https://arxiv.org/api/LZYaXYTwS4tJlVMjpqBpQaMb0DA 2026-03-20T22:57:44Z 2597 150 15 http://arxiv.org/abs/2503.03778v3 2025-07-31T19:40:54Z

Designing generative models for 3D structural brain MRI that synthesize morphologically-plausible and attribute-specific (e.g., age, sex, disease state) samples is an active area of research. Existing approaches based on frameworks like GANs or diffusion models synthesize the image directly, which may limit their ability to capture intricate morphological details. In this work, we propose a 3D brain MRI generation method based on state-of-the-art latent diffusion models (LDMs), called MorphLDM, that generates novel images by applying synthesized deformation fields to a learned template. Instead of using a reconstruction-based autoencoder (as in a typical LDM), our encoder outputs a latent embedding derived from both an image and a learned template that is itself the output of a template decoder; this latent is passed to a deformation field decoder, whose output is applied to the learned template. A registration loss is minimized between the original image and the deformed template with respect to the encoder and both decoders. Empirically, our approach outperforms generative baselines on metrics spanning image diversity, adherence with respect to input conditions, and voxel-based morphometry. Our code is available at https://github.com/alanqrwang/morphldm.

2025-03-04T20:30:42Z Provisional Acceptance at MICCAI 2025 Alan Q. Wang Fangrui Huang Bailey Trang Wei Peng Mohammad Abbasi Kilian Pohl Mert Sabuncu Ehsan Adeli http://arxiv.org/abs/2508.10910v1 2025-07-31T12:21:52Z

Understanding cancer cell differentiation is essential for advancing its detection, diagnosis, and treatment. Mathematical models significantly contribute to this by providing a theoretical framework to understand the complex interactions between cancer stem cells, differentiated cancer cells, and immune system components. Such models depend on experimental data and computational simulations to predict tumor dynamics, offering insights into how different cell populations evolve over time. However, to ensure their realistic and consistent outcomes, rigorous mathematical analysis is required, including verification of solution uniqueness, stability, viability, positivity, and boundedness. Such validation guarantees that model's results can be used in both oncological research and clinical applications. In this study, we conduct a comprehensive analysis of an integrative mathematical model of cancer cell differentiation, with a particular focus on its interactions with immune cells. The model captures the dynamic balance between cancer stem cell self-renewal, differentiation into mature tumor cells, and immune-mediated elimination. By employing analytical and numerical techniques, we assess the model's feasibility, stability, and long-term behavior under various biological conditions. Our findings demonstrate that immune system engagement can significantly influence tumor composition and growth, highlighting potential therapeutic targets. This work not only advances theoretical cancer modeling but also provides a foundation for future experimental validation and the development of combined differentiation-immunotherapy approaches. The results underscore the importance of interdisciplinary collaboration in the fight against cancer.

2025-07-31T12:21:52Z Marcela V. Reale Gustavo Paccosi David H. Margarit Lilia Romanelli http://arxiv.org/abs/2507.22092v1 2025-07-29T12:35:08Z

Computational pathology (CPath) has shown great potential in mining actionable insights from Whole Slide Images (WSIs). Deep Learning (DL) has been at the center of modern CPath, and while it delivers unprecedented performance, it is also known that DL may be affected by irrelevant details, such as those introduced during scanning by different commercially available scanners. This may lead to scanner bias, where the model outputs for the same tissue acquired by different scanners may vary. In turn, it hinders the trust of clinicians in CPath-based tools and their deployment in real-world clinical practices. Recent pathology Foundation Models (FMs) promise to provide better domain generalization capabilities. In this paper, we benchmark FMs using a multi-scanner dataset and show that FMs still suffer from scanner bias. Following this observation, we propose ScanGen, a contrastive loss function applied during task-specific fine-tuning that mitigates scanner bias, thereby enhancing the models' robustness to scanner variations. Our approach is applied to the Multiple Instance Learning task of Epidermal Growth Factor Receptor (EGFR) mutation prediction from H\&E-stained WSIs in lung cancer. We observe that ScanGen notably enhances the ability to generalize across scanners, while retaining or improving the performance of EGFR mutation prediction.

