HistoNeRF: An Accessible and Intelligent Approach for Comprehensive 2D-to-3D Histological Assessment

Kubilay Doğan KILIÇ; Kaan Özyazıcı; Hüseyin Kısaoğlu; Zeynep Simge Yılmaz; Büşra Horuz; Ayşegül Taşkıran Özyazıcı

doi:10.1002/jemt.70124

HistoNeRF: An Accessible and Intelligent Approach for Comprehensive 2D-to-3D Histological Assessment

Makaleler > Tam Makale · 2026

Yazarlar: Kubilay Doğan KILIÇ, Kaan Özyazıcı, Hüseyin Kısaoğlu, Zeynep Simge Yılmaz, Büşra Horuz, Ayşegül Taşkıran Özyazıcı

MICROSCOPY RESEARCH AND TECHNIQUE, SCI İndekslerine Giren Dergi, Uluslararası, Wos, 2026, cilt 0, ss 0-0

DOI: 10.1002/jemt.70124

Özet

Histological analysis is central to biomedical research and diagnostic pathology, yet conventional two- dimensional (2D) section-ing captures only limited aspects of tissue architecture. Critical spatial relationships—such as tumor boundaries, stromal or-ganization, and vascular networks—remain obscured, restricting diagnostic accuracy and biological interpretation. HistoNeRFaddresses these limitations by adapting Neural Radiance Fields (NeRF) to reconstruct three-dimensional (3D) tissue volumesfrom routine histological sections. In this study, 84 toluidine blue (TB)–stained murine ovarian sections were digitized, alignment-corrected, and integrated into volumetric models. Tissue segmentation was performed using a convolutional neural network,while visualization was achieved through an interactive, GPU-accelerated interface. To ensure accessibility and reproducibility,a Python-based graphical application (HistoNeRF GUI) was developed following Human–Computer Interaction (HCI) princi-ples and containerized with Docker, allowing installation-free deployment via Docker Hub. HistoNeRF produced high-fidelity3D reconstructions (SSIM = 0.92; Dice similarity coefficient = 0.88), enabling expert histologists to better visualize follicularstructures, stromal compartments, and vascular elements. The containerized GUI was deployed successfully from Docker Hub,providing immediate access to 3D reconstruction without a complex local setup. By overcoming the inherent constraints of 2Dmicroscopy, HistoNeRF enhances the visualization, interpretability, and reproducibility of histological architecture. The HCI-guided, cross-platform interface supports scalability and rapid adoption in digital pathology workflows. Although validation waslimited to murine ovarian tissue and one staining protocol, this framework can be extended across tissue types and clinical data-sets. HistoNeRF bridges routine histology and 3D volumetric analysis through accurate, interactive reconstructions that advancediagnostic precision and biomedical research. While demonstrated on 84 serial TB-stained ovarian sections, broader validationacross tissues, stains, and pathological conditions remains future work; to support this, we provide a Dockerized, modular pipe-line for straightforward extension.

Detay görüntüle