Dear Artificial Intelligence Insider,
Artificial intelligence (AI) is showing promise as a potential tool for helping to decrease the workload of radiologists in reading screening mammograms.
In a pair of studies presented at ECR 2020, researchers outlined how AI software could triage a significant percentage of normal mammograms, cutting the workload of radiologists by more than half without affecting the cancer detection rate.
Full AI automation is arguably the only way to make medical imaging as safe as possible, wrote Ross Upton, PhD, in a guest editorial. In 2019, Upton was one of Forbes' 30 under 30 in Europe's science and healthcare sector, so as he is a rising star, his opinions are worth noting.
Meanwhile, researchers from the Karolinska Institute in Sweden have found that radiologists have a wide range of sensitivity for detecting cancer on mammograms. The authors believe that the findings make the case for the use of computer-aided detection software for breast screening.
A freely available AI algorithm can yield comparable diagnostic performance to radiologists for assessing the likelihood and severity of COVID-19 on noncontrast CT studies in patients suspected of having the disease. In addition, a deep-learning model was recently found to improve radiologist performance in identifying hip fractures on radiographs.
A Swiss team has reported that the combination of AI and radiomics can be a valuable approach to predicting if a particular treatment strategy will be successful for acute ischemic stroke. The results could potentially pave the way for personalized endovascular patient management, according to the group. In addition, AI and radiomics can also be highly accurate for distinguishing between glioblastoma and brain metastasis on standard T1-weighted MRI, making it easier to quickly diagnose these brain lesions.
It can be difficult to keep track of all of the radiology AI software applications that are commercially available in Europe. To help, a group from Radboud University Medical Center in the Netherlands has created a website that aims to provide detailed overviews of all CE-marked AI software for radiology.
More involvement by radiologists and more assistance from data scientists are needed to improve the performance of AI algorithms and expand their use in clinical practice, according to a Spanish expert.
Is there a story you'd like to see covered in the Artificial Intelligence Community? Please feel free to drop me a line.