A new approach to concordance in mid-infrared spectromicroscopy mapping of malignant tumors

Abstract

Mid-infrared spectromicroscopy studies on biological tissue sections require accurate identification of tumor-bearing areas in histology-stained and infrared-unstained tissue sections. Concordance was achieved as follows: paired stained and unstained thin (5 microm) human brain tumor cryosections mounted on slides were scanned with a Nikon Coolscan 4000 film scanner at 4000 dpi, edited with Adobe Photoshop CS2 software, and both digital images saved. A digital tractile grid, developed in our laboratory, was overlaid onto both images. Boundaries of tumor-containing areas in stained sections were identified by light microscopy, and a digital boundary map constructed. The map was transferred onto the unstained spectromicroscopy tissue image, and finally layered onto the gridded, equisized, spectromicroscope-generated overview image prior to Fourier transform infrared spectromicroscopy. Accurate identification of tumor-bearing areas, normal brain tissue and transitional zones allowed for meaningful interpretation of respective spectral patterns in detecting subtle differences within biochemical profiles. This is the first reported method of a standardized technique for ensuring concordance in mapping of malignant tumors by mid-infrared spectromicroscopy. This technique is applicable to all biological thin tissue sections, and serves to enhance accuracy of concordance between globar- and synchrotron-light generated infrared data with that obtained by conventional light microscopy.