부산대학교 도서관

Detecting heavy metal contamination, such as cadmium, in seafood, particularly beche-de-mer (BDM), is very important for food safety and human health. However, conventional methods for detecting cadmium are often invasive, time-consuming, labor-intensive, and very expensive. So in this research, these challenges were addressed by utilizing hyperspectral imaging (HSI) combined with an Extreme Learning Machine (ELM) algorithm to predict cadmium concentrations in beche-de-mer. This approach provides a non-invasive, rapid, cheaper, and precise alternative for assessing heavy metal contamination. We applied the ELM model to hyperspectral data acquired from six beche-de-mer specimens, which were purchased from local markets in Indonesia, to derive spectral signatures correlating with cadmium concentrations. Our findings reveal a significant correlation between the hyperspectral data and the measured cadmium concentrations, achieving a prediction accuracy of $R_{P}^{2}=0.95186$ and $RMSEP=3.7802$. The results demonstrate the effectiveness of integrating HSI with advanced machine learning algorithms such as ELM in addressing detection of heavy metal contamination in seafood, especially in BDM.