Rubin Jing , Yutao Lu , Zhiheng Zhang , Honghua Jiang , Bo Zhou , Wenjin Hou , Peiyu Xin , Huan Peng
Abstract
Quantifying nematode status and population dynamics before and after stress exposure is essential for drug screening and environmental monitoring. Current assessment workflows lack integrated automation solutions and still rely on manual microscopic observation after sampling and separation, a labor-intensive process that creates a throughput bottleneck for large-scale nematode control research. We present a six-channel automated detection system integrating a custom image acquisition device, the Nematode Efficient Identification and Detection (Nem-EID) algorithm, and a software platform. In terms of artificial intelligence contributions, we developed the Nem-EID algorithm, which leverages hardware-specific imaging characteristics and combines elliptical pre-screening with inter-frame Intersection over Union analysis to achieve accurate, real-time nematode identification and viability assessment. Regarding engineering application contributions, we constructed a fully automated microplate scanning imaging device based on an optical–mechanical co-design strategy, and built a multi-interface integrated management platform to support efficient device control and data visualization. Experiments on root-knot nematode samples achieved a counting accuracy of 91.7%, a mortality detection accuracy of 93.1%, and an average processing time of 0.832 ms per frame. The proposed system provides an efficient, low-cost, and reliable solution for non-destructive nematode monitoring, offering strong support for practical applications in drug screening and environmental monitoring.
Paper Linkage:https://doi.org/10.1016/j.engappai.2026.116612
Chinese