by Lisa Chamoff
, Contributing Reporter | December 12, 2022
Artificial intelligence (AI) and machine learning in medical imaging are not only useful for diagnosis, but can also be used for ordering images, increasing image quality, reducing scan time and coding and billing.
During a session at the RNSA annual meeting in Chicago, Katherine Andriole, director of academic research and education at the Mass General Brigham and Brigham and Women's Hospital Data Science Office, described how her organization looked at using machine learning in appointment scheduling to predict the likelihood of and prevent patient no-shows. AI analyzed age, time from order to scheduled date, the type of exam and whether it required contrast and issues with weather, traffic and parking. It helped provide information on whether they could send reminder messages or help get them transportation to their appointment.
The team also worked to create a machine learning model to detect whether there was motion during a scan, which tells the technologist before the patient has left the exam room.
Ad Statistics Times Displayed: 62888
Times Visited: 855
Carestream Health is a leading provider of quality X-ray systems and detectors that are designed to maximize diagnostic confidence, workflow and patient satisfaction. Follow the link above to see our complete portfolio of digital radiography solutions.
“This isn't the diagnosis piece, but this automates some things that are very, very important for workflow,” Andriole said.
AI can also be used to take family history information from the electronic medical record and recommend imaging exams; for example, in breast MR, explained Dr. Nabile Safdar, vice chair of imaging informatics in the Department of Radiology and Imaging Sciences at the Emory University School of Medicine.
"The clinical decision support system could be using natural language processing ... to look at multiple diagnoses within a single patient, and other factors like genetics or family history, and then make recommendations for multiple exams based on that AI-generated patient profile," Safdar said. "And in a potential future state you can imagine doing this at the population level."
AI could also generate billable codes from the diagnosis and progress notes.
Dr. Nina Kottler, the associate chief medical officer for clinical artificial intelligence at Radiology Partners, described using AI for hanging protocols.
“If you had AI doing this that knew what your series looked like, it could automatically place them where you wanted, correctly, every time,” Kottler said.
Andriole said her institution found it was a good idea to have a governance committee for AI.
“We don't want this to be the ‘wild west’,” Andriole said.