Summary: Researchers introduced a deep-learning artificial intelligence capable of predicting the molecular classification of brain and spinal cord tumors in minutes using standard, universally available tissue sections. The AI system, named “Hetairos”, was developed by an international coalition.
Trained on a massive global dataset of over 11,000 digitized histological sections from 9,606 patients across four continents, Hetairos bypasses the complex, expensive, and time-consuming gold-standard DNA methylation testing timeline, slashing the diagnostic window from twelve days down to just twelve minutes while vastly outperforming senior human neuropathologists in microscopic classification accuracy.
Key Facts
- The DNA Methylation Bottleneck: Modern neuro-oncology requires DNA methylation analysis (mapping chemical modifications on DNA) to accurately classify highly diverse central nervous system (CNS) tumors. However, this gold-standard testing requires specialized laboratories, heavy tissue volumes, and massive financial budgets, routinely taking up to two weeks to generate resultsโa technological barrier completely unavailable in low-resource regions of the world.
- The 11,000-Slide Global Training Grid: Hetairos was built and validated utilizing a massive, demographically diverse library of over 11,000 digitized tissue sections from 9,606 international patients across eleven medical centers on four continents. The ground-truth diagnoses for this training grid were pre-determined using advanced molecular DNA methylation diagnostics.
- Comprehensive WHO Cataloging: The AI does not merely detect cancer; it successfully distinguishes between 102 distinct molecular tumor subtypes, covering almost the entire diagnostic spectrum of the current World Health Organization (WHO) classification for central nervous system tumors.
- Outperforming Senior Neuropathologists: In a head-to-head clinical trial involving 210 complex tumor cases, five international expert neuropathologists were pitted against the AI, using nothing but stained tissue sections. Hetairos achieved a definitive diagnostic accuracy rate of 68%, while the human specialists averaged just 30%. When evaluating the top three most likely diagnoses, the AI scored 84%, doubling the human specialists’ 50% average.
- Twelve Minutes vs. Twelve Days: In prospective clinical testing run parallel to routine hospital practices, traditional full-scale molecular diagnostics required an average of twelve days to return a report. Hetairos generated its precise molecular subtyping predictions in just twelve minutes using standard, off-the-shelf computer hardware once the routine slide was digitized.
- The “High-Certainty” Filter: Hetairos features a self-evaluating confidence metric. In 50% to 70% of all evaluated cases, the AI flag flagged its own predictions with a “high degree of certainty.” Within this filtered tier, its diagnostic accuracy spiked to an exceptional 87% to 88%. Even in highly ambiguous cases, it successfully narrows down a field of over 100 subtypes to a few likely candidates.
- Explainable AI Tissue Mapping: Bypassing the classic “black box” critique of neural networks, Hetairos actively highlights the exact microscopic regions on the tissue slide that drove its computational decision. This feature allows pathologists to visually verify the AI’s logic, isolate clear boundaries for targeted genomic testing, and proceed with treatment planning within 24 to 48 hours of initial biopsy.
Source: DKFZ
Tumors of the brain and spinal cord are extremely diverse. In recent years, it has become clear that many of these tumors can only be reliably diagnosed if their molecular properties are examined in addition to their microscopic appearance. Of particular importance here is so-calledย DNA methylation analysis, which is now considered the gold standard for the accurate classification of many brain tumors.
However, such tests are complex: they require specialized laboratories, expensive equipment, and sufficient tumor material. In addition, it often takes about two weeks for the results to become available. In many regions of the world, the necessary technologies are not even available.
AI learns from over 11,000 tissue sections
A new AI system called โHetairosโ is expected to bring about substantial improvements. It was developed by a team led by Moritz Gerstung (German Cancer Research Center, DKFZ) and Felix Sahm (Heidelberg Medical Faculty of Heidelberg University and Heidelberg University Hospital). The goal of the project was to predict which molecular subgroup a tumor belongs to based solely on routinely prepared and stained histological sections.
