A Deep Learning Framework for Classification of Neuroendocrine Neoplasm Whole Slide Images
Bibliographic record
Abstract
Background/Objectives: Neuroendocrine neoplasms (NENs) are uncommon neoplasms. Grading informs the prognosis and treatment decision of NENs and is determined by cell proliferation, which is measured by mitotic count and Ki-67 index. These measurements present challenges for pathologists as they suffer inter- and intra-observer variability and are cumbersome to quantify. To address these challenges, we developed a machine learning pipeline for identifying tumor areas, proliferating cells, and grading NENs. Methods: Our study includes 385 samples of gastroenteropancreatic NENs from across British Columbia with two stains (247 H&E and 138 Ki-67 images). Labels for these cases are at the patient-level, and there are 186 patients. We systematically investigated three settings for our study: H&E, H&E with Ki-67, and pathologist-reviewed and corrected cases. Results: Our H&E framework achieved a three-fold balanced accuracy of 77.5% in NEN grading. The H&E with Ki-67 framework yields a performance improvement to 83.0% on grading. We provide survival and multivariate analysis with a c-index of 0.65. Grade 1 NENs misclassified by the model were reviewed by a pathologist to assess reasons. Analysis of our AI-graded NENs for the subset of pathologist-assessed G1s demonstrated a significant (p-value = 0.007) survival difference amongst samples the algorithm assigned to a higher grade (n = 20; median survival 4.22 years) compared to concordant G1 samples (n = 60; median survival 10.13 years). Conclusions: Our model identifies NEN grades with high accuracy and identified some grade 1 tumors as prognostically unique, suggesting potential improvements to standard grading. Further studies are needed to determine if this discordant group is a different clinical entity.
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 | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| 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".