Prof. YongKeun Park’s group at KAIST has developed a breakthrough technique for three-dimensional (3D) histopathology that eliminates the need for conventional staining. By combining holotomography with deep learning, the team can generate 3D hematoxylin and eosin (H&E)-like images of thick, label-free cancer tissues. This advance, published in Nature Communications on 6 May 2025, enables researchers and clinicians to visualize intact tissue microarchitecture at subcellular resolution without damaging the specimen.

Traditionally, histopathology has relied on thin, chemically stained sections viewed under bright-field microscopes. While effective, this method captures only two-dimensional (2D) slices and requires labor-intensive processes such as serial sectioning and staining. These limitations make it difficult to analyze thick tissue structures and can introduce artifacts that obscure diagnostic insights.

Prof. YongKeun Park’s group at KAIST overcame these challenges by measuring the 3D refractive index (RI) distribution of tissues with holotomography—a label-free imaging technique that reconstructs the optical properties of a sample with subcellular resolution. They then trained a deep learning framework to translate RI data into virtual H&E images, validated against conventional stained slides. Applied to colon and gastric cancer samples as thick as 50 μm—far beyond the limits of standard pathology slides—the method revealed detailed 3D microanatomical structures, including glands, lumens, nuclei, and necrotic regions.

Importantly, the researchers demonstrated that their virtual staining approach not only reproduces standard histological features but also provides quantitative and volumetric information inaccessible via 2D methods. For example, the 3D images allowed precise tracking of nuclear morphology, lumen contraction, and vascular structures across tissue depths. Moreover, the method proved scalable and reproducible across different institutions and cancer types, underscoring its robustness for clinical and research applications.

“This work brings histopathology decisively into the 3D domain,” said Prof. YongKeun Park, corresponding author and faculty member in the Department of Physics at KAIST. “Our label-free virtual H&E staining enables rapid, non-destructive, and quantitative analysis of cancer tissues, providing insights that are difficult or impossible to obtain with conventional pathology.”

This work was conducted in collaboration with researchers at Vanderbilt University Medical Center, Gangnam Severance Hospital, and Seoul National University Boramae Medical Center, and was supported by the National Research Foundation of Korea (RS-2024-00442348, 2022M3H4A1A02074314, RS-2023-00241278, RS-2024-00351903), an Institute of Information & Communications Technology Planning & Evaluation (IITP; 2021-0-00745) grant funded by the Korea government (MSIT), Korea Institute for Advancement of Technology (KIAT) through the International Cooperative R&D program (P0028463), and a Korean Fund for Regenerative Medicine (KFRM) grant funded by the Korean government (the Ministry of Science and ICT and the Ministry of Health & Welfare) (21A0101L1-12), Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Korea (HI21C0977, HR22C1605).
2025 KI Newsletter