2025-07-29T12:35:08Z Accepted (Oral) in MedAGI 2025 International Workshop at MICCAI Conference Gianluca Carloni Biagio Brattoli Seongho Keum Jongchan Park Taebum Lee Chang Ho Ahn Sergio Pereira http://arxiv.org/abs/2506.12944v3 2025-07-29T10:40:05Z

Risk stratification is a key tool in clinical decision-making, yet current approaches often fail to translate sophisticated survival analysis into actionable clinical criteria. We present a novel method for unsupervised machine learning that directly optimizes for survival heterogeneity across patient clusters through a differentiable adaptation of the multivariate logrank statistic. Unlike most existing methods that rely on proxy metrics, our approach represents novel methodology for training any neural network architecture on any data modality to identify prognostically distinct patient groups. We thoroughly evaluate the method in simulation experiments and demonstrate its utility in practice by applying it to two distinct cancer types: analyzing laboratory parameters from multiple myeloma patients and computed tomography images from non-small cell lung cancer patients, identifying prognostically distinct patient subgroups with significantly different survival outcomes in both cases. Post-hoc explainability analyses uncover clinically meaningful features determining the group assignments which align well with established risk factors and thus lend strong weight to the methods utility. This pan-cancer, model-agnostic approach represents a valuable advancement in clinical risk stratification, enabling the discovery of novel prognostic signatures across diverse data types while providing interpretable results that promise to complement treatment personalization and clinical decision-making in oncology and beyond.

2025-06-15T19:11:10Z Maximilian Ferle Jonas Ader Thomas Wiemers Nora Grieb Adrian Lindenmeyer Hans-Jonas Meyer Thomas Neumuth Markus Kreuz Kristin Reiche Maximilian Merz http://arxiv.org/abs/2507.20345v1 2025-07-27T16:28:47Z

Morphoelasticity represents a foundational theory for tracing back growth, remodelling, and morphogenesis, yet crucial challenges persist. A unified growth law -- independent of a priori assumptions about constitutive relations or specified structures of the growth tensor -- remains in fact elusive, as does an intimate connection between local anisotropic growth, morphogenesis dynamics, diffusion phenomena and mechanics. When anisotropy of growth is not prescribed arbitrarily, current frameworks mainly attribute shape emergence to growth-induced global configurational switches, somehow neglecting the existence of local drivers of spontaneous patterning or distorsions also in stress-free conditions. To overcome these limitations, in this work we propose a theoretical approach that reformulates anisotropic growth, remodeling, and morphogenesis starting from first principles, grounded in mass balance. In particular, by positing mass conservation as the sole dynamic constraint for the growth problem, we generalize the mass balance equation to derive evolution laws for mass distribution and geometric reconfiguration. This reveals how mass transport and associated microstructural evolution of the tissue fabric synergistically guide biological, geometric, and constitutive changes of living matter, with the isochoric component of growth orchestrating local anisotropy and, at the macroscale, even spontaneous shape development. The proposed fully coupled modelling approach integrates chemo-mechanical and biological interactions inherent to living systems, offering a pathway to investigate complex scenarios in growth and morphogenesis, redefining the latter in a Continuum Mechanics framework. It is felt that this strategy could help to take a step forward a unified perspective for biomechanical and mechanobiological problems, bridging scales from local drivers to emergent tissue forms and shapes.

2025-07-27T16:28:47Z Angelo Rosario Carotenuto Stefania Palumbo Arsenio Cutolo Massimiliano Fraldi http://arxiv.org/abs/2507.18887v1 2025-07-25T02:09:44Z

The emergence of spatial patterns and organized growth is a hallmark of developing tissues. While symmetry-breaking and scaling laws govern these processes, how cells coordinate spatial patterning with size regulation remains unclear. Here, we combine quantitative imaging, a Turing activator-repressor model with self-organized reactive boundaries, and in vitro models of early mouse development to study mesodermal pattern formation in two-dimensional (2D) gastruloids. We show that colony size dictates symmetry: small colonies (radius approximately 100 micrometers) spontaneously break symmetry, while larger ones remain centro-symmetric, consistent with size-dependent positional information and model predictions. The mesodermal domain area scales robustly with colony size following a power law, independent of cell density, indicating that cells sense and respond to gastruloid size. Time-lapse imaging reveals a biphasic growth law: an early power-law expansion followed by exponential arrest, marking a dynamical phase transition. These dynamics, conserved across sizes, reflect features of criticality seen in physical systems, where self-organization, scaling, and boundary feedback converge. Our findings uncover a minimal mechanism for size-dependent pattern formation and growth control. This framework enables quantitative investigation of symmetry-breaking and scaling in self-organizing tissues, offering insights into the physical principles underlying multicellular organization.

