Dayton Engineer

Dr. Russell Hardie with the Department of Electrical and Computer Engineering has been awarded an NIH grant to harness artificial intelligence (AI) to revolutionize the detection of metastatic lung disease in pediatric patients by identifying lung nodules in computed tomography (CT) scans. Lung nodule detection is a critically important task that is more difficult in children than in adults. 

Hardie will collaborate with his former doctoral student, Dr. Barath Narayanan, now a senior research scientist at the University of Dayton Research Institute.

The project involves the development and testing of a new AI system to review CT scans and automatically identify the presence and location of any lung nodules. This system will aid radiologists in an otherwise time-consuming and fatiguing task that is subject to human error. The integration of AI into the radiologists review process has been shown to  dramatically enhance  accuracy and efficiency for lung nodule detection in adults. However, to date, no pediatric-specific systems have been developed or tested.

Lung nodules are spherical masses ranging in diameter up to 30 millimeters, and the presence of these can indicate serious health issues, including lung cancer. While lung nodules in adults could be a sign of primary lung cancer, they are often associated with metastatic disease in pediatric cases. 

"In children, the presence of lung nodules can signal that cancer from elsewhere in the body has spread to the lungs," Hardie said.  

Early detection is vital for the best treatment options and outcomes. Unfortunately, lung nodule detection by radiologists is challenging due to the vast amount of image data that must be reviewed and the relatively small size of the nodules in pediatric cases.

"When radiologists review a CT scan, there are hundreds of images they have to look at, sometimes thousands,” Hardie said. “They scroll back and forth through the images along one of the three dimensions, looking for the telltale spherical shape of a nodule. However, there are vessels throughout the lungs. In a single cross-section image it is difficult to distinguish the vessels from a nodule, everything just looks round.”

In response to this challenge, Hardie and his team are collaborating with Cincinnati Children's Hospital Medical Center (CCHMC), where a team of pediatric radiologists have expressed the need for assistance in identifying lung nodules. 

"When we realized that lung nodule detection was a pressing issue for them, it was a perfect fit," Hardie said.

Hardie and Narayanan are not new to using AI to detect lung nodules, as they have already created a system to detect nodules in adults. Hardie started this work back during his first faculty sabbatical in 2003. For the current project, they are building on their prior work to create a new first-of-kind system specifically designed for pediatrics.

“Our first big study was to see if we could take our existing adult system, known as FlyerScan, and use it successfully on pediatric data from CCHMC. We discovered that performance dropped. The vascular structure in children is more compact and the nodules tend to be much smaller in the pediatric cases than in adults, with many as small as one to two millimeters,” Hardie said. 

The team determined that a pediatric-specific lung nodule detection system was needed, leading to the NIH-funded project.

“The project would not be possible without the support and collaboration of the team of radiologists at CCHMC,” Hardie said. “The radiologists provide clinical expertise to guide the system development and evaluate its efficacy. The team at CCHMC is also providing CT scans and expertly annotated data with which to train the AI system.”

As the research progresses, the potential impact of their work looms large. With AI at the forefront of lung nodule detection, the future of pediatric healthcare is looking brighter, offering hope for early diagnosis and treatment in a field where every moment counts.

"The American College of Radiology promotes the principle that, ‘AI used in pediatric patients should be designed for and shown to work in pediatric patients.’ This is exactly what we are aiming to do. It’s our guiding light," Hardie said.