Poster — Thur Eve — 47: Automatic Comparison of Portal Images for the Detection of Radiotherapy Treatment Delivery Errors
Bibliographic record
Abstract
The goal of this project is to develop a real‐time automatic comparison tool to detect portal dose image errors and has two logical parts: 1. A mechanism to detect the presence of a significant discrepancy. 2. A set of components to identify the probable cause(s) of the discrepancy including geometrical positioning errors as well as machine specific delivery errors. Initially, we have only considered in‐plane rotations and translations. A total of ∼700 portal images of an anthropomorphic pelvic phantom were obtained using a Varian aS1000 EPID. The images contained combinations of three types of geometrical patient set‐up errors. The angle of rotation (about the axis perpendicular to the EPID) was varied from [−5, +5] degrees and the translational distance was varied from [−10, +10] mm. The majority of the images were acquired from the AP direction however ∼130 images were of a lateral view. The system was found to be sensitive to in‐plane rotations down to ∼ 2°, out of plane rotations greater than ∼4° and displacements 3mm. For the lateral views, the calculated rotation was correct to within 1/2° and translations to within 5mm 53% of the time. Considering only translations, the accuracy increases to ∼82%. The results are much better for the AP views. The in‐plane rotation was within ±1/2deg; for all cases and the displacement error was greater than 1.5 mm in only 1 case. The system was able to analyze an image pair in ∼10s, however, no effort was put toward optimization of the system.
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 | 0.002 | 0.004 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.015 | 0.005 |
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".