Label-free investigation of infected acute pyelonephritis tissue by FTIR microspectroscopy with unsupervised and supervised analytical methods

Abstract

Acute pyelonephritis (AP) is a severe urinary tract infection (UTI) syndrome with a large population of patients worldwide. Current approaches to confirming AP are limited to urinalysis, radiological imaging methods and histological assessment. Fourier transform infrared (FTIR) microspectroscopy is a promising label-free modality that can offer information about both morphological and molecular pathologic alterations from biological tissues. Here, FTIR microspectroscopy serves to investigate renal biological histology of a rat model with AP and classify normal cortex, normal medulla and infected acute pyelonephritis tissues. The spectra were experimentally collected by FTIR with an infrared Globar source through raster scanning procedure. Unsupervised analysis methods, including integrating, clustering and principal component analysis (PCA) were performed on such spectra data to form infrared histological maps of entire kidney section. In comparison to Hematoxylin & Eosin-stained results of the adjacent tissue sections, these infrared maps were proved to enable the differentiation of the renal tissue types. The results of both integration and clustering indicated that the concentration of amide II decreases in the infected acute pyelonephritis tissues, with an increased presence of nucleic acids and lipids. By means of PCA, the infected tissue was linearly separated from normal ones by plotting confident ellipses with the score values of the first and second principal components. Moreover, supervised analysis was performed based on the supported vector machines (SVM). Normal cortex, normal medulla and infected acute pyelonephritis tissues were classified by SVM models with the best accuracy of 96.11% in testing dataset. In addition, these analytical methods were further employed on synchrotron-based FTIR spectra data and successfully form high-resolution infrared histological maps of glomerulus and necrotic cell mass. This work demonstrates that FTIR microspectroscopy will be a powerful manner to investigate AP tissue and differentiate infected tissue from normal tissue in a renal infected model system.