TruSPECT Processing Station

The TruSPECT Processing Station is a software-only medical device for accepting, displaying, storing, and processing nuclear medicine images (SPECT) to detect radioisotope tracer uptakes in patients. It includes visualization tools, quantitative analysis features for cardiac applications (QPS, QGS, QBS), and can integrate with CT/MR scans for diagnostic purposes. The device operates on Windows and can be configured with various clinical applications and post-processing software modules.

The modified TruSPECT maintains identical fundamental technological characteristics to its predicate device, including manual and automatic segmentation tools and support for multiple imaging modalities. The primary modification is the addition of the TruClear AI module, an artificial intelligence–based post-processing component designed to reduce noise in low-count D-SPECT myocardial perfusion images while using pre-trained neural networks in iterative reconstruction and traditional image processing algorithms. This does not alter the system's intended use, indications for use, or fundamental software architecture.

Testing was performed in compliance with NEMA PS3.1–3.20, IEC 62304, 21 CFR 820, and ISO 13485. Verification and validation activities included software verification, system integration and performance testing, simulated-use testing, and risk-based assessments. Clinical validation used FDA-cleared Cedars-Sinai QPS/QGS for quantitative parameter derivation and independent board-certified nuclear medicine physician review using 5-point Likert scales.

The proposed device has identical intended use and technological characteristics to the predicate (K212230), with the addition of TruClear AI not raising new safety or effectiveness concerns. Verification and validation testing confirmed the system meets all performance specifications, maintains the same safety and effectiveness profile, and operates within intended parameters. The AI module's quantitative performance demonstrated strong correlation (e.g., LVEF r=0.94, TPD r=0.98) to high-count reference images with results meeting pre-specified acceptance criteria.