2.2 AI viewer

AI viewer layout

The standard AI viewer has a 2x1 layout, with the raw DICOM image and information on the left, inference results and information on the right.

The side bar at the far right contains inference result & calculation (optional), image slider at the bottom will appear when the study contains multiple images.

View tools include: contrast (WL/WW for CT images), pan, zoom & heatmap transparency (classification only)

Single-Label Classification:

Inference visualization: heatmap of the inference result
Inference result: class name and confidence score

Multi-Label Classification:

Inference visualization: heatmap of the inference result (can switch between classes with )
Inference result: class names and confidence score of each class that has conf. score >0.5

Object Detection:

Inference visualization: bounding boxes of the inference result (toggle on/off with )
Inference result: Name, object count & max. object width for each class that appears in the image

Object Segmentation:

Inference visualization: object masks of the inference result (toggle on/off with )
Inference result: Name & total area for each class that appears in the image. (Area calculated in pixels if no available spatial information)

Inference calculation

If at least one image is selected using the add button , a calculation table can be shown by clicking on where the number on the icon indicates the numbers of image selected. the calculation table contains different values for different types of models & setting as well.