Abstract This study aimed to explain the non-destructive prediction of water fractions in Shiitake mushrooms (Lentinus edodes) using multispectral imaging (MSI) during far-infrared drying at 70 °C for different periods. Quantitative estimation of water fractions was obtained by combining MSI with chemometrics including partial least-squares (PLS), back propagation neural network (BPNN) and least squares-support vector machines (LS-SVM). Analysis of water fraction mobility of mushroom showed different properties in response of drying process. The proportion of each water fraction corresponds with extracted spectral data of MSI. After comparing the results of different chemometrics, PLS was good for the estimation of water fractions, which showed high accuracy with Rp values of 0.95, 0.92, and 0.83. Moreover, lowest RMSEP were estimated at 8.09%, 8.00% and 4.95% for free, immobilized, and bound water, respectively. The findings of this study demonstrated that MSI was an effective tool to monitor the water fractions in processed shiitake mushrooms.

中文翻译：

---

多光谱成像无损测定红外干燥香菇水分含量的有效性研究

摘要 本研究旨在解释使用多光谱成像 (MSI) 在 70 °C 下不同时期的远红外干燥过程中对香菇 (Lentinus edodes) 中水分含量的无损预测。通过将 MSI 与包括偏最小二乘法 (PLS)、反向传播神经网络 (BPNN) 和最小二乘支持向量机 (LS-SVM) 在内的化学计量学相结合，获得水分数的定量估计。对蘑菇水分流动性的分析表明，干燥过程具有不同的特性。每个水部分的比例对应于提取的 MSI 光谱数据。比较不同化学计量学的结果后，PLS 有利于估计水的分数，其 Rp 值分别为 0.95、0.92 和 0.83，显示出较高的准确度。此外，最低 RMSEP 估计为 8.09%、8。游离水、固定水和结合水分别为 00% 和 4.95%。这项研究的结果表明，MSI 是监测加工香菇中水分含量的有效工具。