Ceevra Reveal 3+

Ceevra Reveal 3+ is a medical image processing software that transforms CT and MR scans into interactive 3D anatomical models for enhanced preoperative planning and intraoperative visualization. It utilizes machine learning for automated segmentation in adults and provides quantitative measurement tools, with performance validated through rigorous testing on diverse patient data and phantom models.

Ceevra Reveal 3+ is an advanced software medical device designed to assist healthcare professionals by generating and displaying interactive three-dimensional (3D) images derived from patient CT and MR scans. The system involves two primary functions: company personnel use the software to create these 3D visualizations, which clinicians subsequently access and interact with via dedicated mobile or desktop applications, external displays, or compatible virtual reality (VR) headsets. This interactive visualization capability supports preoperative surgical planning and enables intraoperative display of anatomical models, allowing clinicians to rotate, zoom, pan, selectively show or hide structures, and notably, utilize newly incorporated features for quantitative measurements like volume, diameter, and inter-structural distances. The software leverages machine learning (ML) and computer vision algorithms to perform preliminary segmentation of normal anatomy, specifically for adult patients aged 22 and older; images for younger individuals or those of unknown age are processed without ML involvement. The robustness of its four ML models was extensively validated using 141 distinct patient imaging studies, ensuring data diversity in terms of patient demographics and scanner manufacturers (GE, Siemens, Toshiba, Philips), and maintaining strict separation between training and testing datasets. Performance assessments, using Sørensen–Dice coefficient (DSC) and Hausdorff distance (HD-95), yielded strong results for various anatomical structures, including prostate (0.87 DSC), bladder (0.90 DSC), neurovascular bundles (7.8 mm HD-95), kidneys (0.87-0.89 DSC), and major arteries and veins (0.81-0.87 DSC). Furthermore, the accuracy of the quantitative measurement features was confirmed through tests on phantom data and actual patient studies, demonstrating a mean difference within +/- 10%.

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 machine learning segmentation algorithms in Ceevra Reveal 3+ vary across different CT/MR scanner manufacturers and imaging protocols beyond those tested?; What is the specific methodology used by Ceevra personnel to generate the 3D images, and what level of manual intervention or quality control is involved before clinicians access them?; For the new quantitative measurement features, what is the range of error observed, and how does this compare to manual measurements or other established measurement tools in clinical practice?; How does the intraoperative display functionality integrate into existing surgical workflows, and what is the latency or update speed when interacting with 3D models in real-time during a procedure?; Given that machine learning is only used for patients 22 and over, what are the differences in image quality, segmentation accuracy, or utility for preoperative planning when ML-assisted segmentation is not employed for younger patients?