P5-08-03: How Reader's Training, Software, and Image Formats Impact Percent Dense Area Measures.
Bibliographic record
Abstract
Abstract BACKGROUND: Mammographic percent dense area, the percent ratio of dense to total breast area in a mammogram, is one of the strongest measures of a woman's risk of breast cancer. However, systematic differences have been observed between readers and mammography technologies (film and digital) that could cause clinically inconsistent associations with risk. The purpose of this study was to evaluate inter- and intra-reproducibility of percent dense area between readers and between film and digital technologies. METHODS: One hundred digitized film mammograms were randomly selected with 25 films in each of quartile of percent density and read by two readers at two different sites (Mayo Clinic and UCSF). The readers had extensive experience and were also jointly trained at university of Toronto using Cumulus software. After training, all films were read twice with at least one year between duplicate readings. The Mayo clinic reading used Cumulus while UCSF used custom semiautomatic software to estimate total and dense tissue area. In addition, digitized films and unprocessed full field digital mammograms of the same women were assessed by one reader. The time between the film and digital acquisitions ranged from nine to twenty-four months. Interclass correlation coefficient (ICC) was calculated for each comparison. RESULTS: The intra- and inter-observer ICCs, consistency for film images, were 0.96 (UCSF) and 0.97 (Mayo), and 0.96 (UCSF vs. Mayo). We found ICC between film and digital mammograms for percent dense area was 0.88. The digital mammogram had 9% significantly higher total breast area and 5% significantly lower percent density area compared to film. CONCLUSIONS: Similarly trained readers had a high reproducibility regardless of the software used. Our results suggest centralized reader training should enable pooling of film breast density results from different clinics. However, pooling film and digital results would need careful calibration due to lower measured percent dense areas than on film. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P5-08-03.
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 | 0.066 | 0.218 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.007 | 0.003 |
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".