Automated PD-L1 Scoring for Non-Small Cell Lung Carcinoma Using Open-Source Software
Bibliographic record
Abstract
PD-L1 expression in non-small cell lung cancer (NSCLC) is predictive of response to immunotherapy, but scoring of PD-L1 immunohistochemistry shows considerable interobserver variability. Automated methods may allow more consistent and expedient PD-L1 scoring. We aimed to assess the technical concordance of PD-L1 scores produced using free open source QuPath software with the manual scores of three pathologists. A classifier for PD-L1 scoring was trained using 30 NSCLC image patches. A separate test set of 207 image patches from 69 NSCLC resection cases was used for comparison of automated and manual scores. Automated and average manual scores showed excellent correlation (concordance correlation coeffecient = 0.925), though automated scoring resulted in significantly more 1–49% scores than manual scoring ( p = 0.012). At both 1% and 50% thresholds, automated scores showed a level of concordance with our ‘gold standard’ (the average of three pathologists’ manual scores) similar to that of individual pathologists. Automated scoring showed high sensitivity (95%) but lower specificity (84%) at a 1% threshold, and excellent specificity (100%) but lower sensitivity (71%) at a 50% threshold. We conclude that our automated PD-L1 scoring system for NSCLC has an accuracy similar to that of individual pathologists. The detailed protocol we provide for free open source scoring software and our discussion of the limitations of this technology may facilitate more effective integration of automated scoring into clinical workflows.
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 | 0.006 | 0.018 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.004 | 0.002 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.002 | 0.001 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.009 | 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".