Page 192 - Vibrational Spectroscopic Imaging for Biomedical Applications
P. 192

168    Cha pte r  S i x


     Widefield Raman Image of Tissue
             2930 cm –1
    Pre-Instrument Response Correction
                                     Overall Average
                               60
                               50

                               40

                               30
                     5 lm      20

                (a)
                                   800 1200  1600  2000  2400  2800 3200
     Widefield Raman Image of Tissue      Raman Shift (cm )
                                                      –1
             2930 cm –1
   Post-Instrument Response Correction
                                     Overall Average
                               60
                               50

                               40
                               30
                               20

                (b)
                                   800 1200  1600  2000  2400  2800 3200
                                                      –1
                                          Raman Shift (cm )
   FIGURE 6.3  (a) Widefi eld Raman image and mean image spectrum prior to
   instrument response correction, (b) the same fi eld of view after correction with
   instrument response.


        6.4.6 Baseline Correction
        Once instrument response has been eliminated, the remaining fluo-
        rescence is removed from the background. While the photobleach-
        ing process eliminates the fluorescence masking the Raman signal,
        background fluorescence still exists, evident in the sloping baseline.
        A common method for correcting the sloping baseline is using a
        polynomial fit.
            A polynomial fit may be completed in various ways. Often times
        the spectral range is selected first upon which the polynomial will be
        fitted. This is done if the existent errant slope would alter the outcome
        of the fit. Points are then selected on the spectrum along with a poly-
        nomial order. For example, end points and a first order polyno-
        mial will result in a linear baseline fit. This example is used in
   187   188   189   190   191   192   193   194   195   196   197