An AI-Assisted Protocol for Quantifying Superficial Chorioallantoic Membrane Vasculature

Area of Science:

• Vascular biology
• Bioimaging
• In vivo models

Background:

• The chicken chorioallantoic membrane (CAM) is a valuable in vivo model for studying angiogenesis and tumor growth.
• Distinguishing superficial and deep vascular networks in the CAM is challenging with conventional 2D imaging.
• Existing AI methods for vascular quantification often require extensive preprocessing or user-specific model training.

Purpose of the Study:

• To develop an AI-assisted protocol for quantitative analysis of the superficial vascular plexus in the CAM.
• To simplify and standardize the analysis of CAM vasculature for angiogenesis and anti-angiogenesis studies.
• To enable 2D quantification of superficial CAM vasculature without custom AI model training.

Main Methods:

• An AI-assisted protocol was established to focus vascular quantification on the superficial CAM capillary plexus.
• Bovine whipped cream was injected into the CAM to create an optical mask, isolating the superficial vasculature.
• A U-Net-based model was employed for automated 2D quantification of vessel parameters like area, length, and thickness.

Main Results:

• The segmentation model achieved a Dice similarity coefficient of 0.831.
• The study demonstrated remodeling of superficial CAM vasculature at embryonic days 11 and 15, with increased branching and altered perfusion.
• The workflow simplified preprocessing and allowed standardized analysis using a pre-trained segmentation model.

Conclusions:

• The novel method combines experimental layer isolation with automated segmentation for 2D quantification of superficial CAM vasculature.
• This approach eliminates the need for custom AI model training, making it more accessible.
• The protocol supports practical, reproducible CAM angiogenesis studies through a standardized analysis workflow.