KAIST Expects Innovation in Non-Invasive Pathological Diagnosis: Observing Cancer Tissues Without Dissection

An international collaborative research team has developed an innovative technology that enables the observation of cancer tissues without the need for dissection. Departing from the traditional method of slicing and staining cancer tissues for observation, the core of the newly developed technology lies in applying artificial intelligence (AI)-based deep learning algorithms to the three-dimensional structure of cancer tissues, creating virtually stained images that resemble real-life samples.

(From the bottom left) Juyeon Park, PhD candidate in Physics, Yonggeun Park, Professor of Physics, (From the top left) Sujin Shin, Professor at Gangnam Severance Hospital, Taehyun Hwang, Professor at Vanderbilt University Medical Center Mayo Clinic. Provided by KAIST

On May 26, KAIST announced that the research team led by Professor Yonggeun Park from the Department of Physics, in collaboration with Professor Sujin Shin’s team at Gangnam Severance Hospital of Yonsei University, Professor Taehyun Hwang’s team at Mayo Clinic in the United States, and the Tomocube AI research team, has developed an innovative technology that allows vivid observation of the three-dimensional structure of cancer tissues without any staining process.

In pathology, the conventional method of observing cancer tissues under a microscope only allows for the examination of specific cross-sections of the three-dimensional tissue, which limits the ability to understand the spatial arrangement and three-dimensional intercellular connections.

To address this, the collaborative research team utilized an advanced optical technology called holotomography (HT) to measure the three-dimensional refractive index information of tissues, and applied an AI-based deep learning algorithm to generate “virtual staining” (Hematoxylin & Eosin·H&E) images. Virtual staining is the most widely used staining method for observing pathological tissues.

Comparison data between the conventional 3D histopathology procedure and the 3D virtual H&E staining technology proposed by the collaborative research team. The conventional method requires the preparation and staining of dozens of tissue slides. In contrast, the technology developed by the collaborative research team can reduce the number of slides by up to 10 times and rapidly generate H&E images without the staining process. Provided by KAIST

The research team quantitatively confirmed that the images generated by this technology are highly similar to actual stained tissue images. They also demonstrated consistent performance across various organs and tissues, proving the versatility and reliability of the technology as a next-generation pathology analysis tool.

Furthermore, by verifying the feasibility of the technology in collaboration with hospitals and research institutions in Korea and the United States using Tomocube’s holotomography equipment, the research team opened up the possibility for the technology to be introduced and utilized in real-world pathology research settings.

Professor Yonggeun Park stated, “This study is a highly significant achievement that expands the unit of analysis in pathology from two dimensions to three dimensions. The research results (the developed technology) are expected to be widely used in various biomedical research and clinical diagnostics, such as analyzing the boundaries of cancer tumors and the spatial distribution of surrounding stromal cells in micro-tumor environments.”

The research was supported by the Leader Research Program of the National Research Foundation of Korea, the Global Industry Technology Cooperation Center Project of the Korea Institute for Advancement of Technology, and the Korea Health Industry Development Institute.

Juyeon Park, an integrated master’s and doctoral student at KAIST, participated as the first author of the study. The research results (paper) were published online in the May 22 issue of Nature Communications.

This content was produced with the assistance of AI translation services.