YUAN to Showcase Real-Time AI Imaging Workflows at NVIDIA GTC Berlin

Two live demonstrations combine NVIDIA IGX T7000, NVIDIA Holoscan Sensor Bridge, and YUAN video capture technology to connect raw ultrasound data with edge AI inference and multi-source video processing.
YUAN today announced it will showcase two live demonstrations at NVIDIA GTC Berlin, integrating NVIDIA IGX T7000, NVIDIA Holoscan Sensor Bridge, and YUAN’s multi-source video capture technology. The demonstrations will present an end-to-end raw-to-insights workflow—from high-bandwidth ultrasound data transfer to real-time AI inference and low-latency video fusion—highlighting opportunities for AI-enabled medical imaging development. As developers integrate AI into medical imaging systems, bringing high-volume sensor data and multiple video sources into a coordinated processing pipeline become increasingly important. YUAN brings together video capture, sensor connectivity, and system integration to connect data acquisition with accelerated computing at the edge.
Connecting Raw Ultrasound Data With Real-Time AI Inference
The first demonstration, “Ultrasound Raw-to-Insights,” streams raw data from a Texas Instruments (TI) ultrasound AFE5932EVM evaluation module through an Altera FPGA-based HSB implementation with 100GbE connectivity directly to the IGX T7000 for real-time AI processing.
The NVIDIA IGX T7000 board kit unlocks real-time sensor processing and AI reasoning for physical AI agents and robotics with unprecedented performance, enterprise software, and functional safety. Powered by an NVIDIA Blackwell Architecture iGPU and optional dGPU, it delivers up to 5581 FP4 TFLOPS of AI compute to effortlessly run multiple generative AI models at the edge.
Combining high-bandwidth data streaming with IGX’s powerful AI computing, the demonstration shows how raw ultrasound data can flow into an integrated inference pipeline, providing a foundation for real-time imaging analysis applications.
Bringing Multiple Live Video Sources into a Unified Workflow
The demonstration brings multiple live camera feeds into a unified workflow using YUAN’s SC750N101 capture card. Featuring an onboard hardware video processor with image signal processing (ISP) capabilities, the SC750N101 supports sub-frame latency output and failover functionality. It also connects to Holoscan Sensor Bridge via PCIe, enabling efficient transfer of RAW data and YUV video streams to the system. The demonstration highlights low-latency processing, synchronization, and data fusion across multiple video streams. A flexible video switching architecture enables different imaging sources to operate within a unified, resilient pipeline, supporting real-time visualization and AI application development.
For more information on YUAN’s computing and sensor connectivity solutions, explore the SmartVDO Near workstation family and YUAN HSB series.
Visit YUAN at NVIDIA GTC Berlin 2026
Hall 7 | Booth 7030
Messe Berlin
