Univariate imaging refers a group of imaging methods which utilize single spectral parameters such as absorbance, or intensity values,
band positions (frequency values), band widths, or combinations thereof. While multivariate imaging techniques are based on parameters
generated by complex chemometric procedures (e.g. cluster membership values in cluster analyses, or principal component scores in PCA),
univariate parameters can be generally derived directly from the spectra.
Help is available for the following functions of the 'Univariate Imaging' menu bar:
chemical imaging also known as functional group mapping chemical movie - plays a movie of successive chemical images, or frequency slices frequency imaging - univariate imaging based on frequency / wavenumber positions FWHM imaging - imaging using full half width at half maximum (FWHM) values
Video tutorial univariate imaging with CytoSpec (Youtube):
Chemical Imaging
Chemical Imaging (also known as functional group mapping): This function permits producing chemical images, which can be considered color-scaled
representations of univariate spectral parameters such as absorbances, or intensity values at a given wavenumber, or frequency position.
In chemical imaging, parameters are color-encoded and plotted according to the spectra's spatial [x,y] coordinates. In the CytoSpec implementation
color coding is done according to the selected colormap. For univariate chemical
imaging, the following options are available:
Method A: calculates the area below the spectral curve in the region between points P1 and P2 (by integration, no baseline correction)
Method B: obtains a simple absorbance, or spectral intensity value at a given wavenumber, or frequency position P1
Method C: calculates the area formed by the linear conjunction line of points P1 and P2 and the spectral curve in the region between both
points.
Method D: obtains a baseline corrected absorbance, or spectral intensity value at a given at a given wavenumber, or frequency position P5.
The baseline is formed by the linear conjunction line of points P1 and P2
Methods A-D/E-H: calculate ratios between absorbance, or spectral intensity values obtained by methods A-D and E-H, respectively.
The dialog box 'Chemical Imaging'
Select first the imaging method. You can choose between methods A-D by activating the appropriate radio button. If you wish
to use ratios for chemical imaging you have to additionally select the 'ratio' button and the method in the
denominator (methods E-H).
Next, type the wavenumber values P1-P6 used as integration borders, or frequency points for construction baselines. Note
that the number of values depends on the integration method.
Select the data block on which chemical imaging should be carried out. The following data blocks can be used for chemical
imaging: original data (i), preprocessed data (ii), derivatives (iii), and deconvolution data (iv). CytoSpec returns an
error message in cases where the selected source data block is empty.
Button 'plot': starts the 'chemical imaging' function. When finished the functions plots a chemical image into the
respective axis of original, or processed data by color-encoding the calculated values.
Button 'cancel': Closes this dialog box.
Note that chemical images produced from the source data block of original data will be displayed in the upper right panel of CytoSpec's
main gui, whereas chemical images re-assembled from preprocessed, derivatives, or deconvolution data are plotted in the axis to the
lower left. Note also that images obtained by multivariate hyperspectral imaging approaches (i.e. cluster, PCA, MCR-ALS ANN, or endmember
extraction methods) are also displayed in this panel.
Mouse clicks by the left mouse button into chemical images result in the following actions (if 'show mode' is active):
the [x,y] coordinate of the mouse pointer is obtained and displayed in the editable text boxes of the main figure,
the corresponding spectrum is extracted from the hyperspectral map and is plotted in the spectra panel located at the left, and
information of how the chemical image was produced is displayed in the area between image panels to the right. Please
refer to the section Main Window - Basics Concepts for more details.
The function 'chemical movie' can be used to play an animation of chemical images, or frequency slices. With this function the
system automatically steps through a range of frequency values, facilitating judgment of the spatial distribution patterns of individual
spectral modes. Chemical images are generated by using option 'B' of the function chemical imaging.
Screenshot of the dialog box 'Chemical Movie'
<
<<
•
>
>>
one step (frequency position) backward
starts an animation of the frequency slices (='chemical movie') in backward direction
stop playing the chemical movie
one frequency slice forward
plays the chemical movie in forward direction
increment: the step size between two consecutive video frames in frequency units, i.e. wavenumber, or Raman shift
positions [cm⁻¹]. Values smaller than 1 are permitted. The default value for 'increment' equals 2.
actual position: frequency position of the actual video frame given in frequency units, i.e. wavenumber, or Raman shift
positions [cm⁻¹].
Frequency Imaging
Frequency imaging: This function permits visualization of peak positions, and their variations, within hyperspectral image data.
Band positions - either maxima or minima - are obtained, its values are color-converted and plotted as functions of the spatial coordinates.
