ClearView cCAD

ClearView cCAD is an image analysis software designed to assist physicians by automatically classifying shape and orientation characteristics of selected regions in breast ultrasound images using multivariate pattern recognition. The system operates as a web service, processing DICOM images and leveraging novel descriptors to support ACR BI-RADS assessment, with its performance in discerning these parameters demonstrated to be statistically equivalent to experienced radiologists in non-clinical testing.

ClearView cCAD is a software application designed to assist skilled physicians in the analysis of breast ultrasound images. It functions by automatically classifying the shape and orientation characteristics of user-selected regions of interest (ROIs), employing multivariate pattern recognition methods and novel statistical, texture, shape, and orientation descriptors to characterize lesions from DICOM compliant digital medical files queried from ultrasound devices or PACS servers. Users select an ROI, and the software analyzes it, displaying results for automated reporting, storage, or modification, thereby aiding compliance with the ACR BI-RADS® ultrasound lexicon classification form. Operating as a platform-agnostic web service compatible with Windows 7 or higher systems, ClearView cCAD notably does not perform automatic lesion segmentation, a step deemed unnecessary for its classification accuracy when compared to predicate devices. Its web-based client architecture, distinct from standalone applications, includes mitigations to ensure image analysis occurs on original full-resolution images. Non-clinical bench testing, involving over 1200 cases evaluated against three MQSA certified radiologists, demonstrated that ClearView cCAD achieved overall accuracy for BI-RADS based shape and orientation assessment statistically equivalent to the radiologists.

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 do the 'novel statistical, texture, shape, and orientation descriptors' used by ClearView cCAD differ from standard image analysis algorithms, and what is their specific contribution to its performance in BI-RADS classification?; Given that the system does not perform automatic lesion segmentation, what are the potential implications for workflow efficiency and inter-observer variability in ROI selection by different users?; What specific 'mitigations were taken to ensure that all image analysis was performed on the original full resolution images' for the web-based client, and how does this address potential concerns about image fidelity compared to standalone applications?; The non-clinical performance data showed statistical equivalence to radiologists; were there specific types of lesions or BI-RADS categories where the software performed exceptionally well or where its performance was less aligned, and what might account for these variations?; How does the system handle images with poor quality or significant artifacts, and does its performance degrade predictably in such scenarios when assessing lesion characteristics?