arXiv subjects

Gizem Mert

Publications and source records attributed to Gizem Mert.

3 recordsLinked to original sources

arXiv2026

Multimodal Hierarchical Mixture-of-Experts for 3D Medical Vision-Language Modeling

Medical 3D scans and reports are inherently hierarchical, with diagnostic cues concentrated in specific anatomical regions across multiple levels of organization. Yet, most medical vision-language models ignore this structure and fail to focus on clinically relevant anatomy. We introduce MedHiVE, which combines hierarchical vision-language Soft Mixture-of-Experts (SMoE) with contrastive learning to jointly model fine-grained anatomy and global scan context. To our knowledge, MedHiVE is the first framework to route both vision and text tokens through a two-level SMoE hierarchy organized by anatomical scale, coupling this with report-derived anatomy-level contrastive alignment for 3D medical imaging. This replaces the direct region-text alignment of statically pooled features and vision-only hierarchical routing of prior work with dynamic, slot-based cross-modal information flow across anatomical scales. MedHiVE operates via a two-level, learnable token-routing mechanism, where local image and text features are first routed through a local SMoE and then fused by a global SMoE into unified multimodal representations. Contrastive learning aligns both anatomy-level and global vision-language representations in a unified space. Pre-trained on 25,000+ CT scans with anatomical supervision, MedHiVE consistently surpasses 11 state-of-the-art methods across 6 datasets on 3 tasks: segmentation, classification, and detection, demonstrating the importance of hierarchical modeling. The code for MedHiVE is at github.com/szczurek-lab/MedHiVE.

cs.CV↗

arXiv2026

Mamba Goes HoME: Hierarchical Soft Mixture-of-Experts for 3D Medical Image Segmentation

In recent years, artificial intelligence has significantly advanced medical image segmentation. Nonetheless, challenges remain, including efficient 3D medical image processing across diverse modalities and handling data variability. In this work, we introduce Hierarchical Soft Mixture-of-Experts (HoME), a two-level token-routing layer for efficient long-context modeling, specifically designed for 3D medical image segmentation. Built on the Mamba Selective State Space Model (SSM) backbone, HoME enhances sequential modeling through adaptive expert routing. In the first level, a Soft Mixture-of-Experts (SMoE) layer partitions input sequences into local groups, routing tokens to specialized per-group experts for localized feature extraction. The second level aggregates these outputs through a global SMoE layer, enabling cross-group information fusion and global context refinement. This hierarchical design, combining local expert routing with global expert refinement, enhances generalizability and segmentation performance, surpassing state-of-the-art results across datasets from the three most widely used 3D medical imaging modalities and varying data qualities. The code is publicly available at https://github.com/gmum/MambaHoME.

eess.IV↗

arXiv2024

Multimodal Analysis of White Blood Cell Differentiation in Acute Myeloid Leukemia Patients using a β-Variational Autoencoder

Biomedical imaging and RNA sequencing with single-cell resolution improves our understanding of white blood cell diseases like leukemia. By combining morphological and transcriptomic data, we can gain insights into cellular functions and trajectoriess involved in blood cell differentiation. However, existing methodologies struggle with integrating morphological and transcriptomic data, leaving a significant research gap in comprehensively understanding the dynamics of cell differentiation. Here, we introduce an unsupervised method that explores and reconstructs these two modalities and uncovers the relationship between different subtypes of white blood cells from human peripheral blood smears in terms of morphology and their corresponding transcriptome. Our method is based on a beta-variational autoencoder (ß-VAE) with a customized loss function, incorporating a R-CNN architecture to distinguish single-cell from background and to minimize any interference from artifacts. This implementation of ß-VAE shows good reconstruction capability along with continuous latent embeddings, while maintaining clear differentiation between single-cell classes. Our novel approach is especially helpful to uncover the correlation of two latent features in complex biological processes such as formation of granules in the cell (granulopoiesis) with gene expression patterns. It thus provides a unique tool to improve the understanding of white blood cell maturation for biomedicine and diagnostics.

cs.CV↗