uCT ATLAS Astound with uWS-CT-Dual Energy Analysis; uCT ATLAS with uWS-CT-Dual Energy Analysis

The uCT ATLAS Astound and uCT ATLAS are computed tomography X-ray systems that produce cross-sectional images of the whole body for head, cardiac, vascular, and general body imaging. They are also intended for low-dose CT lung cancer screening to detect lung nodules that may represent cancer. The devices include uWS-CT-Dual Energy Analysis software that processes CT images acquired at different tube voltages to visualize material composition based on energy-dependent attenuation coefficients.

The subject devices add six new software functions to the predicate device (K231482): CardioBoost (deep learning cardiac reconstruction), AIIR (modal-based iterative reconstruction with deep learning), CardioXphase optimized (AI-based coronary detection for optimal phase selection), Motion Freeze (deep learning head motion artifact reduction), Ultra EFOV (deep learning-enhanced field-of-view extension), and optimized CardioCapture (AI motion correction for cardiac imaging). These differences involve dataset augmentation and deep learning network optimization rather than fundamental changes to scanning or reconstruction methodology.

IEC 61223-3-5 for general image quality performance metrics including CT HU number and thickness section testing; CTIQ White Paper guidance for low contrast detectability testing; AAPM report guidance for high contrast spatial resolution testing. Reader studies evaluated image quality using five-point scale assessments of noise, artifact reduction, structure fidelity, and diagnostic benefit.

The devices are substantially equivalent because they maintain identical intended use and indications for use as the predicate device (K231482), and the new software functions represent optimizations of existing deep learning-based reconstruction and analysis methods rather than new safety or effectiveness concerns. Bench testing confirms the devices meet IEC 61223-3-5 image quality requirements and show equivalent or improved performance versus predicate comparisons (FBP, KARL 3D). Reader studies validate that images are diagnostically sufficient with equal or better quality than predicate comparisons. From the user workflow perspective, operation remains unchanged, and the differences do not raise new safety or effectiveness issues.