MYCOLab Ai Logo - gold and army green

Deep learning application to hyphae and spores identification in fungal fluorescence images

Summary

Researchers developed an artificial intelligence system using two deep learning models to automatically identify fungal infections in microscope images. The system analyzes fluorescence-stained samples to detect fungal spores, hyphae, and mycelium with accuracy matching experienced doctors. This automated approach can significantly reduce the time clinicians spend examining samples and help prevent misdiagnosis, especially in hospitals with fewer specialist technicians.

Background

Fungal infections, particularly invasive fungal diseases, pose significant diagnostic challenges due to lack of rapid and efficient diagnostic techniques. Traditional microscopy-based diagnosis requires skilled clinicians and is prone to misinterpretation. Deep learning has shown promise in medical image analysis but remains understudied in fungal identification from fluorescence images.

Objective

To develop and evaluate a dual-model deep learning framework combining YOLOX and MobileNet V2 models for automated detection of fungal hyphae, spores, and mycelium in clinical fluorescence images. The study aims to create a system that achieves clinician-level accuracy in fungal diagnosis.

Results

The YOLOX model achieved 85% precision for hyphae and 77% for spores, with recall values of 90% and 85% respectively. The MobileNet V2 model achieved 83% precision with 100% recall for mycelium detection. The combined framework achieved 92.5% precision, 99.3% recall, and a Kappa value of 0.857 in fungal positivity detection, demonstrating high agreement with clinician evaluations.

Conclusion

The proposed dual-model framework successfully identifies fungal hyphae, spores, and mycelium in fluorescence images with clinician-level accuracy. The integration of YOLOX and MobileNet V2 models provides a comprehensive and effective automated diagnosis system that can significantly reduce workload and improve diagnostic efficiency in clinical practice.
  • Published in:Scientific Reports,
  • Study Type:Validation Study,
  • Source: PMID: 40715216, DOI: 10.1038/s41598-025-11228-y
Scroll to Top