Viz RV/LV

The Viz RV/LV software device employs an artificial intelligence algorithm to analyze volumetric CTPA scans, automatically identifying and measuring the maximal diameters of the heart's right and left ventricles to calculate their ratio. It provides clinicians with annotated DICOM images and a summary report displaying these measurements, intended as an adjunctive tool rather than a standalone basis for clinical decision-making in the assessment of right ventricle enlargement.

The Viz RV/LV is a software-only medical device that utilizes a locked artificial intelligence machine learning (AI/ML) algorithm to analyze volumetric computed tomography pulmonary angiogram (CTPA) scans. Its primary function is to automatically identify the right and left ventricles of the heart, measure their maximal diameters, and report the ratio of these measurements. This assists clinicians, including radiologists, pulmonologists, and cardiologists, in the assessment of potential right ventricle enlargement. The software processes CTPA scans forwarded to Viz.ai’s Backend Server and is not intended for standalone diagnostic use, but rather as an adjunctive tool to support clinical assessment. Upon analysis, Viz RV/LV generates results in DICOM format, specifically an Annotated Image Series and an RV/LV Summary Report. The annotated images provide visual overlays on scan slices, highlighting maximum ventricular diameters and the interventricular septum. The summary report details these maximal diameters, their lengths, and the calculated RV/LV ratio. Clinical performance testing demonstrated a mean absolute error (MAE) of less than 7.2 mm when comparing the algorithm’s measurements to a ground truth established by trained radiologists. The device's performance was stratified across various parameters like clinical site, patient demographics, and scanner characteristics, based on studies with distinct training and pivotal datasets.

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 AI algorithm's performance in measuring ventricular diameters vary across different CT scanner manufacturers and models not included in the pivotal study?; What is the range of CTPA slice thicknesses for which the Viz RV/LV software maintains its stated accuracy (MAE < 7.2 mm)?; Are there specific patient populations or cardiac pathologies (e.g., congenital heart defects, significant cardiac remodeling) where the algorithm's accuracy in identifying ventricular boundaries might be compromised?; How does the processing time of the Viz RV/LV software scale with the size and resolution of the input CTPA dataset, and what are the typical processing times in a clinical workflow?; Beyond the RV/LV ratio and maximal diameters, does the software provide any additional quantitative metrics or qualitative assessments of ventricular morphology from the CTPA scans that could be clinically useful?