Kenneth Tang

M.S. in Biotechnology

AIRdx

Founder / CEO

NLST (Learn more about this study)

NLST-702

Aug 18, 2020

Development of deep-learning software for the detection, classification, and malignancy risk assessment of lung nodules on LDCT

Lung cancer is most common cause of cancer death worldwide. Emphasis has been placed on developing guidelines and technologies that promote early detection and diagnosis of lung cancer, with the goal of reducing the mortality rate and improving the chance of survival for patients diagnosed with lung cancer. Over the past decade, numerous governmental agencies and medical societies have published guidelines to promote routine screening of high-risk patients utilizing low-dose computed tomography (LDCT). This has allowed for earlier detection and diagnosis of lung cancer, increasing the overall cure and survival rate. However, the implementation of LDCT screening has also resulted in a greater number of false-positive exams, many of which led to invasive testing, such as subsequent standard CT imaging, biopsy, and occasionally even surgery. Not to mention, radiologists are expected to process, read, and dictate a greater volume of studies a day, which may result in overall reduction in quality of readings and physician burnout.

The goal of this project is to develop a CAD software utilizing CNN algorithms to assist radiologists in the detection, Lung-RADS classification, and malignancy risk assessment of lung nodules on LDCT. This would allow for greater diagnostic confidence amongst radiologists and reduction in false positive findings, subsequently mitigating the risks of lung cancer screening and preventing unnecessary testing.