Hetairos was trained and validated using more than 11,000 digitized tissue sections from 9,606 patients. The diagnoses were primarily determined using DNA methylation diagnostics. The data came from eleven medical centers on four continents. In total, Hetairos distinguishes 102 different molecular tumor subtypes, covering nearly the entire spectrum of the current WHO classification of central nervous system tumors.
The AI not only evaluates its diagnosis but also indicates how confident it is in it. In approximately 50 to 70 percent of all cases, Hetairos made predictions with a high degree of certainty. In these cases, accuracy was around 87 to 88 percent. Even when the AI was uncertain, it was usually able to significantly narrow down the number of possible diagnoses.
Instead of having to distinguish between more than a hundred tumor subtypes, Hetairos often provides neuropathologists with only a few likely candidates. This can significantly simplify the selection of further diagnostic tests.
โThe study shows that artificial intelligence is capable of deriving molecular information directly from routine tissue sections and thus fundamentally changing cancer diagnostics,โ said Darui Jin, one of the lead authors of the study.
Hetairos outperforms experienced specialists
Particularly noteworthy was the direct comparison with human experts. Five experienced neuropathologists from various international centers were given 210 cases and asked to make a diagnosis based solely on the tissue sections. Hetairos achieved an accuracy rate of 68 percent, while the specialists averaged 30 percent. When considering the three most likely diagnoses in each case, the AI scored 84 percent, while the specialists scored about 50 percent.
โThe results show that modern AI systems are now capable of recognizing extremely subtle morphological patterns that are difficult even for experienced specialists to distinguish,โ says Felix Sahm.
โCurrently, the diagnosis of very rare tumor types still poses a major challenge for Hetairos; in this regard, experienced neuropathologists appear to be at least on par. However, we expect the systemโs performance to improve even further with larger and more diverse datasets,โ adds Moritz Gerstung.
Diagnosis in twelve minutes instead of twelve days
In a prospective study, Hetairos was used in parallel with routine clinical practice. The system analyzed 210 tumor samples without the AI result influencing the actual diagnosis or treatment decision.
While complete molecular diagnostics took an average of about twelve days, Hetairos generated its findings in just twelve minutes on standard computer hardware after digitizing the stained tissue sections. Including preparation and digitization of the tissue sections, results could often be available within 24 hours to two days.
Assistance with difficult and unclear cases
Hetairos could be particularly valuable in situations where traditional molecular methods reach their limits, when there is insufficient tumor material for genetic testing, or when molecular tests do not yield clear results. In addition, the system highlights the areas in the tissue section that were particularly important for its decision. This allows doctors to understand the basis of the AIโs diagnosis and identify which regions may be suitable for further investigation.
โWe developed Hetairos primarily as a tool to support diagnostics,โ explains neuropathologist Felix Sahm. โIt is not intended to replace molecular analyses, but rather to specifically complement and accelerate them. The technology could make an important contribution, particularly in countries or regions with limited resources, as it is based on standard tissue sections used worldwide.โ
The method could also offer economic advantages. While a DNA methylation analysis typically costs several hundred euros, Hetairos uses existing tissue sections for its analysis.
Moritz Gerstung confirms: โHetairos demonstrates the enormous potential of AI-supported digital pathology to provide rapid and widely available diagnostic methods that were previously only possible with considerable technical effort.โ
Key Questions Answered:
A: By recognizing incredibly complex, microscopic structural patterns that are invisible to the naked human eye. When tumors mutate genetically, those molecular changes subtly alter how the cells cluster, shape, and bind together. By training on over 11,000 global tissue samples where the molecular data was already known, Hetairos learned to map these ultra-subtle physical shapes back to their specific genetic subtypes, extracting deep molecular data straight from a routine visual scan.
