Novel convolutional neural network for bacterial identification of confocal microscopic datasets

Area of Science:

• Microbiology
• Bioinformatics
• Computer Science

Background:

• Artificial intelligence (AI) is increasingly used for rapid data analysis.
• Confocal microscopy generates complex biological image data.
• Accurate and efficient bacterial identification is crucial in diagnostics and research.

Purpose of the Study:

• To develop a novel deep-learning algorithm, CM-Net, for classifying bacterial species from confocal microscopy images.
• To automate and accelerate the process of microbial identification.
• To enhance the accessibility of advanced bacterial analysis for non-expert users.

Main Methods:

• A deep-learning algorithm, CM-Net, was developed.
• Image augmentation techniques were used to increase dataset size to 7066 images (224x224 dimensions).
• The dataset included images of Escherichia coli and Staphylococcus aureus, augmented and fed into CM-Net for training and testing with 5-fold cross-validation.

Main Results:

• CM-Net achieved high performance across seven metrics, including 96.08% accuracy, 95.98% sensitivity, and 96.19% specificity.
• The algorithm processed bacterial identification results in just 8.9 minutes.
• The model demonstrated high reliability and accuracy in classifying bacterial species.

Conclusions:

• CM-Net offers a significant advancement in automated bacterial identification using AI.
• The algorithm drastically reduces analysis time and minimizes human error in microbial identification.
• CM-Net empowers non-expert personnel to perform accurate microbial identification, simplifying laboratory workflows.