Multimodal multisample spectroscopic imaging analysis of tumor tissues using multivariate curve resolution

Chemical imaging aims to characterize the molecular patterns present in a tissue or surface. High-resolution images with high content of molecular information represent a clear advantage in many research areas, such as in biomedicine. Specific molecular patterns, when related to pathological traits, could help understanding the molecular events related to the condition studied. Chemical composition of the different regions within the tissue may reveal interesting patterns that cannot be detected using conventional staining techniques. In this work, a new chemometrics approach is proposed for the simultaneous multimodal chemical imaging analysis of six tissue slices from two types of tumor tissues (EG and 1070) exhibiting different sensitivity to the docetaxel drug. Therefore, a deeper chemical imaging study of these tissues could reveal new chemical fingerprints related to the different susceptibility and response to this drug treatment. The spectral imaging techniques used in this study were MALDI-TOF mass spectrometry imaging (MSI) and FT-IR hyperspectral imaging. In addition, a RGB color picture of the hematoxylin/eosin staining was also included in the joint image analysis. The huge spectral data information generated by these multiple imaging techniques in the simultaneous analysis of the different tissue samples was organized in combined multiset and multimodal spectral image data structures which were then analyzed by multivariate curve resolution. Using this approach, the nature of the chemical constituents and their high-resolution spatial distribution were detangled showing what were the main chemical patterns present in specific areas of the tissues which distinguished the two types of tumors. In summary, in the present work, the application of multivariate curve resolution is confirmed to be a powerful and flexible data analysis tool able to extract chemical information from the simultaneous analysis of huge, fused image data sets from obtained by different type of spectroscopic imaging techniques having different spatial resolution powers.