Imaging Informatics

Nicole Murphy, MS, a medical physicist at Northwestern Memorial Hospital in Chicago, and Christina Sammet, PhD, research assistant professor of Radiology at the Northwestern Feinberg School of Medicine and medical physicist at Ann and Robert H. Lurie Children's Hospital of Chicago, targeted three main objectives in relation to radiation dose management at RSNA 2017. 

Artificial intelligence (AI) is here to stay in radiology—and so are radiologists.

An analysis of nearly three million radiologic exams has confirmed prior research showing that physicians’ concentration tends to fall off toward the tail end of on-duty shifts. And yes, the diminishment in radiologists’ accuracy may be increased when they’re working especially long shifts and/or plowing through long worklists.

Free-text radiology reports can be automatically classified by convolutional neural networks (CNNs) powered by deep-learning algorithms with accuracy that’s equal to or better than that achieved by traditional—and more labor-intensive—natural language processing (NLP) methods.

Researchers in the radiology department at the University of California, San Francisco (UCSF)—led by of Sabrina Ronen, PhD, director of the Brain Research Interest Group (RIG) and professor in the department of radiology and biomedical imaging at UCSF—are in the process of developing new, non-invasive imaging biomarker indicators to address multiple types of cancer, according to a recent UCSF press release. 

Many radiologists use Twitter and LinkedIn for staying up on matters related to their work. A study published online Nov. 12 in the Journal of the American College of Radiology shows they’d do well to tap, for the same purposes, the social-media platform that’s commonly thought of as a purely personal online space.

A group of German researchers has developed a nuclear medicine test that can detect infections in kidney transplant tissue, according to a study published in Journal of Nuclear Medicine. 

Members of the online radiology community, take note: Personally tweeting links to articles posted ahead of print in online medical journals doesn’t increase overall pageviews of these articles. It just increases the number of people who find their way to any given “article in press” via Twitter.

Stanford researchers have developed a deep-learning neural network model that can determine the bone age of children from a hand radiograph about as accurately as both an expert radiologist and an existing software package that uses a feature-extraction approach and has been cleared for clinical in use in Europe.

A medical AI startup is offering its image-interpretation algorithms for a flat $1 per read.

Health informaticists at UC-San Francisco have tracked down the source of many cases of Clostridium difficile infection that had vexed a UCSF hospital. The clues—there for the piecing together in the EMR—led to a CT scanner in the emergency department.

Magnetic resonance enterography (MRE) could be the new gold standard for children presenting obscure gastrointestinal bleeding (OGIB).