uAI Portal

uAI Portal is a medical image management and processing software that utilizes artificial intelligence algorithms to provide automatic segmentation and analysis tools for various CT angiography studies, including coronary, aorta, pulmonary artery, head/neck, and lower extremity vessels. Its purpose is to aid clinicians in the viewing, manipulation, evaluation, and storage of these medical images, with its segmentation performance validated through Dice coefficient comparisons against radiologist-established ground truth across diverse datasets.

uAI Portal is a software solution designed for processing, reviewing, and analyzing CT studies, facilitating the viewing, manipulation, communication, and storage of medical images, including DICOM 3.0 formatted functional and anatomical datasets across various time-points or frames. It supports interpretation and evaluation within healthcare settings through multiple display formats like VR, MIP, and MPR, and offers specialized applications for Lower Extremity Vessel Analysis (CTA), Head and Neck Vessel Analysis (CTA), Coronary Analysis (CCTA), Pulmonary Artery Analysis (CTPA), and Aorta Analysis (CTA), enabling clinicians to view, manipulate, and evaluate specific imaging datasets for their respective anatomical regions. The system's core technology involves artificial intelligence algorithms for automatic segmentation of anatomical structures from CT images, facilitating reconstruction and analysis. It operates by parsing DICOM images, applying these algorithms in memory to generate segmented results which are then stored and can be interactively reviewed by clinicians who can perform operations like measurements and save or print results. Performance verification, using 150 US images and validated against ground truth established by radiologists, demonstrated high accuracy for its AI-based segmentation, with Dice coefficients for key structures like Coronary Artery vessels (average 0.920) and Aorta trunk (average 0.946) meeting predefined acceptance criteria and showing consistency across demographic and imaging subgroups.

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 performance of the AI segmentation algorithms in uAI Portal vary when encountering rare anatomical variants or complex pathologies not extensively represented in the training/testing datasets?; What are the specific differences in segmentation output (e.g., vessel-only vs. vessel-plus-organ) between uAI Portal's Aorta and Pulmonary Artery analysis applications and the predicate device's combined CT Vessel Analysis, and how do these differences impact clinical workflow or diagnostic interpretation?; Given that algorithm training was conducted on images from China and verification on US images, what steps were taken to ensure the robustness and generalizability of the AI algorithms across diverse patient populations and imaging equipment beyond those explicitly tested?; The document mentions "Basic Documentation Level" for software. What specific cybersecurity measures are implemented within uAI Portal to protect patient data integrity and privacy, especially concerning network transfer and database storage?; How does the "user-friendly solution for reconstructing medical images" manifest in terms of processing speed and user interaction time for typical cases in each of the five analysis applications, and how does this compare to manual methods or the predicate device?