The deep learning hybrid may particularly help less-experienced readers or trainees spot subtle femoral neck breaks, experts wrote in the Journal of Digital Imaging.

H. Lee Moffitt Cancer Center and Research Institute scientists used low-dose CT and chest x-ray imaging data from the National Lung Screening Trial to create their model.

The agreement affords Life Image’s network of 58,000 global clinics—including primary and comprehensive stroke centers—access to RapidAI’s data-driven neuroimaging technology.

Typically, distinguishing between tumor types involves immunohistochemical sample staining and complex genetic analysis, often a day-long process.

An automated system to predict knee replacement surgery risk is critical, given that more than half of the estimated 14 million Americans suffering from knee osteoarthritis will undergo the procedure.

With further refinement, the researchers say their convolutional neural network-based platform could help lighten sonographers’ ever-increasing workloads.

The technique—known as 3D cartilage surface mapping—detects subtle changes in a person’s knee joint that cannot be picked up via conventional x-ray or MRI.

With such a model, clinicians can implement preemptive measures for higher risk patients, experts wrote in Academic Radiology.

The algorithm's sensitivity edged out that of trained radiologists at identifying signs of the disease on chest CT scans, Mount Sinai researchers explained recently.

If validated further, the algorithm could be used to flag urgent scans in radiology workflows, especially in resource-strapped regions.

With further testing and validation, Hyungjin Kim, with Seoul’s National University College of Medicine and colleagues believe radiologists may one day use the tool to individualize treatment and achieve better outcomes.

"Using this new tool may uncover the characteristics of different types of clots that were previously unrecognized by humans," researchers wrote in the journal eLife.