The New Era of Medical AI: Advantech AIR-510 Hands-on with NVIDIA Holoscan & MONAI—Exploring the Future of Medical Imaging!

Introduction: Medical Image Analysis, Now Smarter Than Ever!

#

Do you remember when doctors used to spend countless hours searching for clues in one CT or ultrasound image after another? With breakthroughs in AI technology, medical image analysis is evolving rapidly! This time, the Advantech team conducted real-world tests of NVIDIA’s latest Medical AI solutions—Holoscan and MONAI—on the AIR-510 platform, making medical analysis faster, more accurate, and more intelligent. Curious about how we achieved this? Join us as we unveil the secrets behind this experiment!

Background & Technology Overview: AI + Medical Imaging—Empowering Doctors with Enhanced Insight

#

Today’s healthcare institutions face overwhelming demands for image interpretation, whether it’s whole-body CT scans, cardiac ultrasounds, or endoscopic examinations. All require efficient and precise support tools. NVIDIA Holoscan and MONAI form a super-combination in modern medical AI:

• MONAI Label: An open-source medical imaging AI platform that automatically assists doctors in annotating and segmenting various organs or lesions, reducing manual workload.
• NVIDIA Holoscan: A platform optimized for real-time medical image processing, supporting multiple AI models to enable medical devices to make instant decisions.
• Advantech AIR-510: Our in-house high-performance AI computing platform, providing robust support for medical AI applications.

With the global surge in demand for medical AI applications, our primary goal in this experiment is to truly implement AI technology and enhance medical efficiency!

Experiment Process & Key Findings: Hands-on Experience with AI-powered Medical Image Processing

#

We focused on two main topics: “Automatic Whole-body CT Segmentation” and “Real-time Medical Image Recognition.” Here’s a look at our workflow and exciting discoveries!

1. MONAI Label Whole-body CT Automatic Segmentation (TotalSegmentator)

#

Step-by-step Experiment Record

#

1. Start the Backend Server

   • First, run the following commands in the CMD window to launch the MONAI Label service:

     cd /home/eiotae/Downloads
     source ~/miniconda3/bin/activate
     conda activate monai
     monailabel start_server --app apps/monaibundle --studies datasets/Task09_Spleen/imagesTs --conf models wholeBody_ct_segmentation
     

   • Once the screen below appears, you can proceed (do not close this window)
     

     

2. Launch the 3D Slicer Visualization Platform

   • In a new CMD window, run:

     cd /home/eiotae/Downloads/Slicer-5.6.2-linux-amd64
     ./Slicer
     

   • Click on “MONAI Label” on the screen
     

     

3. Connect to the MONAI Service

   • Enter the URL displayed in the first CMD window (e.g., http://0.0.0.0:8000)
     

     

4. Load and Run AI Segmentation

   • Click “Next Sample” to load a sample image
     

     

   • Click “Run” to start AI segmentation
     

     

5. 3D Structure Display and Operation

   • Click “Show 3D” to display the 3D segmentation results
     

     

   • Right-click the icon in the upper right corner and select “Center view” to adjust the perspective
     

     

   • Done!
     

     

Key Highlights

#

• The operation process is simple, and the AI can automatically segment multiple organs and structures.
• The 3D visualization results are clear, helping doctors quickly grasp the distribution of lesions.
• The entire workflow runs on the Advantech AIR-510, demonstrating powerful AI inference capabilities.

2. Holoscan Real-time Medical Image Recognition DEMO

#

We further tested three major Holoscan AI medical imaging applications and found their real-time performance and versatility impressive!

Entering the Holoscan Docker Environment

#

cd /home/eiotae/Downloads/holohub
sudo  ./dev_container launch

Three Major AI Application Demos

#

• Automatic Cardiac Ultrasound Measurement

  ./run launch multiai_ultrasound cpp
  

  • The model automatically annotates four key linear cardiac indicators (right ventricular diameter, interventricular septal thickness, left ventricular diameter, posterior wall thickness).
  • Accurately assists doctors in interpreting cardiac structural abnormalities.
    

    

    
    Watch the live demo video

• Real-time Colon Polyp Segmentation

  ./run launch colonoscopy_segmentation cpp
  

  • The AI model instantly segments polyps in colonoscopy images, enhancing early detection rates.
    

    

    
    Watch the live demo video

• Endoscopic Surgical Instrument Tracking

  ./run launch multiai_endoscopy cpp
  

  • Utilizes an LSTM model to track surgical instrument movements, aiding in surgical training and safety monitoring.
    

    

    
    Watch the live demo video

Results & Applications: Advantech Makes Medical AI Implementation Simpler

#

This experiment showcased the powerful integration of Advantech AIR-510 with NVIDIA Holoscan and MONAI. Not only does it allow doctors to easily benefit from AI assistance, but it also enables healthcare institutions to:

• Accelerate Diagnostic Workflows: Automatic segmentation and real-time measurements reduce manual labor time.
• Enhance Diagnostic Accuracy: AI-assisted annotation and detection lower the risk of missed diagnoses.
• Promote Smart Healthcare Adoption: Applicable to radiology, cardiology, gastroenterology, and operating room scenarios.
• Highly Flexible and Scalable: The platform supports various AI models and image formats for future upgrades.

Compared to traditional medical image analysis, AI-powered healthcare environments are more efficient, precise, and aligned with clinical needs.

Conclusion & Future Outlook: Continuous Innovation, Bringing AI Healthcare to Every Hospital

#

Advantech continues to invest in the integration of medical AI hardware and software. The results of the AIR-510 hands-on tests prove that we not only align with top international AI platforms but can also rapidly deploy solutions at the medical frontline. Moving forward, we will keep optimizing platform compatibility, advance real-time AI applications, and actively collaborate with medical institutions to develop more innovative solutions.

The future of medical AI is already here. Advantech will stand by your side at the forefront, working together with partners to bring smart healthcare closer to every patient’s needs!

Reference Links & Technical Resources

• Official MONAI Label Example
• Holoscan Official DEMO & Documentation
• For all operational screenshots and tutorials, refer to the images and descriptions in the article

Advantech — Continuous innovation, making AI healthcare accessible every moment!