases within the ipsilateral cortex. Harmonization improved statistical power and frequency of significant differences in areas with existing group differences, thus improving the ability to detect regions affected by injury rather than by other confounds. These findings indicate the utility of NeuroCombat in reproducible data collection, where biological differences can be accurately revealed to allow for greater reliability in multi-site neuroimaging studies.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.022 | 0.017 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.003 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".