nsherry4 / peakaboo

XRF Analysis Software

Peakaboo

XRF Visualization Software

Peakaboo allows users to identify the spectral origins of the XRF spectrum using a technique that fits all components of the K, L, or M spectrum including escape peaks and pileup peaks, and then plots their spatial intensity distributions as maps.

Downloads for Windows, Mac, Linux

Learn how to use Peakaboo on our Wiki

Ask a question in our Discussions

Peak Fitting

The spectra produced are fitted with several K, L, or M lines for each element. The line positions and relative intensities for each line series were taken from several tabulated sources, including Xraylib. For fitting of the spectral peaks, a Pseudo-Voigt function is used by default. Thus, the identification of a particular element involves a close fit of multiple lines in the spectrum, each with its own shape.

Noise Reduction

Noise reduction is essential for spectra taken at very brief intervals while the sample is scanned in the X-Ray beam. The software provides a number of mathematical filters that are used in noise reduction or in background attenuation or removal. Noise filters include moving average, fast Fourier transform (FFT) low pass, Savitsky-Golay, and others.

Where applicable, noise removal filter parameters can be adjusted to suit the data in question. Best noise reduction with minimum change in peak shape is often achieved using a Savitsky-Golay filter.

Background Removal

Background removal or reduction is particularly important for spectra acquired using white radiation. Peakaboo has a number of background removal filters of varying sophistication and performance. Varying levels of background can be removed.

Results Mapping

Once peak fitting is complete, one or two dimentional data sets can be mapped to show the distribution of the selected elements. Peakaboo can show individual elements, composites of several elements, ratios of sets of elements, overlays of up to four sets of elements, as well as two dimensional histograms to compare element concentrations.

User Extendable

Peakaboo allows users who are comfortable with Java programming to extend Peakaboo in three ways:

• Filters
• Data Sources
• Data Sinks

To get started creating your own filters or file format support, all you need is the Peakaboo JAR to build against. For more information on creating plugins, see the user manual.

Library and Plugins

Peakaboo can also be used as a library. This is also the easiest target to build custom plugins against. Read more about Peakaboo as a library without the GUI.

Building

Peakaboo builds and manages its dependencies with maven. It also has a dependency on xraylib. At the time of writing, xraylib was not available in maven, so a prebuilt version is included in a local maven repository. To build Peakaboo, run mvn package and it will produce a runnable jar file you can invoke with the java -jar <JARNAME> command.

peakaboo's Issues

masked (marked bad) scan information does not properly carry through to subplots

Because the dimensions of the data change, the index values used to store the bad scans are no longer accurate, and now mark the wrong scan as bad, or cause index out of bounds exceptions

for that reason, 'bad scan' data is intentionally discarded in Peakaboo 3 when creating a subplot view

This could be addressed by recalculating the index values, or by storing the 'bad scan' data as a boolean list in the DataSource object so that CopiedDataSource could do the index adjustments for it, as well

Load Session dialog shows all extensions

Loading a session should only show .peakaboo file extensions, but it shows everything

Plotting a subset should unminimize plot window

Composite and Overlay map log view UI not linked

Composite and Overlay map log view check boxes both control the same log view parameter in the model, but they don't update their value when the other alters the model.

Fix or Remove LeastSquaresFittingSolver

Interpolation cap

Currently, the maximum number of interpolation iterations is limited by total map size as a matter of performance. This can lead to very small maps having very high levels of interpolation allowed, which doesn't really make sense. Interpolation should be limited to 2 or 3 levels regardless of map size.

Peakaboo manual missing citation

In the section "Details of Peak Fitting", there is a citation which does not have a match in the bibliography.

Filters should be queryable to see if they'll work on a given spectrum

Some filters, like the FFT low-pass filter, need to operate on spectra with a length which is a power of 2. Having a check for each filter if it can handle non-standard lengths will prevent this sort of problem.

Determine peak height reporting units

Should peak height be reported in units of counts, or be scaled to the relative intensity of the strongest line in the TransitionSeries?

Guided fitting doesn't work above ~20keV?

Observed in dataset w/ ~160keV calibration

Fitting list controls are incorrect when opening a subplot

The remove/clear icons should be enabled, but they are not.

Persistent Settings

Should Peakaboo remember preferences like log/linear, axes, etc between sessions?

Get Intensities does not work with Points Selection

Remove FileFormat's compatibility(Path)

Save Images with Explicit Sizing

Currently, plot/map images are saved at whatever size the window is at. This can make it hard to produce maps the same size on two different computers/monitors. When exporting images, the size should be confirmed with the user.

