Categorizing segmentation quality using a quantitative quality assurance algorithm
Bibliographic record
Abstract
INTRODUCTION: Obtaining high levels of contouring consistency is a major limiting step in optimizing the radiotherapeutic ratio. We describe a novel quantitative methodology for the quality assurance (QA) of contour compliance referenced against a community set of contouring experts. METHODS: Two clinical tumour site scenarios (10 lung cases and one prostate case) were used with QA algorithm. For each case, multiple physicians (lung: n = 6, prostate: n = 25) segmented various target/organ at risk (OAR) structures to define a set of community reference contours. For each set of community contours, a consensus contour (Simultaneous Truth and Performance Level Estimation (STAPLE)) was created. Differences between each individual community contour versus the group consensus contour were quantified by consensus-based contouring penalty metric (PM) scores. New observers segmented these same cases to calculate individual PM scores (for each unique target/OAR) for each new observer-STAPLE pair for comparison against the community and consensus contours. RESULTS: Four physicians contoured the 10 lung cases for a total of 72 contours for quality assurance evaluation against the previously derived community consensus contours. A total of 16 outlier contours were identified by the QA system of which 11 outliers were due to over-contouring discrepancies, three were due to over-/under-contouring discrepancies, and two were due to missing/incorrect nodal contours. In the prostate scenario involving six physicians, the QA system detected a missing penile bulb contour, systematic inner-bladder contouring, and under-contouring of the upper/anterior rectum. CONCLUSION: A practical methodology for QA has been demonstrated with future clinical trial credentialing, medical education and auto-contouring assessment applications.
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 | 0.020 | 0.045 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.007 | 0.003 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.003 | 0.002 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.002 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
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".