Minimal PySide6 app for streaming Clarius ultrasound and showing two views:

• Original ultrasound image
• Processed image with YOLO detection or UNet segmentation

sidecaster.py imports the UNet model from UNet.py.

sidecaster.py treats its parent directory as the asset directory:

<asset-dir>/
├── libraries/
│   ├── cast.dll              # Windows
│   ├── libcast.so            # Linux
│   └── pyclariuscast.so      # Linux
├── models/
│   ├── yolo_x_phantom_best.pt
│   ├── unet_phantom_latest.pth
│   └── training_parameters.txt   # optional
└── <app-dir>/
    ├── sidecaster.py
    ├── UNet.py
    └── src/                  # optional guidance icons

• Python 3
• PySide6
• NumPy
• Ultralytics
• PyTorch
• Clarius Cast library files
• Clarius scanner on the network

pip install PySide6 numpy ultralytics torch

python sidecaster.py

Default connection:

IP:   192.168.1.1
Port: 5828

Click Connect, then Run.

• YOLO / UNet – choose detection or segmentation
• ROI – limit processing to the centre of the image
• UNet Threshold – adjust the segmentation threshold
• RL / AP / SI – show measurements on the processed image
• Record Boundaries – plot UNet boundary points in 3D when IMU data is available
• Save Local – save current original and processed images

Saved images:

~/Pictures/clarius_original_image.png
~/Pictures/clarius_processed_image.png

• YOLO and UNet are mutually exclusive in the processed view.
• UNet uses models/unet_phantom_latest.pth and the architecture in UNet.py.
• If the app connects but no image appears, check firewall and network settings.