2025-07-25T02:09:44Z 9 pages, 3 figures Daniel Aguilar-Hidalgo Joel Ostblom M Mona Siu Divy Raval Ajinkya Ghagre Tiam Heydari Benjamin McMaster Jonathan Gui Nicolas Werschler Mukul Tewary Peter W. Zandstra http://arxiv.org/abs/2507.21149v1 2025-07-23T16:58:17Z

Tumor organoid-on-a-chip platforms represent a cutting-edge fusion of patient-derived organoids with microfluidic technologies, offering unprecedented capabilities for personalized cancer research. These systems overcome limitations of conventional models by enabling precise control over the tumor microenvironment, including nutrient gradients, fluid flow, and immune interactions. Tumor organoids recapitulate patient-specific tumor heterogeneity and genetic landscapes, while microfluidic chips provide dynamic perfusion and mechanical stimuli, enhancing physiological relevance. Together, they facilitate advanced applications such as high-throughput drug screening, immunotherapy testing, and metastasis modeling, showing superior predictive power for clinical outcomes. Despite challenges in standardization, scalability, and integration of complex tumor components, ongoing advances in hydrogel engineering, automation, and artificial intelligence are poised to accelerate their clinical translation. This review highlights current technologies, applications, and future directions of tumor organoid-on-a-chip systems, emphasizing their transformative potential in precision oncology.

2025-07-23T16:58:17Z 16 pages, 1 figure, 2 tables Keyvan Alavi http://arxiv.org/abs/2503.19832v2 2025-07-21T12:38:48Z

The fields of mechanobiology and biomechanics are expanding our understanding of the complex behavior of soft biological tissues across multiple scales. Given the intricate connection between tissue microstructure and its macroscale mechanical behavior, unraveling this mechanistic relationship remains an ongoing challenge. Reconstituted fiber networks serve as valuable in vitro models to simplify the intricacy of in vivo systems for targeted investigations. Concurrently, advances in imaging enable microstructure visualization and, through generative pipelines, modeling as discrete element networks. These mesoscale models provide insights into macroscale tissue behavior. However, there is still no clear way to systematically incorporate experimentally observed microstructural changes into in silico models of biological networks. In this work, we develop a novel framework to generate topologically-driven discrete fiber networks using high-resolution images that account for how environmental changes during polymerization influence the resulting structure. Leveraging these networks, we generate models of interconnected load-bearing fiber components that exhibit softening under compression and are bending-resistant. The generative topology framework enables control over network-level features, such as fiber volume fraction and cross-link density, along with fiber-level properties, like length distribution, to simulate changes driven by different polymerization conditions. We validate the robustness of our simulations against experimental data in a collagen-specific study case where we examine nonlinear elastic responses of collagen networks across varying conditions. TopoGEN provides a tool for tissue biomechanics and engineering, helping to bridge microstructural insights and bulk mechanical behavior by linking image-derived microstructural topological organization to soft tissue mechanics.

2025-03-25T16:49:47Z Journal of the Mechanics and Physics of Solids, Volume 205, 2025, 106257 Sara Cardona Mathias Peirlinck Behrooz Fereidoonnezhad 10.1016/j.jmps.2025.106257 http://arxiv.org/abs/2507.15353v1 2025-07-21T08:06:30Z

BACKGROUND: Primary aldosteronism is the most common form of secondary hypertension. The most frequent genetic cause of aldosterone-producing adenomas is somatic mutations in the potassium channel KCNJ5. They affect the ion selectivity of the channel, with sodium influx leading to cell membrane depolarization and activation of calcium signaling, the major trigger for aldosterone biosynthesis. METHODS: To investigate how KCNJ5 mutations lead to the development of aldosterone-producing adenomas, we established an adrenocortical cell model in which sodium entry into the cells can be modulated on demand using chemogenetic tools [H295R-S2 $α$7-5HT3-R ($α$7-5HT3 receptor) cells]. We investigated their functional and molecular characteristics with regard to aldosterone biosynthesis and cell proliferation. RESULTS: A clonal cell line with stable expression of the chimeric $α$7-5HT3-R in H295R-S2 (human adrenocortical carcinoma cell line, Strain 2) cells was obtained. Increased sodium entry through $α$7-5HT3-R upon stimulation with uPSEM-817 (uPharmacologically Selective Effector Molecule-817) led to cell membrane depolarization, opening of voltage-gated Ca 2+ channels, and increased intracellular Ca 2+ concentrations, resulting in the stimulation of CYP11B2 expression and increased aldosterone biosynthesis. Increased intracellular sodium influx did not increase proliferation but rather induced apoptosis. RNA sequencing and steroidome analyses revealed unique profiles associated with Na + entry, with only partial overlap with Ang II (angiotensin II) or potassium-induced changes. CONCLUSIONS: H295R-S2 $α$7-5HT3-R cells are a new model reproducing the major features of cells harboring KCNJ5 mutations. Increased expression of CYP11B2 and stimulation of the mineralocorticoid biosynthesis pathway are associated with a decrease of cell proliferation and an increase of apoptosis, indicating that additional events may be required for the development of aldosterone-producing adenomas.

