MIRA: an open source and user-friendly software to automate counting and sizing of fungal spores

Background: The precise quantification of fungal spores is crucial in phytopathology, serving as a key metric for inoculum standardization and disease severity assessment. Traditionally, manual hemocytometry using tools like the Malassez slide has been the most accurate method for spore counting, but it is labor-intensive, time-consuming, and susceptible to operator variability.

To address these challenges, researchers have developed MIRA (Microscopy Image Recognition & Analysis), an open-source software that utilizes YOLO deep learning algorithms for automated spore counting and sizing.

MIRA boasts a user-friendly graphical interface and is compatible with a variety of camera systems, supporting advanced object detection models such as YOLOv11 and YOLOv26. The software's capabilities include accurate spore detection and counting, automated measurement of spore surface area, and differentiation of spores across genera.

Comparative analysis: To evaluate MIRA's performance, researchers conducted an extensive comparative study using Pyricularia oryzae spores as a representative example. When loaded with the pre-trained MIRA model, the software demonstrated an impressive correlation (R = 0.96) with the gold-standard manual counting methods (Malassez and Kova) while reducing processing time for high-concentration samples (10 spores/mL) by over 90%.

Furthermore, MIRA successfully tested custom YOLO models specifically designed to identify macro- and microconidia of Fusarium oxysporum f. sp. cubense, a model for Pseudocercospora fijiensis, and a multi-class model recognizing six different rice pathogenic fungi.

Ease of use and accessibility: To facilitate adoption by researchers without programming expertise, MIRA provides comprehensive tutorials for operating the software and training custom detection models using Roboflow and Google Colab. The software is available as both open-source Python code and standalone executables for Windows and Linux platforms.

Conclusions: MIRA represents a significant advancement in spore counting techniques by offering a rapid, accurate, and highly reproducible alternative to manual methods. The integration of cutting-edge YOLO-based deep learning algorithms with an intuitive interface and extensive training resources empowers researchers to effectively automate image analysis tasks.

By addressing a major bottleneck in phytopathology workflows, MIRA enhances the efficiency of high-throughput disease phenotyping and has the potential to be adapted for a wide array of microscopic quantification tasks in various biological disciplines.