MPXA-2000

MPXA-2000 is a software application designed to assist clinicians by providing automated quantitative analysis of coronary arteries from X-ray angiographic images using deep learning and computer vision algorithms. It aims to improve the efficiency and reproducibility of assessing coronary artery disease in adult patients by automatically segmenting vessels, classifying them, and calculating key diagnostic parameters.

MPXA-2000 is a stand-alone cardiovascular image analysis software designed for the viewing and quantification of X-ray angiographic images of coronary arteries in adult patients with Coronary Artery Disease (CAD). Its primary purpose is to provide validated and reproducible quantified results to support clinical decision-making in diagnosis and treatment evaluation, though it is not intended as the sole basis for clinical diagnosis. The software utilizes a combination of deep learning algorithms for automatic vessel segmentation and classification (identifying LAD, LCX, and RCA vessel types) and computer vision algorithms for precise measurement of vessel dimensions from DICOM-formatted images. Key functionalities include image uploading, frame selection, automated analysis of regions of interest (ROIs), and visualization and export of these results. The MPXA-2000 system processes X-ray angiographic images to calculate metrics such as percent diameter stenosis, minimum lumen diameter (MLD), and reference diameters. Validation was performed using images from X-ray angiography systems by Philips Medical Systems, Siemens Healthineers, and GE Healthcare. Standalone performance testing on 305 coronary angiograms from a U.S. population demonstrated high accuracy, with a vessel classification accuracy of 0.9934 and a vessel segmentation Dice Similarity Coefficient (DSC) of 0.9200. The software's estimation of minimum luminal diameter was found to be comparable to its predicate device, QAngio XA. The analysis results can be exported in PDF or Excel formats for clinical review by cardiologists and 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 does the deep learning model in MPXA-2000 account for variability in image quality and patient anatomy to maintain its reported vessel segmentation accuracy of 0.9200 DSC across diverse clinical scenarios?; What specific features or algorithms contribute to MPXA-2000's ability to automatically classify LAD, LCX, and RCA vessel types with an accuracy of 0.9934, and how robust is this classification in atypical anatomies or poor contrast images?; Given that MPXA-2000 does not include stent analysis or ostium segmentation like its predicate QAngio XA, what is the clinical impact or workflow adjustment for users who previously relied on these features for comprehensive coronary assessment?; How does the real-time analysis capability of MPXA-2000, leveraging deep learning and computer vision, integrate into existing catheterization laboratory workflows, and what is the typical processing time per study?; Beyond the reported Dice Similarity Coefficient and classification accuracy, what are the performance metrics for specific quantitative measurements like Minimum Lumen Diameter or percent stenosis when compared against a gold standard, such as QCA performed by expert core labs, especially in complex lesion morphologies?