A neural cell automated analysis system based on pathological specimens in a gerbil brain ischemia model

Area of Science:

• Neuroscience
• Computational Biology
• Pharmacology

Background:

• Growing interest in natural products for health due to milder effects compared to medical drugs.
• Limited quantitative systems exist for assessing the impact of natural products on living organisms.
• Ischemia poses a significant health challenge, necessitating effective therapeutic strategies.

Purpose of the Study:

• To develop and validate a deep-learning system for automated cell counting in a gerbil model.
• To quantitatively assess the effectiveness of natural products against ischemia using automated cell counting.
• To provide a more efficient method for evaluating natural products compared to manual assessment.

Main Methods:

• Utilized a deep-learning model, specifically fine-tuned Detectron2, for image analysis.
• Images were acquired from paraffin-embedded gerbil brain tissue.
• The system's performance was benchmarked against expert visual judgment.

Main Results:

• The automated cell counting system demonstrated a 79% positive predictive value.
• The system achieved 85% sensitivity when compared to expert assessments.
• The system successfully evaluated hydrogen water's potential efficacy against ischemia.

Conclusions:

• The developed deep-learning system offers a promising and efficient approach for evaluating natural products, particularly those with milder effects.
• The system's findings on hydrogen water's anti-ischemic potential align with expert evaluations.
• Automated cell counting addresses the need for large datasets in natural product evaluation, mitigating the labor-intensive nature of manual measurements.