Lunit INSIGHT DBT v1.1

Lunit INSIGHT DBT v1.1 is an AI-powered CADe/x software designed to assist radiologists in identifying and characterizing suspicious breast lesions in digital breast tomosynthesis (DBT) exams. It leverages deep learning and an updated AI engine trained on expanded data to enhance diagnostic performance, providing details like lesion type, location, and abnormality scores, and has demonstrated effectiveness and safety comparable to predicate devices.

A regulatory body has reviewed the Lunit INSIGHT DBT v1.1, a software product designed for radiological computer-assisted detection and diagnosis of suspicious breast lesions. This review concluded that the device is substantially equivalent to already marketed products for its specified uses, thereby permitting its commercialization. Marketing the device is subject to various general regulatory controls, including requirements for annual registration, device listing, adherence to good manufacturing practices, and accurate labeling. Further regulations, such as quality system requirements and unique device identification rules, also apply to the product. Lunit INSIGHT DBT v1.1 is a computer-assisted detection and diagnosis (CADe/x) software that aids interpreting physicians in identifying and characterizing suspected breast cancer lesions in digital breast tomosynthesis (DBT) exams. The software employs artificial intelligence and deep learning to analyze DBT images, providing details like lesion type, location, and abnormality scores. This v1.1 version represents an enhancement over its predecessor, incorporating an updated AI engine and expanded training data, which has led to improved diagnostic performance. Non-clinical testing and standalone performance studies confirmed its effectiveness and safety, demonstrating that it is comparable to the predicate device in detecting suspicious lesions.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. What specific metrics were used to assess the "improved diagnostic performance" of v1.1, and how do they compare quantitatively to the previous version and the predicate device?; How does the Lunit INSIGHT DBT v1.1 integrate into existing clinical workflows, and what are the requirements for data input and output formats?; Beyond identifying and characterizing lesions, does the software offer any features for tracking lesion progression over time or for risk stratification?; What was the demographic and pathological diversity of the expanded training data, and how does it ensure the software's generalizability across different patient populations?; Are there any known limitations or specific types of breast lesions or image artifacts where the Lunit INSIGHT DBT v1.1's performance might be suboptimal?