Qrater: collaborative imaging quality control tool
Bibliographic record
Abstract
Abstract Background Quality control (QC) is particularly important in vulnerable populations like patients with dementia where movement artifacts as well as registration failures may be more prevalent. From our need to review the quality of tens of thousands of raw MR images and reviewing the steps of our pre‐ and post‐processing pipelines, we developed Qrater. Qrater is a web‐based application installed centrally on a server where our data is stored. We can then rate the images and access the database of our team’s stored ratings (QC status and potential comments for each image). Method In Qrater the images are uploaded as datasets that can be restricted to specific users (Fig. 1). Each image is viewed and rated with available features such as a magnifying glass, seeing other users’ ratings, adding specifying comments, moving between images, and rating with key bindings to store QC outcomes other than pass/fail/warning (Fig. 2.). Looking for a specific image or rating is made easier by the search and sorting features (Fig. 3). Four experienced researchers rated three different datasets of MRI (A: 1275 unprocessed T1w brain images; B: 1623 unprocessed T1w brain images; C: 3000 preprocessed T1w brain images non‐linearly registered to a standard template). We looked at timing from the ratings’ timestamps to determine how much time it took to rate the datasets. All analyses were done in R v 4.1. Result On average, it took between 10 and 25 seconds to rate each image (14.2 s dataset A, 25.5 s dataset B and 14.1 s dataset C. Figure 1); less than 10 seconds to mark a clearly failed image for both raw acquisition and registration tasks, while the more questionable images (marked as Warning) took between one and two minutes (Table 1, Fig. 4). Conclusion Our team has found Qrater’s usability to be a significant improvement over previous quality control methods. The built‐in SQL database supports thousands of images without affecting its performance, and the web‐based interface enables remote QC. Since Qrater belongs to the initiative towards a more open and reproducible science, the code is currently available for download in GitHub [https://github.com/soffiafdz/Qrater].
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 | 0.023 | 0.083 |
| Meta-epidemiology (narrow) | 0.003 | 0.002 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.005 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.004 | 0.004 |
| Open science | 0.004 | 0.006 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.063 | 0.026 |
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".