187		13

ImageJ and Fiji

13.1\ The ImageJ Universe – 188

13.2\ Fiji – 188

13.3\ Plugins – 190

13.4\ Where to Learn More – 191

\References – 193

© Springer Science+Business Media LLC 2018

J. Goldstein et al., Scanning Electron Microscopy and X-Ray Microanalysis, https://doi.org/10.1007/978-1-4939-6676-9_13

\188 Chapter 13 · ImageJ and Fiji

Software is an essential tool for the scanning electron microscopist and X-ray microanalyst (SEMXM). In the past, software was an important optional means of augmenting the electron microscope and X-ray spectrometer, permitting powerful additional analysis and enabling new characterization methods that were not possible with bare instrumentation. Today, however, it is simply not possible to function as an SEMXM practitioner without using at least a minimal amount of software. A graphical user interface (GUI) is an integral part of how the operator controls the hardware on most modern microscopes, and in some cases it is the only interface. Even many seemingly analog controls such as focus knobs, magnification knobs, or stigmators are actually digital interfaces mounted on hand-panel controllers that connect to the microscope control computer via a USB interface.

In addition to its role in data acquisition, software is now indispensable in the processing, exploration, and visualization of SEMXM data and analysis results. Fortunately, most manufacturers provide high-quality commercial software packages to support the hardware they sell and to aid the analyst in the most common materials characterization tasks. Usually this software has been carefully engineered, often at great cost, and smart analysts will take advantage of this software whenever it meets their needs. However, closed-source commercial software suffers from several limitations. Because the source code is not available for inspection, the procedures and algorithms used by the software cannot be checked for accuracy or completeness, and must be accepted as a “black box.” Further, it is often very difficult to modify closed source

13 software, either to add missing features needed by the analyst or to customize the workflow to meet specific job requirements. In this regard, open source software is more flexible and more extensible. The cost of commercial software packages can also be a downside, especially in an academic or teaching environment or in any situation where many duplicate copies of the software are required. Clearly a no-cost, open source solution is preferable to a high-cost commercial application if you need to install 50 copies for instructional purposes.

One of the most popular free and open source software packages for SEM image analysis is ImageJ, a Java program that has grown over the decades from a small application started at the National Institutes of Health (NIH) into a large international collaboration with hundreds of contributors and many, many thousands of users (7  http://imagej.net).

13.1\ The ImageJ Universe

ImageJ has grown into a large and multifaceted suite of related tools, and how all these parts fit together (and which are useful for SEM and X-ray microanalysis) may not be immediately obvious. The project began in the late 1970s when Wayne Rasband, working at NIH, authored a simple image processing program in the Pascal programming language that he called Image. This original application ran only on the PDP-11, but in 1987 when the Apple Macintosh II was

becoming popular, Rasband undertook the development of a Mac version of the tool called NIH Image. Largely to enable cross-platform compatibility and to allow non-Macintosh users to run the program, it was again rewritten, this time using the Java programming language. The result was the first version of ImageJ in 1997 (Schneider et al. 2012, 2015).

The availability of ImageJ on the Microsoft PC and Unix platforms as well as Macintosh undoubtedly added to its popularity, but just as important was the decision to create an open software architecture that encouraged contributions from a large community of interested software developers. As a result, ImageJ benefitted from a prodigious number of code submissions in the form of macros and plugins as well as edits to the core application itself. Partly to manage this organic growth of the package, partly to reorganize the code base, and in part to introduce improvements that could not be added incrementally, NIH funded the ImageJ2 project in 2009 to overhaul this widely useful and very popular program, and to create a more robust and more capable foundation for future enhancements (7  http://imagej.net/ImageJ2).

Both ImageJ and ImageJ2 have benefitted from independent software development projects that interoperate with these programs. The Bio-Formats file I/O library as well as other related projects led by the Laboratory for Optical and Computational Instrumentation (LOCI) at the University of Wisconsin (7  https://loci.wisc.edu) are important resources in the ImageJ universe and have added valuable functionality. The Bio-Formats project responded to the community’s need for software that would read and write the large number of vendor-supplied image file formats, mostly for light microscopy (LM). Today the Bio-Formats library goes well beyond LM vendor formats and encompasses 140 different file types, including many useful for SEMXM, such FEI and JEOL images, multi-image TIFFs (useful for EDS multi-­ element maps), movie formats like AVI for SEM time-lapse imaging, etc. A follow-on LOCI project called SCIFIO aims to extend the I/O library’s scope to include N-dimensional files (Hiner et al. 2016). Both projects are closely associated with the Open Microscopy Environment (OME) project and the OME consortium (7  http://www.openmicroscopy.org). Similarly, the ImgLib2 project aims to provide a neutral, Java-­ based computational library for processing N-dimensional scientific datasets of the kind targeted by SCIFIO (Pietzsch et al. 2012).

