Detection of Cytopathic Effects Induced by Influenza, Parainfluenza, and Enterovirus Using Deep Convolution Neural...

Area of Science:

• Virology
• Computational Biology
• Medical Diagnostics

Background:

• Virus identification in clinical samples traditionally relies on observing cytopathic effects (CPEs) in cell cultures.
• This method is time-consuming and requires specialized expertise for accurate interpretation of cell morphology changes.

Purpose of the Study:

• To develop and evaluate an artificial intelligence (AI) based approach for efficient and accurate identification of viruses through CPE analysis.
• To compare the performance of single-task and multi-task deep learning models for virus identification.

Main Methods:

• Utilized ResNet-50 as a deep learning backbone for analyzing CPEs induced by influenza, enterovirus, and parainfluenza viruses.
• Implemented and compared single-task and multi-task learning models, incorporating multiplexer and de-multiplexer layers for enhanced performance on known cell lines.

Main Results:

• Achieved high accuracies with single-task (97.78%) and multi-task (98.25%) learning models.
• The multi-task model demonstrated improved accuracy (97.13%) even with limited parainfluenza CPE data, outperforming the single model (95.79%).

Conclusions:

• A deep learning framework using ResNet-50 and multi-task learning effectively identifies virus-induced CPEs.
• The proposed AI models offer a promising, efficient, and accurate alternative to traditional methods for virus detection.