MIDS-PNAS aided lung nodule detection on CT in China's review

What happened

Deepwise built MIDS-PNAS, software that uses deep learning to detect lung nodules of 4 mm or more on chest CT and analyze their features. China's Center for Medical Device Evaluation (CMDE) reviewed the company's application for Class III registration as an innovative device. The trial was the maker's. It was prospective and randomized, and doctors read cases with and without the software in a crossover design. The consensus of three senior chest radiologists served as the reference standard. The report says that doctors' sensitivity for nodules and for patients met clinical needs with the software and that their specificity did not fall. It gives no figures, and it reports no result for reading time, a secondary endpoint. CMDE judged that the product “在临床性能满足临床使用需求的基础上，提高阅片效率” (improves reading efficiency, provided that its clinical performance meets clinical needs). On 23 November 2020, it found that the benefits outweigh the risks and recommended registration. It asked the maker to keep studying how the algorithm generalizes and to review clinical use every year for its next registration.