Objective: Enhance the alignment for chest and lumbar X-rays using AI-generated view position and registration data.​

Feature Highlights:

AI View Position Inference: The model identifies the view position of each image as AP, PA, Left LAT, Right LAT, or Other.​

• Output is stored in a DSR file.​

• The original DICOM tag for view position remains unchanged.​

• This additional metadata supports downstream alignment logic without altering source data.

AI-Based Image Registration: The AI also outputs pan and rotate transformation data in the DSR file to align images for flickering.

• Viewer reads this data and applies transformations to improve image registration.​

• Supports alignment across current and prior studies for accurate comparison.

Viewer Integration:

• Viewer reads both the view position DSR and the registration DSR (including transformation matrix).​

• Applies transformations dynamically to enhance the flickering experience.

Objective: Enhance image registration and flickering accuracy by leveraging AI-generated bounding box and contour data.

 AI Integration: The viewer now utilizes bounding box and breast contour data provided via the DSR file.

Improved Alignment: We’ve introduced a new enhancement that automatically scales mammogram images to optimally fill the available viewport. This ensures that both current and comparison images are consistently aligned, without requiring manual zoom or pan adjustments.

Benefits:

• Use of Display Space: Images are now scaled to use the full viewport.

• Flicker Mode Registration: When using flicker display mode this auto-scaling ensures that both images are spatially aligned.


• How to activate this feature? Select the 'S' key to toggle the feature on and off

Visual Overlays: Bounding boxes and contours are displayed as overlays on the images for better visual context and verification.