How should  AI be used in breast ultrasound? In a recent online article, AuntMinnie.com reported on research presented by Leo Barinov, VP of Clinical Excellence & Applied Research at Koios Medical, Inc., at SIIM 2018. The research demonstrated significantly higher diagnostic performance when radiologists used AI concurrently during the interpretation process.

Researchers led by Lev Barinov of AI software developer Koios Medical, Princeton University, Rutgers University, and Robert Wood Johnson Medical School, found that three radiologists produced a significantly higher diagnostic performance when interpreting breast ultrasound exams concurrently with AI software than they achieved without the aid of the software. 

The researchers evaluated two breast ultrasound reading models: “second-read” and using AI concurrently. Using Koios™ DS software to review 500 pathology-proven breast ultrasound results, the researchers also compared the AI software’s performance with the performance of experienced radiologists reading of the same pathology data.

The study demonstrated that the AI software outperformed all three radiologists at baseline by an average of 11.5% in the area under the curve (AUC). Furthermore, the radiologists gained statistically significant improvement in performance when using AI concurrently.

Read the full article on AuntMinnie.com. Note: Free registration is required.