An Optical Smartphone-Based Inspection Platform for Identification of Diseased Orchids

Area of Science:

• Plant pathology
• Spectroscopy
• Artificial Intelligence

Background:

• Orchid viruses, Odontoglossum ringspot virus and Cymbidium mosaic virus, cause significant economic losses in orchid farming.
• Current diagnostic methods like immunoassays are costly, time-consuming, and labor-intensive, relying on limited sampling.

Purpose of the Study:

• To develop a rapid, noncontact, and convenient platform for detecting orchid viral infections.
• To utilize a spectrometer, smartphone, and AI for efficient disease screening.

Main Methods:

• A noncontact inspection platform integrating a spectrometer and an Android smartphone was developed.
• An Internet of Things and AI-based Android application processed fluorescence spectrum data.
• A remote server-hosted algorithm, trained on optical and PCR data, determined infection status.

Main Results:

• The platform provided infection status determination within 3 seconds.
• The AI algorithm achieved a testing accuracy of 89%.
• The area under the receiver operating characteristic curve was 91%, indicating high diagnostic performance.

Conclusions:

• The developed platform is accurate and convenient for screening orchid infections.
• This noncontact method offers a promising alternative to traditional diagnostic techniques for managing orchid diseases.