Band positions can be obtained from all types of data blocks, including also derivative spectra (for details refer to chapter
Internal Data Organization).
Two different methods of the 'frequency map' routine are available: Peak maxima/minima can be obtained from spectra contained
in one of the four data blocks, or for overlapping bands from second derivative data blocks (this may also be second derivatives from
spectra of the derivative data block!).
The use of second derivatives for peak picking may be useful for the detection of peaks in the presence of strongly overlapping signals,
i.e. when the band is only a small shoulder on a strong signal. To some extent 2nd derivatives compensate also for baseline effects.
Derivatives should be used with care as noise is considerably amplified.
The algorithm of the peak picking routine works as follows:
A. If the option 'obtain peak positions from derivatives' was NOT selected:
1. Spectra of the data block of your choice (see option 'use data block') are interpolated. For interpolation, the
spline method is used. Interpolation is carried out in the frequency range indicated in the edit fields 'select spectral region
for peak search'. Furthermore, a factor of interpolation can be selected. This factor indicates how many times the number of
data points will be increased by interpolation (only in the spectral region selected for peak picking).
2. If the option 'search maxima' was checked, the algorithm is then searching for the x-positions (frequencies) of the
maxima within the spectral region indicated by the user. If minima are chosen, the program is searching for minima. IMPORTANT: If
band positions are obtained from the data block of derivative spectra, maxima appear in second derivative spectra as minima and vice
versa. There is no check for i) the order of the derivative and consequently ii), no compensation for the inversion of maxima and
minima!
3. The frequency values of the maxima/minima are color scaled and plotted as a function of the spatial coordinates. If you
wish to further analyze the band positions by other programs you can access the data matrix of frequency values by using the
Export Maps function.
B. The option 'obtain peak positions from derivatives' was checked:
1. Second derivative spectra from the data block of your choice (see option 'use data block') are calculated
by applying the Savitzky-Golay algorithm with 5 smoothing points (see also chapter Calculation of Derivative Spectra).
2. Derivative spectra are interpolated. For interpolation, the spline method is used. Interpolation is carried out in the
frequency range indicated in the edit fields 'select spectral region for peak search'. Furthermore, a factor of
interpolation can be selected. This factor indicates how many times the number of data points will be increased upon interpolation
in the spectral region selected for peak picking.
3. If the option 'search maxima' was checked, the algorithm is then searching for the x-positions (frequencies)
of the maxima within the spectral region indicated by the user. If minima are chosen, the program is searching for minima.
IMPORTANT: If band positions are obtained from the data block of derivative spectra, maxima appear in second derivative spectra
as minima and vice versa. There is no check for i) the order of the derivative and consequently ii) no compensation for the
inversion of maxima and minima!
4. The frequency values of the maxima/minima are color scaled and plotted as a function of the spatial coordinates.
If you wish to further analyze the band positions by other programs you can access the data matrix of frequency values by using
the Export Maps function.
Screenshot of the dialog box 'Frequency Imaging'
search maxima or minima: indicate whether you want to search for peak maxima or minima.
obtain peak positions from derivatives: if this checkbox is checked, second derivatives of the chosen data block will
be calculated. In this case, maxima or minima will be obtained from second derivative spectra. Otherwise, peak positions
are obtained directly from the selected source data block.
interpolation factor: this factor indicates how many times the number of data points will be increased by interpolation.
Large interpolation factors will result in smoother frequency images.
select spectral region for peak search: the spectral region which will be used for peak picking.
use data block: selected the appropriate source data block from which peak picking will be carried out.
plot: starts the peak picking routine and draws the univariate frequency image
cancel: the 'frequency plot' window will be closed. No frequency image will be drawn.
FWHM imaging: This is a function introduced with CytoSpec version 2.00.05 (Feb 2018). The FWHM imaging function is suitable for
creating pseudo-color images based on the full width at half maximum (FWHM) of vibrational bands. The function requires the presence of
a band in a given spectral interval, ideally in all spectra of the hyperspectral image.
Screenshot of the dialog box 'FWHM Imaging'
interpolation factor: select the source data block and chose then the interpolation factor which indicates how many times
the number of spectral data points will be increased by interpolation. Interpolation is performed in the given spectral region
and is thought to be helpful to improve the numerical accuracy when determining FWHM values of IR, or Raman bands.
select spectral region for peak search: indicate the spectral region in wavenumber / frequency units to be used for
determining FWHM values of a specific band.
use data block: please choose the source data block to be analyzed.
plot: starts the 'FWHM imaging' function. When finished the functions plots a false-color FWHM image into the
respective axis of original, or processed data by color-encoding the FHMW values.