PATH-08. Establishing the utility of deep-learning for H&E-based meningioma molecular classification and outcome prediction
Bibliographic record
Abstract
Abstract The introduction of genomic profiling as a tool for molecular classification and outcome prediction has revolutionized the care of patients with brain tumors. Artificial intelligence (AI) provides advanced avenues to convert complex genomic information into routinely available patient-level information. In this study, we leverage deep learning to demonstrate that H&E can robustly characterize the molecular subtypes of the most common brain tumor, meningioma. To do this, we created a cohort of 605 meningiomas with paired DNA methylation and matched digitized H&E images. We trained and validated dedicated deep learning models to predict molecular groups, relevant chromosomal arm aneuploidies (1p loss, 1q gain, 22q loss), and clinical outcomes using the H&E alone. AUROCs and balanced accuracy were used to assess classifier performance and risk-group-specific outcomes were compared using the log-rank test. Our deep learning classifier achieved a balanced accuracy of 87-97% for predicting molecular groups of meningiomas. Similar performance was achieved in the prediction of chromosomal aneuploidies. Our dedicated outcome prediction model was remarkably prognostic even when adjusting for WHO grade, extent of resection, and age (HR 3.49, 95% CI 1.54-7.91, p = 0.003), demonstrating the clear and immediate translational value of deep learning in this context. Beyond this direct translational utility, the generated AI models also provide novel insights into group-specific molecular heterogeneity which have not been detected using bulk genomic approaches to date. This work is the first to demonstrate the ability to apply deep learning models that can, beyond diagnosis, determine molecular subtypes and predict outcomes in a single brain tumour entity using H&E alone, which to date is only possible using resource-intensive genomic profiling. It further demonstrates the broad clinical utility of applying AI modelling to readily available H&E to democratize access to genomic information globally.
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 | 0.003 | 0.006 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| 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".