Harnessing the Power of AI and Plasmonics for Early Cancer Detection

A group of researchers from the Korea Institute of Materials Science's Advanced Bio and Healthcare Materials Research Division, under the direction of Dr. Ho Sang Jung, have created a novel sensor material that uses artificial intelligence to analyze and diagnose cancer by amplifying the optical signals of cancer metabolites in bodily fluids like urine, mucus, and saliva. The journal Biosensors and Bioelectronics published the study.

In contrast to conventional blood draws or biopsies, this technique provides a non-invasive method of diagnosing cancer by quickly and sensitively detecting metabolites and changes in the bodily fluids of cancer patients.

Working with Professor Soo Woong Yoo of Chonnam National University Hospital, the team diagnosed colorectal cancer using a plasmonic needle that amplifies the Raman signals of molecules. This needle is inserted into a 1 mm hole, which can be used with a colonoscopy camera to swab the tumor’s surface without causing bleeding, allowing for examination of its composition.

In partnership with Professor Byung-Ho Chung of Samsung Medical Center, the researchers also created a device that gathers saliva from lung cancer patients and classifies the cancer stage. The amount of volatile organic chemicals (VOCs) in the breath of lung cancer patients differs from that of healthy individuals. 

These VOCs, found as metabolites of lung cancer, dissolve in saliva. The team has created a device using paper-based sensors and artificial intelligence to stage lung cancer and distinguish between healthy individuals and cancer patients.

Numerous stories describe dogs barking persistently at their owners, leading the owners to suspect something was wrong. Upon visiting the doctor, they discovered cancer. This is due to dogs’ extraordinary sense of smell, which allows them to detect metabolites like VOCs in bodily fluids.

Inspired by this, the research team aimed to incorporate these principles into a cancer-detection device. Their approach uses plasmonic materials, which amplify Raman signals by more than 100 million times, to detect metabolites in bodily fluids with high sensitivity—without the need for complex or expensive traditional equipment.

The team proposed biomarkers for diagnosis through mathematical modeling calculations and artificial intelligence analysis.

The developed technology can be expanded not only to diagnose cancer but also to diseases with poorly understood diagnostics, such as synaptic diseases. We will enter the global diagnostic market based on domestic source technologies and take the lead in developing technologies that people can experience.

Dr. Ho Sang Jung, Project Lead, Korea Institute of Materials Science

The team's invention was named the top research achievement in KIMS's 'Top 10 Outstanding Research Achievements of the Year' survey last year, and they are still creating cutting-edge technology.

The research was supported by the Fundamental Research Program of KIMS and the Biomedical Technology Development Project of the National Research Foundation of Korea. The study's findings have been acknowledged for their excellence and published in three journals.

The journal Biosensors, and Bioelectronics published two publications on January 15, 2024, and August 3, 2024. On August 1, 2024, a paper was also published in Sensors and Actuators B-Chemical.

Furthermore, a total of 11 patents have been submitted for this research in Korea, the United States, and Europe.

Last year, the research team developed a urine-based cancer diagnosis tool, which they handed over to SOLUM Healthcare for licensing and incorporation into products. The technology has now advanced to the point where it can simultaneously detect multiple types of tumors in urine.

The team tested urine samples from over 250 individuals with colorectal, lung, prostate, and pancreatic cancer. Using artificial intelligence and rapid analysis, they were able to obtain results for 100 patients in about two hours.

The researchers also reported achieving clinical sensitivity and specificity rates of over 98 %.

Journal Reference:

Nhat, T., et al. (2023) 3D plasmonic hexaplex paper sensor for label-free human saliva sensing and machine learning-assisted early-stage lung cancer screening. Biosensors and Bioelectronics. doi.org/10.1016/j.bios.2023.115779.