Study: iCAD's AI technology for digital breast tomosynthesis improves clinical performance, reading times

iCAD announced Thursday that a study on its digital breast tomosynthesis (DBT) cancer detection software showed "significant positive results" for clinical performance and workflow efficiency. The company stated that the results confirmed that the latest artificial intelligence (AI) software solution to support DBT improved reader sensitivity and specificity, and also cut reading times.

Jeff Hoffmeister, medical director at iCAD, said the technology has been shown "to significantly improve detection rates while reducing unnecessary recalls resulting from false positives." He noted that "radiologists are concerned about the extra workload and long reading times of tomosynthesis cases in everyday practice…[and] as tomosynthesis moves into the screening arena, our AI software provides important benefits to radiologists, their practices and their patients."

Specifically, the study showed that radiologists' reader sensitivity improved by 8 percent on average, while specificity improved by 6.9 percent. Additionally, when reading tomosynthesis cases with the solution, radiologists' reading times were reduced by 52.7 percent on average, the study found.

According to iCAD, the concurrent read, cancer detection and workflow solution for breast tomosynthesis "is trained to detect malignancies and determine the probability of malignant findings, providing radiologists with a 'certainty of finding' score for each case and each detected lesion." These scores signify the "algorithm's confidence that the detected soft tissue densities and calcifications are malignant," the company said.

In March, iCAD received the CE mark for its PowerLook Tomo Detection 2.0 AI solution, the company's second deep-learning-based, computer-aided detection solution for the interpretation of DBT data. The product is pending clearance by the FDA.

Did you like this article?