Detection of Rice Fungal Spores Based on Micro- Hyperspectral and Microfluidic Techniques

Area of Science:

• Agricultural Science
• Biotechnology
• Spectroscopy

Background:

• Protecting rice, a vital global food crop, from fungal diseases is crucial for agricultural productivity.
• Current methods for early-stage rice fungal disease diagnosis are limited, lacking rapid detection capabilities.

Purpose of the Study:

• To develop a novel method for early detection of rice fungal disease spores using a microfluidic chip combined with microscopic hyperspectral detection.
• To effectively separate, enrich, and classify spores of Magnaporthe grisea and Ustilaginoidea virens.

Main Methods:

• Design and fabrication of a microfluidic chip with a dual inlet and three-stage structure for spore separation and enrichment.
• Utilizing microscopic hyperspectral instrumentation to acquire spectral data from enriched fungal spores.
• Applying the Competitive Adaptive Reweighting Algorithm (CARS) for characteristic band selection and Support Vector Machine (SVM) and Convolutional Neural Network (CNN) for classification model development.

Main Results:

• The microfluidic chip achieved an enrichment efficiency of 82.67% for Magnaporthe grisea spores and 80.70% for Ustilaginoidea virens spores.
• The CARS-CNN classification model demonstrated superior performance, achieving F1-scores of 0.960 for Magnaporthe grisea and 0.949 for Ustilaginoidea virens.

Conclusions:

• The integrated microfluidic chip and hyperspectral detection system effectively isolates and enriches rice fungal spores.
• This approach offers a promising new strategy for the rapid and accurate early detection of rice fungal diseases, aiding agricultural management.