Work

Multispectral sorting to reduce aflatoxin in bulk maize retains some efficacy when tested under less controlled field conditions

Year: 2026

Type: article

Abstract: Aflatoxin contamination in post-harvest maize threatens food safety and security in lower-resource settings. Our study used a pre-calibrated, small-scale, multi-spectral sorting machine based on visual characteristics previously associated with mycotoxin contamination of maize kernels to reduce aflatoxin under less controlled field settings than previously evaluated. Maize samples (n=45 each) were collected from Ghana and Kenya. A pre-existing linear discriminant analysis (LDA) classification algorithm was used to reject kernels with high risk of aflatoxin contamination. Five g subsamples of the accepted and rejected portions were analyzed for total aflatoxins using a lateral flow immunochromatographic assay and ELISA for Ghana and Kenya samples, respectively. After sorting, aflatoxin mean difference between the accept and the reject stream for Ghana was -0.28 log 10 (ng/g), paired t-test showed significant difference (p<2.2e-16) between the accept and reject levels for each individual sample. Kenya sample’s mean difference was -0.28 log 10 (ng/g) (p=0.08 by paired t test). The Ghana and Kenya samples had an average mass rejection of 37.9% (range 15.9-10.7) and 7.17% (range 1.10-25.2%), respectively, with average aflatoxin rejection of 65.5% (range 44.79-76.1) and 54.3% (range 10.2-99.97%), respectively. This shows that performing multi-spectral sorting under less controlled field conditions still retained efficacy to reduce aflatoxin in maize.

Source: Food and Humanity

DOI: 10.1016/j.foohum.2026.101314