Multispectral imaging-based detection of Acidovorax citrulli: from colony identification to infested seed discrimination

Abstract

Abstract Bacterial fruit blotch (BFB) caused by Acidovorax citrulli , is a destructive seed-transmitted disease that seriously threatens global cucurbit production. To address the need for detecting A. citrulli -infested seeds, this study developed a colony identification model and a seed infestation detection model based on multispectral imaging. The combined nMahalanobis and nCDA colony identification models achieved a high recall of 0.999 and a low false-positive rate of 0.149 when tested on samples. For infested melon seed detection, we evaluated and compared the classification performance of seven machine learning models. The results showed that LDA, logistic regression, and MLP exhibited stable performance on artificially infested seed samples. Furthermore, multi-cultivar modeling improved model generalizability and demonstrated the feasibility of using multispectral imaging to identify naturally infested seeds. When a qPCR Ct threshold of 37 was used to define seed infestation status, the logistic regression model achieved a validation accuracy of 0.82. Overall, these findings demonstrate the potential of multispectral imaging for colony identification and seed infestation detection, providing a new technical approach and a scientific basis for seed health testing of bacterial fruit blotch in cucurbit crops.

Plant phenotyping relevance

マルチスペクトル画像と機械学習により、感染種子という植物器官の状態を検出する手法を開発・比較・検証しており、表現型取得が研究の中心である。

Code and data availability

The supplied preprint blocks describe multispectral imaging of A. citrulli colonies and infested melon seeds with machine-learning models, but contain no data availability statement, no public dataset or image deposit, and no author code repository or URL. Models were built with proprietary VideometerLab software and a

No evidence-backed public reproduction asset is currently recorded.