2025-07-21T08:06:30Z Hypertension, 2025, 82 (4), pp.716-732 Bakhta Fedlaoui Teresa Cosentino Zeina Al Sayed Rita Alexandre Coelho Isabelle Giscos-Douriez Nicolo Faedda May Fayad Jean-Sebastien Hulot Christopher Magnus Scott Sternson Simon Travers-Allard Stephanie Baron David Penton Fabio Fernandes-Rosa Maria-Christina Zennaro Sheerazed Boulkroun http://arxiv.org/abs/2507.14380v1 2025-07-18T22:01:03Z

Model-based clustering is widely used for identifying and distinguishing types of diseases. However, modern biomedical data coming with high dimensions make it challenging to perform the model estimation in traditional cluster analysis. The incorporation of factor analyzer into the mixture model provides a way to characterize the large set of data features, but the current estimation method is computationally impractical for massive data due to the intrinsic slow convergence of the embedded algorithms, and the incapability to vary the size of the factor analyzers, preventing the implementation of a generalized mixture of factor analyzers and further characterization of the data clusters. We propose a hybrid matrix-free computational scheme to efficiently estimate the clusters and model parameters based on a Gaussian mixture along with generalized factor analyzers to summarize the large number of variables using a small set of underlying factors. Our approach outperforms the existing method with faster convergence while maintaining high clustering accuracy. Our algorithms are applied to accurately identify and distinguish types of breast cancer based on large tumor samples, and to provide a generalized characterization for subtypes of lymphoma using massive gene records.

2025-07-18T22:01:03Z Kazeem Kareem Fan Dai http://arxiv.org/abs/2507.13330v1 2025-07-17T17:48:19Z

We propose and analyze a hierarchy of three models of blood perfusion through a tissue surrounding a thin arteriole or venule. Our goal is to rigorously link 3D-3D Darcy--Stokes, 3D-1D Darcy--Poiseuille, and 1D Green's function methods commonly used to model this process. Here in Part II, we consider the most detailed level, a 3D-3D Darcy-Stokes system coupled across the permeable vessel surface by mass conservation and pressure/stress balance conditions. We derive a convergence result between the 3D-3D model and both the 3D-1D Darcy--Poiseuille model and 1D Green's function model proposed in Part I [Ohm \& Strikwerda, arXiv preprint July 2025] at a rate proportional to $ε^{1/6}|\logε|$, where $ε$ is the maximum vessel radius. The rate is limited by the inclusion of a degenerate endpoint where the vessel radius vanishes, i.e. becomes indistinguishable from a capillary. Key to our proof are \emph{a priori} estimates for the 1D integrodifferential model obtained in Part I.

2025-07-17T17:48:19Z Laurel Ohm Sarah Strikwerda http://arxiv.org/abs/2507.13316v1 2025-07-17T17:34:05Z

We propose and analyze a family of models describing blood perfusion through a tissue surrounding a thin blood vessel. Our goal is to rigorously establish convergence results among 3D-3D Darcy--Stokes, 3D-1D Darcy--Poiseuille, and 1D Green's function methods commonly used to model this process. In Part I, we propose a 3D-1D Darcy--Poiseuille system where the coupling across the permeable vessel surface involves an angle-averaged Neumann boundary condition coupled with a geometrically constrained Robin boundary condition. We show that this model is well-posed and moreover limits to a 1D Green's function model as the maximum vessel radius $ε\to 0$. In the 1D model, the exterior blood pressure is given by an explicit Green's function expression involving the interior blood pressure. The interior pressure satisfies a novel 1D integrodifferential equation in which the integral term incorporates the effects of the exterior pressure and the vessel geometry. Much of this paper is devoted to analyzing this integrodifferential equation. Using the \emph{a priori} bounds obtained here, we show that the solution to the 1D model converges to the 3D-1D solution with a rate proportional to $ε^{1/2}|\logε|$. In Part II [Ohm \& Strikwerda, arXiv preprint July 2025], we rely on the 1D estimates to show that both the 1D and 3D-1D models converge to a coupled 3D-3D Darcy-Stokes system as $ε\to 0$, thereby establishing a convergence chain among all hierarchy levels.

