Genius AI Detection 2.0 with CC-MLO Correlation

The Genius AI Detection 2.0 with CC-MLO Correlation is a software utilizing deep learning to aid radiologists in identifying and characterizing suspicious lesions in digital breast tomosynthesis exams. Its core methodology involves analyzing DBT images for potential abnormalities, providing confidence scores, and uniquely correlating findings between CC and MLO views to enhance diagnostic workflow and accuracy.

The Genius AI Detection 2.0 with CC-MLO Correlation is a software device designed to assist clinicians in interpreting digital breast tomosynthesis (DBT) exams for breast cancer screening and diagnosis. It functions as a computer-aided detection and diagnosis (CADe/CADx) tool, utilizing deep learning algorithms to analyze DBT images to identify and mark suspicious regions of interest, such as soft tissue densities (masses, architectural distortions, asymmetries) and calcifications; for each detected finding, it provides a location, an outline, a confidence score, and an overall case score for the exam, with these CAD findings presented concurrently to the interpreting physician for immediate assessment. This latest version introduces the CC-MLO Correlation feature, a significant enhancement that correlates suspected lesions in one mammographic view with corresponding findings in the orthogonal view, aiming to improve physician workflow and navigation. The system processes 1-mm 3D DBT slices and can co-register findings to 6-mm SmartSlices; the CC-MLO correlation feature's effectiveness in lesion detection and inter-view correlation was established through standalone performance testing on a dataset of 106 malignant and 561 negative cases, evaluated against expert radiologist-defined ground truth.

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. How does the CC-MLO correlation algorithm handle cases with multiple lesions or highly dense breast tissue, and what is its reported accuracy in such scenarios compared to less complex cases?; What specific deep learning architectures are employed in Genius AI Detection 2.0, and how were they trained and validated to ensure generalizability across diverse patient populations and DBT systems?; Could the 'Certainty of Findings' and 'Case Score' be integrated with other clinical data (e.g., patient history, genetic markers) within the software to provide a more comprehensive risk assessment?; What were the quantitative performance metrics (e.g., sensitivity, specificity, area under the ROC curve) for the CC-MLO correlation feature specifically in the standalone evaluation with 106 malignant and 561 negative cases?; How does the concurrent display of CAD findings, particularly the correlated views, impact radiologist reading time and diagnostic accuracy in real-world clinical workflows compared to non-AI assisted reading or AI without correlation?