Neurologica, the U.S. subsidiary of Samsung, has scored FDA clearance for its Auto Lung Nodule Detection tool, which is designed to use AI to detect pulmonary nodules between 10 and 30 mm in size.

The computer-assisted detection solution is expected to be an aid to physicians when reviewing PA chest radiographs in adults. It is part of S-Station, an operation software installed on Samsung Digital X-ray imaging scanners. "ALND serves as a second reader, on-device AI algorithm. It will help identify nodules that may otherwise have been missed or ambiguous. Early detection is key regarding treatment and survival rates, and ALND allows you to have more diagnostic confidence due to its deep learning algorithm," Lynne Gagne, marketing director at Samsung NeuroLogica, and the clinical research team told HCB News.

The solution indicates the location of suspected lung nodules on chest X-ray images and provides an option known as Autorun to automatically perform nodule detection immediately after scanning. It also provides PACS transmission options that correlate to simplify workflow.

Samsung is collaborating with Vuno, an AI developer in the healthcare space, on further enhancing the solution, and will expand the uses of its chest CAD solution, while continuing to improve diagnostic accuracy and workflow.

Clinically verified in multiple university hospitals, the deep-learning technology has an approved sensitivity of 80% or more and has retrospectively identified 600 chest radiographs with lung cancer and 200 normal chest radiographs. Universities that have tested the tool include Freiburg University Hospital in Germany; Massachusetts General Hospital in Boston; Samsung Medical Center in Seoul, South Korea; and Severance Hospital, also in Seoul.

Additionally, the Radiological Society of North America has extensively tested and validated it in clinical settings. It found the solution performed effectively when used on an "unprecedented" diversity of images acquired under different conditions and demographics. It only produced a small number of false positives per image (0.15). These clinical evaluations showed significant statistical increases in reader performance for nodule detection.

"Currently, there is no commercially available algorithm which 'continuously learns' after deployment. Along with technical difficulties in achieving this, there are also regulatory boundaries which such algorithms must abide. Although, therefore, ALND is a fixed algorithm at the moment, we do have plans to expand the number of detectable radiographic findings in the future in collaboration with VUNO," said Gagne and the team.

Vuno’s Med-Chest X-ray software was recently integrated on Samsung Electronics’ mobile digital X-ray system, GM85, to instantly retrieve and deliver AI results upon scanning an image while the patient is still in the hospital.

NeuroLogica Corp. also, just this month, introduced the V8, a high-end ultrasound system that provides enhanced image quality, usability and convenience for ultrasound professionals. It is equipped with NerveTrack, an AI technology that detects the location of nerves during live scanning, and UterineAssist, an AI application that detects tissue changes and assists the user with measurements of the uterus.

X-Ray Homepage