2025-07-17T17:34:05Z Laurel Ohm Sarah Strikwerda http://arxiv.org/abs/2503.01248v4 2025-07-13T15:52:05Z

Diabetic retinopathy (DR) is a leading cause of vision loss, requiring early and accurate assessment to prevent irreversible damage. Spectral Domain Optical Coherence Tomography (SD-OCT) enables high-resolution retinal imaging, but automated segmentation performance varies, especially in cases with complex fluid and hyperreflective foci (HRF) patterns. This study proposes an active-learning-based deep learning pipeline for automated segmentation of retinal layers, fluid, and HRF, using four state-of-the-art models: U-Net, SegFormer, SwinUNETR, and VM-UNet, trained on expert-annotated SD-OCT volumes. Segmentation accuracy was evaluated with five-fold cross-validation, and retinal thickness was quantified using a K-nearest neighbors algorithm and visualized with Early Treatment Diabetic Retinopathy Study (ETDRS) maps. SwinUNETR achieved the highest overall accuracy (DSC = 0.7719; NSD = 0.8149), while VM-UNet excelled in specific layers. Structural differences were observed between non-proliferative and proliferative DR, with layer-specific thickening correlating with visual acuity impairment. The proposed framework enables robust, clinically relevant DR assessment while reducing the need for manual annotation, supporting improved disease monitoring and treatment planning.

2025-03-03T07:23:56Z 18 pages, 11 figures S. Chen D. Ma M. Raviselvan S. Sundaramoorthy K. Popuri M. J. Ju M. V. Sarunic D. Ratra M. F. Beg http://arxiv.org/abs/2507.07604v1 2025-07-10T10:06:13Z

Synthetic molecular communication (SMC) is a key enabler for future healthcare systems in which Internet of Bio-Nano-Things (IoBNT) devices facilitate the continuous monitoring of a patient's biochemical signals. To close the loop between sensing and actuation, both the detection and the generation of in-body molecular communication (MC) signals is key. However, generating signals inside the human body, e.g., via synthetic nanodevices, poses a challenge in SMC, due to technological obstacles as well as legal, safety, and ethical issues. Hence, this paper considers an SMC system in which signals are generated indirectly via the modulation of a natural in-body MC system, namely the gut-brain axis (GBA). Therapeutic GBA modulation is already established as treatment for neurological diseases, e.g., drug refractory epilepsy (DRE), and performed via the administration of nutritional supplements or specific diets. However, the molecular signaling pathways that mediate the effect of such treatments are mostly unknown. Consequently, existing treatments are standardized or designed heuristically and able to help only some patients while failing to help others. In this paper, we propose to leverage personal health data, e.g., gathered by in-body IoBNT devices, to design more versatile and robust GBA modulation-based treatments as compared to the existing ones. To show the feasibility of our approach, we define a catalog of theoretical requirements for therapeutic GBA modulation. Then, we propose a machine learning model to verify these requirements for practical scenarios when only limited data on the GBA modulation exists. By evaluating the proposed model on several datasets, we confirm its excellent accuracy in identifying different modulators of the GBA. Finally, we utilize the proposed model to identify specific modulatory pathways that play an important role for therapeutic GBA modulation.

2025-07-10T10:06:13Z Sebastian Lotter Elisabeth Mohr Andrina Rutsch Lukas Brand Francesca Ronchi Laura Díaz-Marugán http://arxiv.org/abs/2507.07131v1 2025-07-08T21:55:39Z

Plain X-ray is one of the most common image modalities for clinical diagnosis (e.g. bone fracture, pneumonia, cancer screening, etc.). X-ray image segmentation is an essential step for many computer-aided diagnostic systems, yet it remains challenging. Deep-learning-based methods have achieved superior performance in medical image segmentation tasks but often require a large amount of high-quality annotated data for model training. Providing such an annotated dataset is not only time-consuming but also requires a high level of expertise. This is particularly challenging in wrist bone segmentation in X-rays, due to the interposition of multiple small carpal bones in the image. To overcome the data annotation issue, this work utilizes a large number of simulated X-ray images generated from Computed Tomography (CT) volumes with their corresponding 10 bone labels to train a deep learning-based model for wrist bone segmentation in real X-ray images. The proposed method was evaluated using both simulated images and real images. The method achieved Dice scores ranging from 0.80 to 0.92 for the simulated dataset generated from different view angles. Qualitative analysis of the segmentation results of the real X-ray images also demonstrated the superior performance of the trained model. The trained model and X-ray simulation code are freely available for research purposes: the link will be provided upon acceptance.

2025-07-08T21:55:39Z 4 pages Youssef ElTantawy Alexia Karantana Xin Chen