Julius Hense

Julius Hense

Auto-generated profile, based on publicly available information

Julius Hense

Berlin Institute for the Foundations of Learning and Data (BIFOLD), Berlin, Germany

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
15 March 2024 in Communications Medicine
Verified
Journal Article
Open access
PDF file
SASalim Arslan
Salim Arslan
JSJulian Schmidt
Julian Schmidt
CBCher Bass
Cher Bass
DMDebapriya Mehrotra
Debapriya Mehrotra
AGAndre Geraldes
Andre Geraldes
PPPahini Pandya
Pahini Pandya
Background: The objective of this comprehensive pan-cancer study is to evaluate the potential of deep learning (DL) for molecular profiling of multi-omic biomarkers directly from hematoxylin and eosin (H&E)-stained whole slide images. Methods: A total of 12,093 DL models predicting 4031 multi-omic biomarkers across 32 cancer types were trained and validated. The study included a broad range of genetic, transcriptomic, and proteomic biomarkers, as well as established prognostic markers, molecular subtypes, and clinical outcomes. Results: Here we show that 50% of the models achieve an area under the curve (AUC) of 0.644 or higher. The observed AUC for 25% of the models is at least 0.719 and exceeds 0.834 for the top 5%. Molecular profiling with image-based histomorphological features is generally considered feasible for most of the investigated biomarkers and across different cancer types. The performance appears to be independent of tumor purity, sample size, and class ratio (prevalence), suggesting a degree of inherent predictability in histomorphology. Conclusions: The results demonstrate that DL holds promise to predict a wide range of biomarkers across the omics spectrum using only H&E-stained histological slides of solid tumors. This paves the way for accelerating diagnosis and developing more precise treatments for cancer patients.

24 October 2024 in NEJM AI
Verified
Journal Article
Closed access
JDJonas Dippel
Jonas Dippel
NPNiklas Prenißl
Niklas Prenißl
JHJulius Hense
Julius Hense
PLPhilipp Liznerski
Philipp Liznerski
TWTobias Winterhoff
Tobias Winterhoff
FKFrederick Klauschen
Frederick Klauschen

11 June 2026 in Nature Communications
Verified
Journal Article
Open access
PDF file
JKJonah Kömen
Jonah Kömen
EJEdwin D. de Jong
Edwin D. de Jong
JHJulius Hense
Julius Hense
HMHannah Marienwald
Hannah Marienwald
JDJonas Dippel
Jonas Dippel
KMKlaus-Robert Müller
Klaus-Robert Müller
Biomedical Foundation Models (FMs) are transforming AI-enabled healthcare research and entering clinical validation. However, their susceptibility to learning non-biological features — including variations in laboratory procedures and scanner hardware — poses risks for clinical deployment. We introduce PathoROB, a public benchmark quantifying FM robustness to non-biological features. Representation-level robustness is assessed using the robustness index, while output-level robustness is evaluated across clinically relevant settings, including patch- and slide-level prediction, case retrieval, and clustering tasks. Our experiments reveal robustness deficits across all 20 evaluated FMs, with substantial differences between them. We find that non-robust FM representations can cause major diagnostic downstream errors preventing safe clinical adoption. Using more robust FMs, vision-language alignment, and post-hoc robustification reduces (but does not yet eliminate) the risk of such errors. This work establishes that robustness evaluation is essential for validating pathology FMs before clinical adoption and provides a blueprint for assessing and improving robustness across biomedical domains.

3 June 2026 in Medical Image Analysis
Verified
Journal Article
Open access
MIMina Jamshidi Idaji
Mina Jamshidi Idaji
JHJulius Hense
Julius Hense
TNTom Neuhäuser
Tom Neuhäuser
AKAugustin Krause
Augustin Krause
YLYanqing Luo
Yanqing Luo
KMKlaus-Robert Müller
Klaus-Robert Müller
Multiple instance learning (MIL) has enabled substantial progress in computational histopathology, where a large amount of patches from gigapixel whole slide images are aggregated into slide-level predictions. Heatmaps are widely used to validate MIL models and to discover tissue biomarkers. Yet, the validity of these heatmaps has barely been investigated. In this work, we introduce a general framework for evaluating the quality of MIL heatmaps without requiring additional labels. We conduct a large-scale benchmark experiment to assess six explanation methods across histopathology task types (classification, regression, survival), MIL model architectures (Attention-, Transformer-, Mamba-based), and patch encoder backbones (UNI2, Virchow2). Our results show that explanation quality mostly depends on MIL model architecture and task type, with perturbation ("Single"), layer-wise relevance propagation (LRP), and integrated gradients (IG) consistently outperforming attention-based and gradient-based saliency heatmaps, which often fail to reflect model decision mechanisms. We further demonstrate the advanced capabilities of the best-performing explanation methods: (i) We provide a proof-of-concept that MIL heatmaps of a bulk gene expression prediction model can be correlated with spatial transcriptomics for biological validation, and (ii) showcase the discovery of distinct model strategies for predicting human papillomavirus (HPV) infection from head and neck cancer slides. Our work highlights the importance of validating MIL heatmaps and establishes that improved explainability can enable more reliable model validation and yield biological insights, making a case for a broader adoption of explainable AI in digital pathology. Our code is provided in a public GitHub repository: https://github.com/bifold-pathomics/xMIL.

26 November 2025 in bioRxiv
Verified
Preprint
Open access
JHJulius Hense
Julius Hense
MIMina Jamshidi Idaji
Mina Jamshidi Idaji
LCLaure Ciernik
Laure Ciernik
JDJonas Dippel
Jonas Dippel
FEFatma Ersan
Fatma Ersan
AMAndreas Mock
Andreas Mock
Head and neck squamous cell carcinoma (HNSCC) is a morphologically and molecularly heterogeneous disease with limited effectiveness of genotype-informed therapies. Transcriptome-derived estimates of signaling pathway activity carry prognostic and therapeutic potential but remain inaccessible in routine diagnostics due to cost and tissue constraints. Here, we introduce Digital Spatial Pathway Mapping , an AI-based computational pathology framework that infers signaling pathway activities directly from routine hematoxylin and eosin (H&E) slides, enabling in-silico spatial molecular readouts from standard histology. Models trained on HPV-negative HNSCC from TCGA and externally validated on CPTAC robustly predicted transcriptome-derived activities in cancer-relevant signaling pathways. To achieve spatial interpretability, we applied layer-wise relevance propagation (LRP) to generate heatmaps that highlight positive versus negative evidence for pathway activation. These LRP heatmaps were technically validated by patch-flipping tests and biologically validated against pathway-relevant immunohistochemistry in an independent patient cohort. From these explanations, we derived a tumor area pathway activity score (TAPAS) quantifying the spatial fraction of activated tumor regions within a slide. Applied to a retrospective HNSCC cohort of 1,066 slides from 112 resection specimens, TAPAS captured intratumoral heterogeneity and revealed two biologically dis-tinct tumor states - an oncogenic growth phenotype with widespread pathway activation and a pathway quiescent phenotype associated with higher recurrence risk independent of clinicopathological variables. These findings establish Digital Spatial Pathway Mapping as a scalable, in-silico approach to recover systems-level molecular information from standard histopathology, enabling prognostic and mechanistically grounded patient stratification in head and neck cancer.

1 October 2026 in Medical Image Analysis
Verified
Journal Article
Open access
AMAlexander Möllers
Alexander Möllers
JHJulius Hense
Julius Hense
FSFlorian Schulz
Florian Schulz
TMTimo Milbich
Timo Milbich
MAMaximilian Alber
Maximilian Alber
LRLukas Ruff
Lukas Ruff