Physiological mechanism-guided hyperspectral modeling for cadmium quantification in spinach (Spinacia oleracea L.) leaves.

Abstract

Achieving reliable and accurate detection of heavy metals in vegetables remains a critical challenge in food safety. While spectroscopic techniques enable rapid and nondestructive measurements, their practical application is often hindered by insufficient accuracy. Here, this study presents a hyperspectral imaging approach guided by the physiological response mechanisms of spinach to accurately quantify cadmium content in leaves. Cadmium-induced disruption of the chlorophyll-flavonoid system and impairment of cellular integrity established the physiological basis for selecting two characteristic spectral regions (388.34-430 nm and 1000-1036.34 nm). Leveraging these insights, we designed Cd-SpiNet with multiple convolutional branches to specifically extract cadmium-specific spectral features, thereby enhancing model sensitivity to cadmium-related signals. This physiologically guided strategy significantly enhances detection performance, achieving a coefficient of determination of 0.9753 and a root mean square error of prediction of 0.0287 mg/kg on the prediction set. The high accuracy is attributed to the cadmium-triggered physiological changes that directly influence spectral absorption and reflection characteristics within the selected bands. This study thus offers a novel strategy that integrates plant physiological mechanisms into spectral detection, providing a robust and intelligent solution for precise heavy metal monitoring in agriculture, which is crucial for timely risk assessment and effective management of ecological hazards.

Code and data availability

公開本文の所在を確認できませんでした。非公開または購読が必要な可能性があります。

No evidence-backed public reproduction asset is currently recorded.