As stated by the RSNA, innovative AI systems are capable of recognizing non-smokers with significant risks of lung cancer.
In Brief
A new AI program has been created that can pinpoint non-smokers who may be at an increased risk of developing lung cancer, utilizing standard chest X-ray images for assessment.
At the Cardiovascular Imaging Research Center (CIRC), a significant breakthrough has been made by researchers in developing an AI tool tool aimed at identifying individuals who do not smoke but are still at elevated risk of lung cancer. lung cancer using a routine chest X-ray image.
Research indicates that around 10-20% of lung cancer cases occur in individuals who have never smoked—defined as those who have either never touched a cigarette or have smoked less than 100 cigarettes in their lifetime.
The upcoming study, set to be showcased at the annual meeting of the Radiological Society of North America (RSNA) next week, focuses on enhancing the prediction of lung cancer risk among non-smokers. The researchers are examining how effective a deep learning model is at recognizing high-risk non-smokers through chest X-rays. X-rays This model specifically extracts data from electronic medical records to discover patterns that may indicate the presence of the disease.
Utilizing deep learning—an advanced subset of artificial intelligence—the model is trained to identify these patterns present in X-ray images.
"One of the significant benefits of our method is that it relies solely on a single chest X-ray image, a standard medical test readily accessible within electronic medical records,\" explained Anika S. Walia, the lead author and a medical student at Boston University School of Medicine.
The 'CXR-Lung-Risk' model was crafted based on a dataset of 147,497 chest X-rays from 40,643 individuals, both symptomatic smokers and never-smokers, aiming to evaluate lung-related mortality risk through just one chest X-ray.
Additionally, the researchers tested their model on a different cohort of non-smokers who underwent routine chest X-rays from 2013 to 2014. The main outcome assessed was the development of lung cancer over a six-year span. Out of 17,407 participants (with an average age of 63), 28% were flagged as high risk by the deep learning model, with 2.9% later diagnosed with lung cancer.
The proportion of high-risk individuals significantly surpassed the 1.3% six-year risk threshold that prompts lung cancer screening as advised by National Comprehensive Cancer Network guidelines.
The risk of lung cancer in the high-risk group was still 2.1 times greater compared to those classified as low risk.
"This AI tool paves the way for opportunistic screening of non-smokers who are at a great risk of lung cancer, utilizing existing chest X-rays stored in electronic medical records,\" stated Michael T. Lu, the director and co-director of CIRC at Massachusetts General Hospital (MGH). artificial intelligence \"With smoking rates on the decline, finding methods for early lung cancer detection among non-smokers becomes increasingly crucial.\"
MIT's AI Technology to Spot Early Lung Cancer Signs
In another significant advancement earlier this year, researchers from the Mass General Cancer Center and MIT assert that their AI technology can detect early signs of lung cancer several years before conventional CT scans reveal them.
The tool, known as Sybil, has shown impressive results in recent studies, accurately forecasting whether someone will develop lung cancer within a year with a success rate anywhere between 86% to 94%.
Currently, the CDC recommends annual low-dose CT scans for adults at risk of lung cancer as a preventative measure.
Nevertheless, even with regular screenings, there are inherent limitations to what even expert radiologists can identify, highlighting the role of Sybil in this scenario.
Dr. Lecia Sequist, an oncologist and the program director at the Cancer Early Detection and Diagnostics Clinic at Massachusetts General Hospital, points out the physical limitations of human observation by saying, \"There's so much the naked eye cannot perceive.\" She emphasizes that the AI named Sybil takes a completely novel approach to scan analyses compared to what human radiologists typically do.
Cardiovascular Imaging Research Center
Disclaimer
In line with the Trust Project guidelines Kumar, an adept tech journalist specializing in the intersections of AI/ML, marketing technologies, and emerging domains like cryptocurrencies, blockchain, and NFTs, has amassed over three years of industry experience. He has built a reputation for crafting engaging narratives, conducting insightful interviews, and offering comprehensive analyses. Kumar excels at creating impactful content, including articles, reports, and research works for prominent industry platforms, skillfully communicating complex tech concepts in an accessible and engaging manner.