Towards Lifespan Automation for Caenorhabditis elegans Based on Deep Learning: Analysing Convolutional and Recurrent...

Area of Science:

• Biotechnology
• Computational Biology
• Genomics

Background:

• Automating Caenorhabditis elegans (C. elegans) lifespan assays presents challenges.
• Issues include occlusions, dirt, worm aggregation, and difficulty in determining worm viability due to minimal movement in later life stages.

Purpose of the Study:

• To develop an automated method for C. elegans lifespan assays.
• To improve the accuracy and efficiency of determining worm viability in standard Petri dishes.

Main Methods:

• Combined traditional computer vision with deep learning models (CNNs and RNNs) for live/dead C. elegans classification.
• Utilized low-resolution image sequences for analysis.
• Employed data augmentation techniques to train neural networks with limited sample sizes.

Main Results:

• Achieved low error rates (3.54% ± 1.30% per plate) compared to manual lifespan determination.
• Demonstrated the feasibility of the proposed automated method.
• Successfully addressed challenges like occlusion and aggregation.

Conclusions:

• The proposed method offers a feasible and accurate approach to automate C. elegans lifespan assays.
• The integration of computer vision and neural networks, along with data augmentation, provides a robust solution.
• This automation can significantly aid in aging research and drug screening.