Domestic researchers have developed a technology that can early screen for lung cancer using exhalation. It showed 95% accuracy in clinical trials. This allows for early screening of lung cancer at a simple and low expense without radiation risks, which is expected to help in the prevention and treatment of the disease.
The Electronics and Telecommunications Research Institute (ETRI) announced on the 11th that it has developed a sensor system that detects various volatile organic compounds (VOCs) generated by cancer cell masses in the lungs through exhalation, and an artificial intelligence (AI) deep learning algorithm technology to analyze the data obtained to identify lung cancer patients. The research results were published in the international journal "Sensors and Actuators B: Chemical" in June of last year.
Earlier, researchers developed an "electronic nose" in 2019 that detects lung cancer using breath. This technology is inspired by how human noses detect odors through nerve cells; when breath gases enter, an electronic sensor acts like a human nose, converting the smell into electrical signals and determining the presence or absence of disease through AI deep learning. However, the diagnostic accuracy of the electronic nose technology for lung cancer is about 75%, which has not reached the level suitable for practical screening.
The newly developed technology enables simple lung cancer screening using only human breath. Initially, the exhalation of the test subject is collected in a plastic kit. When exhalation is connected to a Teflon-based bag and a carbon-adsorption tube stick, various gas components released during breathing adhere to the stick. The stick is then reinserted into the lung cancer early diagnosis system to obtain electrical signals varying according to the composition of the exhalation and the amount of VOCs sticking to the carbon tube stick. Learning and analyzing this with the AI deep learning algorithm can help determine the likelihood of lung cancer.
The research team collected clinical samples of breath from 107 lung cancer patients and 74 healthy individuals over the past 10 years in cooperation with Professor Jeon Sang-hoon from the Cardiothoracic Surgery Department of Bundang Seoul National University Hospital, analyzed them through standard devices and gas sensors, and compiled them into databases. Based on this, they developed an AI deep learning algorithm model, and as a result, the team was able to screen lung cancer patients with more than 95% accuracy within 20 minutes after collecting the breath.
The research team noted that this is a "next-generation lung cancer early diagnosis technology that can encompass the strengths of existing immune diagnostics and molecular diagnostics", adding that "it is cheaper and faster to produce compared to existing hospital diagnostic equipment and has a higher accuracy rate compared to existing medical equipment (low-dose lung CT scans), with excellent convenience." The technology is expected to be commercialized through technology transfer and investment to medical device companies in the future, and the research team plans to secure reproducibility and reliability by conducting large-scale clinical trials involving more than 1,000 cases through follow-up research.
Lee Dae-sik, a senior researcher at ETRI, stated, "If the technology developed this time is commercialized, it will enhance the treatment and survival rates of lung cancer patients through early screening," and added, "It is expected to secure domestic market competitiveness in the medical device sector and help reduce the government's health insurance expenditure costs." Professor Jeon Sang-hoon remarked, "I hope to improve the system's reproducibility and reliability by expanding the clinical scale and to enhance the system based on big data to contribute to public health improvement."
Reference materials
Sensors and Actuators B: Chemical(2024), DOI: https://doi.org/10.1016/j.snb.2024.135578