The U.S. Food and Drug Administration (FDA) had regulated medical image analyzers for more than 20 years under the agency’s most stringent regulatory requirements. But the agency recently reclassified a subset of these device types, significantly decreasing regulatory burden and simplifying their path to market.
On January 22, 2020, the FDA reclassified medical image analyzers applied to mammography breast cancer, ultrasound breast legions, radiograph lung nodules, and radiograph dental caries detection from Class III to Class II. At a high level, the new device classification means that instead of fulfilling the standard of reasonable assurance of safety and efficacy through a premarket approval (PMA) application, a device of this type will need to satisfy the FDA’s substantial equivalence standard through a far less onerous 510(k) submission. Similarly, changes to these types of medical image analyzers can now generally be made within the Class II 510(k) framework.
In its reclassification order, the FDA explained that the relevant medical image analyzers are computer-assisted or aided detection (CADe) devices and not their computer-aided diagnostic (CADx) counterparts. For example, the downclassification would apply to a concurrent-read software device that shows CADe marks on the regions of interest intended to draw the clinicians’ attention during their standard review workflow. It would not, however, apply to software that provides quantitative measures of disease risk. Instead, this latter device would likely be considered radiological CADx software for lesions suspicious for cancer. The reclassification also would not apply to software that suggests prioritization of the cases in a review list/worklist for a clinician’s reading session, as this would likely be considered a computer-aided triage device. Although the radiological CADx software for lesions suspicious for cancer and computer-aided triage device are also regulated as Class II devices, they are subject to their own distinct device considerations and 510(k) data requirements, including special controls.
The special controls designated by the agency for the downclassified medical image analyzers include, but are not limited to, design verification and validation that includes a detailed description of the image analysis algorithms and pre-specified performance testing methods and dataset(s) used to assess whether the device will improve reader performance as intended, and results from performance testing that demonstrate that the device improves reader performance in the intended use population, as well as certain labeling requirements. Importantly, the agency considered but declined to apply special controls that would have required periodic retraining of clinicians and the collection of postmarket data.
The FDA’s recent downclassification of medical image analyzers, combined with its decision not to apply more demanding special controls, could indicate a willingness to reconsider how the agency regulates additional types of image analysis software. Indeed, in February, the FDA is hosting a two-day public workshop titled, “Evolving Role of Artificial Intelligence in Radiological Imaging,” that is intended to discuss emerging applications of Artificial Intelligence (AI) in radiological imaging including AI devices intended to automate the diagnostic radiology workflow as well as guided image acquisition.
