Deep learning-based framework for cardiac function assessment in embryonic zebrafish from heart beating videos.

Area of Science:

• Cardiovascular research
• Developmental biology
• Biomedical engineering

Background:

• Zebrafish embryos are vital models for studying cardiovascular development and disease.
• Current methods for assessing zebrafish cardiac function are manual, time-consuming, and prone to variability.
• There is a need for automated, reliable tools to quantify cardiac indices in zebrafish.

Purpose of the Study:

• To develop and validate an automated framework for zebrafish cardiovascular assessment.
• To enable efficient and consistent quantification of cardiac function in zebrafish embryos.
• To provide a widely applicable tool for researchers using zebrafish models.

Main Methods:

• Developed the Zebrafish Automatic Cardiovascular Assessment Framework (ZACAF) utilizing a U-net deep learning model.
• Applied ZACAF to microscopic videos of wildtype and cardiomyopathy mutant zebrafish embryos.
• Validated the framework's performance against manual assessment methods.

Main Results:

• The ZACAF achieved over 90% accuracy in assessing cardiovascular indices like ejection fraction (EF) and fractional shortening (FS).
• The framework operates on standard, low-frame-rate (5-20 fps) black and white microscopic videos.
• Demonstrated potential for high-throughput, consistent analysis of large video datasets.

Conclusions:

• The Zebrafish Automatic Cardiovascular Assessment Framework (ZACAF) offers an efficient, accurate, and reliable method for quantifying zebrafish cardiac function.
• This automated approach overcomes limitations of manual analysis, facilitating large-scale studies.
• ZACAF is broadly applicable across laboratories with standard equipment, supporting diverse research collaborations.