A: No, it is engineered to serve as an advanced, high-velocity diagnostic assistant. As co-developer Dr. Felix Sahm emphasizes, Hetairos is not a replacement for traditional molecular testing or human oversight; it is a complementary tool. While it crushes human averages on massive multi-layered patterns, human neuropathologists remain completely equal or superior when diagnosing exceptionally rare, unique tumor anomalies. The AI acts as a digital copilot, rapidly clearing easy cases and narrowing down complex ones so doctors can work faster and with greater accuracy.
A: Saving critical, life-altering time. When dealing with aggressive brain or spinal cord cancers, waiting twelve days for complex molecular results before a medical team can tailor an exact treatment plan can be incredibly dangerous. Hetairos delivers highly accurate molecular subtyping predictions in just twelve minutes, allowing neuro-oncologists to safely determine the exact character of the tumor and initiate highly targeted, life-saving therapies within 24 to 48 hours of the initial surgical procedure.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- Journal paper reviewed in full.
- Additional context added by our staff.
About this brain cancer and AI research news
Author:ย Sibylle Kohlstรคdt
Source:ย DKFZ
Contact:ย Sibylle Kohlstรคdt โ DKFZ
Image:ย The image is credited to Neuroscience News
Original Research:ย Open access.
โHetairos is a histology-based artificial intelligence model for predicting central nervous system tumor methylation subtypesโ by Darui Jin (ๆ่พพ็ฟ), Artem Shmatko, Areeba Patel, Samuel Rutz, Lukas Friedrich, Rouzbeh Banan, Ramin Rahmanzade, Philipp Sievers, Stefan Hamelmann, Daniel Schrimpf, Kirsten Gรถbel, Henri Bogumil, Sybren L. N. Maas, Martin Sill, Felix E. Hinz, Abigail K. Suwala, Felix Keller, Antje Habel, Gleb Rukhovich, Ferdinand Zettl, Obada T. Alhalabi, Sebastian Ille, Jannik Sehring, Daniel Amsel, Benedikt Wiestler, Pedro Piovesan Lago, Bogdana Suchorska, Olfat Ahmad, Dominik Sturm, David Reuss, Pieter Wesseling, Adelheid Wรถhrer, Frank L. Heppner, Ingmar Blรผmcke, Claire Delbridge, Martin Jakobs, Christel Herold-Mende, Sandro M. Krieg, Wolfgang Wick, David T. W. Jones, Stefan M. Pfister, Maysa Al-Hussaini, Yanghao Hou, Felipe DโAlmeida Costa, Leonille Schweizer, Luca Bertero, Till Acker, Arnault Tauziede-Espariat, Pascale Varlet, Doron Merkler, Kristof Egervari, Hildegard Dohmen, Pablo Zoroquiain, Roger Gejman, Sebastian Brandner, Xiangzhi Bai, Andreas von Deimling, Felix Sahm & Moritz Gerstung.ย Nature Cancer
DOI:10.1038/s43018-026-01186-3
Abstract
Hetairos is a histology-based artificial intelligence model for predicting central nervous system tumor methylation subtypes
Molecular testing is essential for classifying central nervous system (CNS) tumors, with methylation profiling providing the highest diagnostic granularity. However, this requires more resources and time than conventional hematoxylin and eosin (H&E) histopathology, which is widely available globally.
Here we propose Hetairos, an artificial intelligence algorithm that predicts 102 methylation-based CNS tumor subtypes from digital H&E slides. Built and validated on 9,606 patients and over 11,000 slides from 11 centers across four continents, Hetairos identified 50โ70% of cases with high confidence, achieving an accuracy of 0.87 for its highest-rated predictions.
Hetairos outperformed five board-certified neuropathologists in a direct histology-only comparison (0.68 versus 0.30). Prospective evaluation in routine diagnostics confirmed its performance, reducing turnaround time from 12 days (molecular testing) to 12โmin.
Hetairos supports diagnostic decision-making across the full spectrum of pediatric and adult CNS tumors by narrowing differential diagnoses and guiding efficient testing.
