QLAB Advanced Quantification Software

The QLAB Advanced Quantification Software 13.0, particularly its 3D Auto RV feature, aims to enhance the analysis of right ventricular function from ultrasound images by integrating advanced segmentation technology and machine learning algorithms. This approach enables semi-automated, dynamic quantification of Right Ventricle volumes and ejection fraction, with non-clinical studies demonstrating its accuracy and performance relative to predicate technologies and cross-modality imaging.

The QLAB Advanced Quantification Software 13.0 is a software application package designed for viewing and quantifying image data acquired from Philips ultrasound systems. A key modification in this version is the 3D Auto RV application, which focuses on providing dynamic clinical functionality for the Right Ventricle (RV). This application achieves this by integrating the segmentation engine of the QLAB HeartModel and the TomTec-Arena 4D RV-function, employing machine learning algorithms for the automatic segmentation of the RV to identify its endocardial contours. The system offers semi-automatic 3D surface model creation, which users can subsequently review, edit, or accept to refine the quantification. The 3D Auto RV application enhances Right Ventricle quantification by combining HeartModel's auto-segmentation technology with TomTec Arena’s 4D-RV algorithm for improved RV border placement and by introducing new RV measurement parameters. Non-clinical validation studies have demonstrated its performance capabilities. Internal testing revealed that the 3D Auto RV software generates RV end-diastolic volume error rates below 15% for all tested datasets when compared to the predicate 4D RV device. Furthermore, a root mean square error (RMSE) analysis showed that, in comparison to the predicate, the 3D Auto RV measured end-diastolic RV volumes with an 8.3 ml RMSE, end-systolic RV volumes with a 2.7 ml RMSE, and RV ejection fraction with a 2.7% RMSE. An external study published in the Journal of the American Society of Echocardiography corroborated these findings, showing accurate and reproducible quantification with less than 15% difference from cardiac magnetic resonance (CMR) measurements of RV volume, and confirmed that users could effectively identify and revise contours when necessary.

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 in 3D Auto RV adapt to variations in Right Ventricle morphology and image quality across different patient populations?; What specific training datasets were used for the machine learning model, and how does their diversity impact the generalizability of the 3D Auto RV application?; What is the quantitative impact on diagnostic accuracy and clinical decision-making when using the 3D Auto RV measurements compared to traditional 2D or manual 3D methods for Right Ventricle assessment?; How does the semi-automatic contour editing workflow in 3D Auto RV affect inter-observer and intra-observer variability compared to fully manual or fully automatic segmentation methods?; Beyond volume and ejection fraction, what other functional or morphological parameters of the Right Ventricle can the 3D Auto RV software accurately quantify or derive?