Given the complexity of this rapidly evolving ecosystem of interrelated and interoperable tools that support ImageJ, it is not surprising that some users find it difficult to understand how all the pieces fit together and how to exploit all the power available in this software suite. Fortunately, there is a simple way to access much of this power: by installing Fiji.

13.2\ Fiji

Fiji, which is a recursive acronym that stands for “Fiji Is Just ImageJ,” is a coherent distribution of ImageJ2 that is easy to install and comes pre-bundled with a large collection of useful­

189

13

13.2 · Fiji

plugins and enhancements to the bare ImageJ2 application (Schindelin et al. 2012). It is often thought of as “ImageJ with Batteries Included.” The Fiji website provides several convenient installation packages for both the 32-bit and 64-bit versions of Fiji for common operating systems such as Microsoft Windows (currently Windows XP, Vista, 7, 8, and 10) and Linux (on amd64 and x86 architectures). Pre-built and tested versions for Mac OS X 10.8 (Mountain Lion) and later are also available. By default, these bundles include a version of the Java Runtime Environment (JRE) configured for Fiji’s use that can coexist with other instances of Java on the host computer, but “bare” distributions of Fiji are available that will attempt to utilize your computer’s existing JRE if that is preferred. Of course, as an Open Source software project, all of the source code for Fiji can be downloaded.

Installation of Fiji is straightforward because it has been configured as a portable application, meaning it is designed to

run from its own directory as a standalone application. Installation is as simple as downloading the distribution and unpacking it; Fiji does not use an installer, does not copy shared libraries into destination directories scattered around the file system, and it does not store configuration information in system databases (e.g., the Windows registry). Because of this design, once installed it can be moved or copied simply by moving the directory tree. This portability also means it runs quite well from a USB flash drive or removable hard drive.

After launching Fiji you will be presented with the Fiji main window (.  Fig. 13.1a), which contains the Menu bar, the Tools bar, and a Status bar for messages and other application feedback to the user. Selecting “Update…” or “Update Fiji” on the Help menu will trigger the updater, one of the most useful features of Fiji. Because Fiji is configured by default to start the updater immediately after program launch, for many new users this is the first piece of Fiji

. Fig. 13.1  a Fiji main window.	a
b Fiji updater

b

190\ Chapter 13 · ImageJ and Fiji

. Fig. 13.2  Fiji’s Manage Update Sites window

13

functionality­ they encounter. Upon activation the updater will scan your local Fiji installation and calculate checksums for everything to see if any components are out-of-date, or if new features have been added since it was last run. It will then confer with the global Fiji code repositories to look for updated Java Archive files (.jar files) and offer to download and install them for the user. .  Figure 13.1b shows an example of this, where the updater has located numerous changes in the ImageJ, Fiji, and Bio-Formats repositories. By selecting the “Apply changes” button the software will fetch the latest code and apply all the patches to the user’s local Fiji installation. .  Figure 13.2 shows a window listing a selection of available Fiji update sites illustrating the rich community resources.

13.3\ Plugins

One of the most powerful features of Fiji is the enormous collection of plugins, macros, and other extensions that have been developed by third-party contributors in the scientific

­community. Fiji comes with some of the most useful plugins pre-installed, and these are accessible from the Plugins menu item. Hundreds of powerful features are accessible this way, exposed to the user in a series of cascading menus and submenus. Such a large set of choices can be overwhelming at first, but many of the plugins are meant for light microscopy, so the SEM analyst may find it simpler to ignore some of them. However, the Non-local means denoising plugin, the Optic flow plugin, and the myriad of morphological operations under the Plugins|Process menu are all useful for SEM microscopists, as are the dozens of features in the Registration, Segmentation, Stacks, Stitching, Transform, and Utilities submenus.

Sometimes the appearance of a plugin as a single entry in the Fiji menu structure belies the full power of that plugin. Indeed, some of the most impressive plugins available for Fiji might be considered entire image processing packages in their own right. An example of this is the Trainable WEKA Classifier plugin that appears as a single entry on the Segmentation submenu of the Plugins menu. WEKA is an acronym that stands for “Waikato Environment for Knowledge Analysis,” a tool