BraMARS: An Interpretable Histopathology-Driven Deep Learning Model for Brain Metastasis Risk Stratification in Surgically Resected Limited-Stage SCLC.

Overview

1. Institute of Genomic Medicine School of Biomedical Engineering Wenzhou Medical University Wenzhou People's Republic of China
2. Department of Thoracic Radiation Oncology Harbin Medical University Cancer Hospital Harbin People's Republic of China
3. Department of Pathology National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital Chinese Academy of Medical Sciences and Peking Union Medical College Beijing People's Republic of China
4. AMG Nephrology Avera McKennan Hospital Sioux Falls South Dakota USA

Abstract

Brain metastasis (BM) is a major cause of mortality in limited‐stage small‐cell lung cancer (LS‐SCLC). Prophylactic cranial irradiation (PCI) reduces BM incidence but carries neurotoxicity and lacks individualized risk assessment. Here, we developed BraMARS, an explainable deep learning model that estimates future BM risk from routine H&E‐stained whole‐slide images of resected LS‐SCLC. BraMARS demonstrates robust discriminatory performance across independent cohorts, with AUCs ranging from 0.738 to 0.944, and stratifies patients into high‐risk and low‐risk groups with significantly different disease‐free survival, overall survival, and brain metastasis‐free survival. Retrospective simulation shows BraMARS‐guided risk stratification could reduce PCI exposure in 19.3% of low‐risk predicted patients while improving identification of high‐risk‐predicted patients by 84.4%. Histopathologic attribution and proteomic analyses linked higher scores to distinct tissue patterns and programs involving mitochondrial metabolism, reactive‐oxygen‐species detoxification, and DNA repair. Overall, BraMARS provides a biologically interpretable histopathology‐based framework for estimating subsequent BM risk in resected LS‐SCLC, with potential to support individualized intracranial risk assessment, intensified MRI surveillance, and hypothesis generation for prospective BM‐prevention strategies.

Reproduced under the paper's license (CC BY), from the paper cited above.

Repository

Its files are read in the Code ↔ Paper reader above, with 3 matches between paragraphs and lines of code.

Code Availability Statement

The code of this work is available at https://github.com/ZhoulabCPH/BraMARS.

Reproduced under the paper's license (CC BY), from the paper cited above.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

• 1 repository of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
• 7 scripts, each with its path and the digest of its content;
• 3 matches between paragraphs of the paper and lines of the code (method lexical-v1);
• neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

Datasets cited

• ngdc.cncb.ac.cn/omix — at ngdc.cncb.ac.cn; found in “Data Availability Statement”

Data Availability Statement

The whole‐slide histopathology images and associated clinical data generated and analyzed during this study are not publicly available due to patient privacy concerns and institutional data protection policies. These data may be made available to qualified researchers for non‐commercial research purposes upon reasonable request to the corresponding author, Dr. Lin Yang and Dr. Shilong Liu, subject to institutional data transfer agreement and ethical approval. The mass spectrometry proteomics data have been deposited in the OMIX database (https://ngdc.cncb.ac.cn/omix/) under accession code OMIX007230.

Reproduced under the paper's license (CC BY), from the paper cited above.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 1, 27 September 2026: the first record

Recorded: type, language, journal, pages, dates, 11 authors, 4 keywords, 2 funders, 41 references.

Cite

This paper

Yang, Z., Wang, T., Liu, S., Zhang, Z., Zhang, Y., Yang, F., Yu, B., Gao, S., Chen, Y., Yang, L., & Zhou, M. (2026). BraMARS: An Interpretable Histopathology-Driven Deep Learning Model for Brain Metastasis Risk Stratification in Surgically Resected Limited-Stage SCLC. Advanced science (Weinheim, Baden-Wurttemberg, Germany), e77075. https://doi.org/10.1002/advs.77075

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.1186/s11658-026-00957-1

Lactate-activated astrocytes promote NSCLC brain metastasis through extracellular vesicle-mediated miR-8085/TRIM67/ELK1 signaling axis.

Journal: Cellular & molecular biology letters

In common: ngdc.cncb.ac.cn/omix, other condition, 2 references

[2] doi:10.1002/cnr2.70625 [code]

Machine Learning-Based Prediction of Brain Metastasis at Initial Diagnosis in Small-Cell Lung Cancer: Model Development and SHAP Interpretation Study.

Journal: Cancer reports (Hoboken, N.J.)

In common: clinical / translational, other condition, 3 references

[3] doi:10.1038/s41598-026-61605-4 [code]

Learning precise segmentation of neurofibrillary tangles from rapid manual point annotations.

Journal: Scientific reports

In common: Pillow, PyTorch, scikit-learn, 2 other tools, histology / microscopy, other condition

[4] doi:10.1002/alz.71649 [code]

Postmortem brain MRI reveals differential associations of subcortical and limbic volumes with cortical thinning and neurodegenerative pathologies.

Journal: Alzheimer's & dementia : the journal of the Alzheimer's Association

In common: Pillow, PyTorch, scikit-learn, 2 other tools, histology / microscopy, other condition

[5] doi:10.1038/s41398-026-03965-z [code]

Disentangling individual heterogeneity reveals robust network and molecular signatures of major depressive disorder with suicidal ideation.

Journal: Translational psychiatry

In common: Pillow, PyTorch, scikit-learn, 2 other tools, 1 reference

[6] doi:10.1038/s41598-026-57519-w [code]

Automated segmentation of neurons and spinal cord structures in immunofluorescence images using SpineDL.

Journal: Scientific reports

In common: Pillow, PyTorch, scikit-learn, 2 other tools, histology / microscopy, other condition

[7] doi:10.1371/journal.pcbi.1013499 [code]

VesiclePy: A machine learning vesicle analysis toolbox for volume electron microscopy.

Journal: PLoS computational biology

In common: Pillow, PyTorch, scikit-learn, 2 other tools, histology / microscopy

[8] doi:10.1038/s41746-026-02651-0 [code]

AI Augmented Confocal Laser Endomicroscopy for Rapid Intraoperative Diagnosis of Brain Tumors.

Journal: NPJ digital medicine

In common: Pillow, PyTorch, pandas, 1 other tool, histology / microscopy, clinical / translational, other condition

[9] doi:10.1093/braincomms/fcag253 [code]

Disease detection and classification in temporal lobe epilepsy: step-wise versus simultaneous AI decision models in a multisite neuroimaging study.

Journal: Brain communications

In common: Pillow, PyTorch, scikit-learn, 2 other tools, clinical / translational

[10] doi:10.1093/jnen/nlaf152 [code]

Clinical and pathologic correlations of machine learning quantification of Aβ deposits across 3 brain regions of decedents with Alzheimer disease.

Journal: Journal of neuropathology and experimental neurology

In common: Pillow, PyTorch, scikit-learn, 2 other tools, clinical / translational

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.