Maybe the default can be the current size, but we should probably also provide some other sensible defaults to choose from, along with manual entry.

PeakTable relies on call to Xraylib RadRate to reject some lines

This is poor form, and we should find a better way to do this.

Shrinking the map window doesn't play nice with zoom feature

Shrinking the map window should shrink the map when the zoom is set to 1, but it just adds scroll-bars

Revisit Peak Models

Investigate use of xraylib

https://github.com/tschoonj/xraylib

Save/load new peak model settings as part of session

Drag and Drop to reorder fittings

Energy auto-calibration should be more thorough

The energy calibration currently looks for the best "rough" fit, then uses the slower "quality" algorithm to refine the result. Instead, the rough algorithm should be used to rule out, say, 99% of possibilities, leaving the rest for the quality algorithm to verify and then refine.

Savitsky-Golay always only filters weak signal, doesn't honour parameter

SettingsModel and Filter/Fitting Model values kept in sync at startup manually

Settings like escape peak type, fwhmBase, etc are being kept in sync manually by defining the presets twice.

Remove use of SwidgetIO for SimpleIODialogs

Using output/input streams is a holdover from trying to support applets.

Saved session doesn't include visibility of fittings

The session files saved to disk (and presumably passed when opening a subplot) don't include the visibility of the fittings.

FFT filter doesn't update end rolloff spinner when changed to cutoff mode

Selected fittings should be highlighted in the plot

When selecting a fitting in the plot window, it's portion of the fit should be highlighted so the user can see where exactly it is without having to toggle visibility on/off

Maps with interpolation and selections don't make sense.

Trying to select an interpolated region to re-plot it doesn't make any sense, because the actual scan data isn't that granular.

Interpolation and selections should be mutually exclusive.

Fitting functions should work in units of energy, not channel

Fitting functions should accept energy values rather than channel values in order to make them more reliable.

Investigate Apache Commons FastMath

Re-plotting element list linked to original

Re-plotting sub region shows element list which is linked to original plot's element list. If you toggle the visibility of an element fit in one, it impacts the other.

Estimating energy calibration takes too long

Using peak locations rather than full fitting may be able to speed this process up. Maybe even enough for it to still be useful after #5

Undo stack should store state as Strings

Since SavedSettings has been reworked, the UndoController can be simplified to store saved settings as a string instead of a byte array.

Modal Panel not redrawing when window minimized

Taking a subset of a large data set freezes the UI for a time

Because any new SubsetDataSource needs to recalculate the mean/max scans and other values, it can take a long time to load a large subset of a large data set.

The UI should show the "Calculating Values" progress meter that it shows as part of opening data for the first time.

Documentation Update

Documentation needs to be updated with new screenshots, and with information on new features.

FFT Filter works in units of keV, when it doesn't have access to that info

Ctrl 3-click on map is broken

Move escape peak selection to peak model menu?

DataSource/Sink plugin interfaces should use Path instead of File

DataSource/Sink plugins should use Path instead of File for read(). This will make it easier to point a DataSource/Sink plugin at files inside a Zip file, for example.

Map of single element title is Map of Map of

Application title not set

It looks like with Java 9 (and at least my impl of Java 8), setting the application title via the Toolkit's awtAppClassName field is no longer allowed, and generates deprecation warnings. This means that (on linux, at least) the application name is not set, and appears blank.

Map Selections for Ratio

Making map selections for Ratio maps should select a range of ratios.

Allow customization of Curve Fitting

Different XRF data from different equipment is going to have differently properties, including different peak shapes, widths, etc. The user should be able to calibrate peak width and fitting function.

Allow minimum energy to be non-zero

Currently, users can set a max energy to scale their data properly, but if the scan does not start at 0 energy, it can skew the results, especially when points of interest are in different parts of a spectrum

Direct API use to process a map yields slightly different results than through the GUI

Ratio view occationally shows white for one side of the ratio

After some complex series of steps which I have not been able to identify, the ratio view in the mapping window will fail to draw (or draw in white) one side of the ratio data.

Workaround: Close the map window and remap. Failing that, save your session, close Peakaboo, reopen it and reload your data and your session.

Mapping with 0 fittings selected is allowed, throws exception

Mapping with 0 fittings selected doesn't make sense, but it's allowed, and throws an exception when you do it.

Dimensions do not carry through to sub-plots

A subplot generated using a bounding box ought to be able to provide data dimensions, but when the subplot is remapped, the map doesn't show those dimensions.