As we step into a new decade, one of the most important breakthroughs of the previous decade is the rapid proliferation of artificial intelligence (AI) applications. Thanks to significant improvements in computing power and digital storage costs, data-driven AI algorithms are helping us solve many complex problems, especially in the field of computer vision, which benefit a broad spectrum of applications ranging from medical imaging to autonomous cars.
At Cincinnati Children’s, we are fortunate to have one of the largest pediatric dataset in the country which helps us leverage the power of AI to improve the standard of care for kids. In radiology, we have undertaken multiple AI research projects that are focused on improving your child’s safety, comfort and cure.
Low Dose CT Scans
Although our current CT protocols deliver optimal radiation doses to children undergoing CT exams, next generation AI-driven CT algorithms can significantly reduce the dose delivered. Our team worked together to validate newly developed AI-based CT reconstruction algorithms against current methods through several qualitative and quantitative comparisons. The results showed that up to 50% reduction in dose is possible while maintaining the same image quality.
https://pubs.rsna.org/doi/abs/10.1148/radiol.2020202317
Quick Turn Around Time for X-rays
The lateral airway radiography exam that is used to look for foreign bodies or upper airway infections is a little tricky as the children undergoing this exam are young and have trouble maintaining their head in an extended position for a clear view of the airway. Currently, the technologists have to communicate with a radiologist to make sure the exam is adequate for diagnosis before the child can leave the x-ray room. However, our newly developed AI algorithm performs as well as the radiologists in detecting poor quality exams and can speed up the time the kids have to spend in the radiology suite.
https://pubs.rsna.org/doi/abs/10.1148/ryai.2020190226
Sarcopenia Norms for Kids
The depletion of muscle mass, known as sarcopenia, is associated with many medical complications. Early diagnosis of sarcopenia can help with improved care and outcome for various diseases. However, currently there are no standard reference values for diagnosing sarcopenia in children. With the help of our large imaging dataset and the advanced AI tools, we have built a pipeline to automatically calculate the muscle area for thousands of normal children that have undergone a CT scan. This pipeline is currently being used to establish the reference values for sarcopenia diagnosis for children of every age group.
https://pubs.rsna.org/doi/10.1148/ryai.2021200130
Overall, we have just started scratching the surface of all the benefits AI can bring to improve children’s care and the faculty and staff at Cincinnati Children’s are very excited for the AI-driven future.
Elanchezhian (Elan) Somasundaram, PhD, author; Glenn Miñano, BFA, editor; Meredith Towbin, copyeditor
