Cleerly Labs v2.0

Cleerly Labs v2.0 is a web-based software application designed to assist medical professionals in analyzing cardiac CT data for the detection and quantification of coronary artery plaques and stenosis, aiding in the evaluation of Coronary Artery Disease. The device employs machine learning for semi-automated segmentation of coronary arteries and plaque, which users can edit, combined with subsequent mathematical calculations for measurements, and presents findings in a structured CORONARY Report with enhanced visualization and PACS integration.

Cleerly Labs v2.0 is a web-based software application developed for trained medical professionals. Its primary purpose is to serve as an interactive tool for viewing and analyzing cardiac computed tomography (CT) data, specifically from Coronary Computed Tomography Angiography (CCTA) scans. The software assists in determining the presence and extent of coronary plaques, identified as atherosclerosis, and stenosis in patients being evaluated for, or suspected of having, Coronary Artery Disease (CAD), by providing tools for measurement and visualization of coronary arteries. The methodology of Cleerly Labs v2.0 involves post-processing of CT images, utilizing machine learning, including deep learning, to identify high-quality images and perform semi-automatic segmentation of coronary arteries, lumen, and vessel walls, thereby enhancing user efficiency. Users can manually edit these segmentations, adjust plaque thresholds, and demarcate features like stenosis and stents. Subsequent plaque, stenosis, and vessel measurements are derived from mathematical calculations based on this user-editable segmentation. Key enhancements in v2.0 include the CORONARY Report for intuitive visualization of findings, the ability to select coronary anatomical findings, and secured network server integration for data transfer with PACS systems. Performance data from its predicate indicated good correlation with expert readers for volumetric measurements of lumen, vessel, and various plaque types, with core algorithms remaining unchanged in this version.

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 machine learning algorithm's performance in segmentation vary across different CT scanner models and image acquisition protocols, and what are the implications for measurement accuracy?; What specific rule-based mathematical calculations are used for plaque, stenosis, and vessel measurements after user edits, and how sensitive are these outputs to minor variations in user segmentation?; The CORONARY Report is described as 'visually interactive and comprehensive'; what specific interactive elements does it offer, and how have these been validated for usability and clarity in conveying complex findings to clinicians?; What are the defined 'high quality images' criteria that the deep learning methodology uses for identification, and how does the software handle or flag images that do not meet these criteria?; The performance data (Table 6) shows varying Pearson Correlation Coefficients for different plaque types (e.g., 0.53 for Low-Density-Non-Calcified Plaque Volume); what is the clinical impact of this variability, particularly for less correlated parameters?