Lung cancer kills 160,000 people in the U.S. every year and the global death toll from the disease is around 2 million. Lung cancer can be lurking in plain sight, only to be detected too late to be properly treated. Like all cancers, the best chance for successful treatment of lung cancer depends on early detection by screening people at high risk for the disease, such as smokers. However, these screenings aren’t perfect, and the subtle difference between a harmful tumor (malignant) and a harmless tumor (benign) can be difficult to distinguish from a CT scan.
A collaboration between Google and medical partners including Northwestern University revealed a new AI-based tool that can create a better model of a patient’s lung from the CT scan images. This 3-D image helps AI to better predict malignancy of tumors by learning from previous scans. The system was able to confirm evidence of cancer more quickly and had fewer false positive and false negatives than its human counterparts.
“This area of research is incredibly important, as lung cancer has the highest rate of mortality among all cancers, and there are many challenges in the way of broad adoption of lung cancer screening,” Shravya Shetty, the technical lead at Google, said in a University release. “Our work examines ways AI can be used to improve the accuracy and optimize the screening process, in ways that could help with the implementation of screening programs.”
While lung cancer cells can be destroyed if detected soon enough, only a small number of the eligible U.S. population is screened for lung cancer, according to Google. A more advanced system may be able to detect lung cancer quickly and influence more people to screen for the disease.
“Radiologists generally examine hundreds of two-dimensional images or ‘slices’ in a single CT scan but this new machine learning system views the lungs in a huge, single three-dimensional image,” study co-author and anesthesiology and engineering Prof. Mozziyar Etemadi said in the release. “AI in 3D can be much more sensitive in its ability to detect early lung cancer than the human eye looking at 2-D images. This is technically ‘4D’ because it is not only looking at one CT scan, but two (the current and prior scan) over time.”