The US Food and Drug Administration (FDA) approved computer-aided detection and diagnostic software called OsteoDetect to speed the diagnosis of wrist fractures in adults, the agency announced yesterday


Autodetect, from Imagen, in New York City, uses an artificial intelligence algorithm to detect fractures in the distal radius by analyzing standard x-ray images. The software then marks the image for further review by a clinician.

The software is designed for use in multiple settings, including primary care, emergency departments, urgent care, and specialty clinics, according to the FDA press release. "It is an adjunct tool and is not intended to replace a clinician's review of the radiograph or his or her clinical judgment," an FDA news release states.

Approval was based on two retrospective studies that included 1200 cases. In one study, the software's detection rate was compared with that of three board-certified orthopedic hand surgeons on 1000 cases.

"Both studies demonstrated that the readers' performance in detecting wrist fractures was improved using the software, including increased sensitivity, specificity, positive and negative predictive values, when aided by autodetecting, as compared with their unaided performance according to standard clinical practice," the agency release states.

De novo premarket review pathway

An agency spokesperson told Medscape Medical News there was no advisory committee meeting before the approval, which was provided via the de novo premarket review pathway that is used for novel low- to moderate-risk devices.

This is the second FDA marketing approval for artificial intelligence–based device. The first device, IDx-DR, was approved in April for use in the detection of diabetic retinopathy in adults. It was designed for use by nonspecialists.

A request for comment from the company as to whether there are any ongoing prospective trials of the product or plans to determine whether the software improves patient outcomes or the delivery of healthcare services was not answered before publication.