The BoneView algorithm from Gleamer had already shown considerable effectiveness in identifying fractures, as evidenced by the award-winning research by Guermazi et al. at the latest RSNA. A fresh study published in the European Radiology by Lille University Hospital by Jacques et al. has sparked renewed attention by presenting convincing proof of BoneView’s adeptness in detecting hand and wrist fractures.
The innovative research compares radiologists’ sensitivity in identifying trauma on X-ray images, on the hand and wrist, with and without AI assistance. The ground truth for the study was set high, using the opinions of two radiologists corroborated by CT scans.
The results? BoneView has facilitated an increase in sensitivity for all bones, with an impressive 13.1% increase in detecting scaphoid fractures. This type of fracture, often elusive on X-rays due to the complexity of the wrist’s anatomy, is notorious for being challenging to diagnose.
The algorithm didn’t just shine in detecting one of the most intricate fractures. It also aided radiologists in elevating their overall sensitivity and negative predictive value without losing specificity. AI boosts the proficiency of less experienced radiologists to equal that of highly skilled colleagues in the absence of AI.
Employing CT scans to determine the reference standard for X-ray interpretation represents a novel approach. It’s intriguing to observe that AI stand-alone exhibits greater sensitivity compared to when radiologists use it in conjunction. This highlights the potential benefits of utilizing BoneView to detect occult or subtle fractures, demonstrating how Gleamer tools can be a complementary resource for imaging specialists.
Those interested can engage directly with Gleamer at the RSNA 2023 (booth 4373) or learn about BoneView, CE MDR, and FDA-approved fracture detection HERE.