Deep learning-based approach for automated assessment of PTEN status.
Bibliographic record
Abstract
294 Background: PTEN loss is associated with adverse outcomes in prostate cancer and has the potential to be clinically implemented as a prognostic biomarker. Deep learning algorithms applied to digital pathology can provide automated and objective assessment of biomarkers. The objective of this work was to develop an artificial intelligence (AI) system for automated detection and localization of PTEN loss in prostate cancer samples. Methods: Immunohistochemistry (IHC) was used to measure PTEN protein levels on prostate tissue microarrays (TMA) from two institutions (in-house n=272 and external n=125 patients). TMA cores were visually scored for PTEN loss by pathologists and, if present, spatially annotated. In-house cohort (N=1239 cores) were divided into 70/20/10 training/validation/testing sets. Two algorithms were developed: a) Class I=core-based, to label each core for biomarker status and b) Class II=pixel-based, to spatially distinguish areas of PTEN loss within each core. ResNet101 architecture was used to train a multi-resolution ensemble of classifiers at 5x, 10x, and 20x for Class I task and a single classifier at simulated 40x for Class II segmentation. Results: For Class I algorithm, accuracy of PTEN status was 88.3% and 93.4% in validation and testing cohorts, respectively (Table). AI-based probability of PTEN loss was higher in cores with complete loss vs partial loss. Accuracy was improved to 90.7% in validation and 93.5% in test cohorts using the Class II region-based algorithm, with median dice scores 0.833 and 0.831, respectively. Direct application to external set demonstrated a high false positive rate. Loading trained model and conservatively re-training (“fine-tuning”) on 48/320 external cohort cores improved accuracy to 93.4%. Conclusions: Results demonstrate feasibility and robustness for fully automated detection and localization of PTEN loss in prostate cancer tissue samples and possibility for time/cost-effectiveness of sample processing/scoring in research and clinical laboratories.[Table: see text]
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 | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 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) | 0.003 | 0.002 |
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".