Surface Science Laboratories, Toray Research Center, Inc, 3-3-7, Sonoyama, Otsu, Shiga, 520-8567, Japan.
Faculty of Science and Technology, Seikei University, 3-3-1 Kichijoji-kitamachi, Musashino, Tokyo, 180-8633, Japan.
Anal Bioanal Chem. 2022 Jan;414(2):1177-1186. doi: 10.1007/s00216-021-03744-3. Epub 2021 Nov 3.
Time-of-flight secondary ion mass spectrometry (TOF-SIMS) is a useful and versatile tool for surface analysis, enabling detailed compositional information to be obtained for the surfaces of diverse samples. Furthermore, in the case of two- or three-dimensional imaging, the measurement sensitivity in the higher molecular weight range can be improved by using a cluster ion source, thus further enriching the TOF-SIMS information. Therefore, appropriate analytical methods are required to interpret this TOF-SIMS data. This study explored the capabilities of a sparse autoencoder, a feature extraction method based on artificial neural networks, to process TOF-SIMS image data. The sparse autoencoder was applied to TOF-SIMS images of human skin keratinocytes to extract the distribution of endogenous intercellular lipids and externally penetrated drugs. The results were compared with those obtained using principal component analysis (PCA) and multivariate curve resolution (MCR), which are conventionally used for extracting features from TOF-SIMS data. This confirmed that the sparse autoencoder matches, and often betters, the feature extraction performance of conventional methods, while also offering greater flexibility.
飞行时间二次离子质谱 (TOF-SIMS) 是一种用于表面分析的有用且多功能的工具，可获取各种样品表面的详细成分信息。此外，在二维或三维成像的情况下，通过使用团簇离子源，可以提高高分子量范围内的测量灵敏度，从而进一步丰富 TOF-SIMS 信息。因此，需要适当的分析方法来解释这种 TOF-SIMS 数据。本研究探讨了稀疏自动编码器（一种基于人工神经网络的特征提取方法）处理 TOF-SIMS 图像数据的能力。将稀疏自动编码器应用于人皮肤角质形成细胞的 TOF-SIMS 图像，以提取内源性细胞间脂质和穿透外部的药物的分布。将结果与传统上用于从 TOF-SIMS 数据中提取特征的主成分分析 (PCA) 和多变量曲线分辨率 (MCR) 的结果进行比较。这证实了稀疏自动编码器匹配，并且通常优于传统方法的特征提取性能，同时也提供了更大的灵活性。
