Test Platform

Power Macintosh 7200/90 running Mac OS 8 with 32 Mb RAM
UMAX S-12 Vista flatbed scanner

Overview

NIH Image 1.61 is a free, public-domain Macintosh program for image capturing, processing, and analysis. It is one of the most versatile programs available for image analysis on any platform, allowing biologists to process images in a variety of formats. Some of the powerful features include overlaying epifluorescent and phase microscopy images, doing 3-D reconstruction of confocal microscopy images, conducting densitometric analysis of autoradiographic gel images, counting particles, measuring structures from transmission electron micrographs, making QuickTime movies from a series of images, and making montages from images in a stack format.

Most remarkably, users can customize NIH Image to meet their research needs. This can be accomplished in several ways. The easiest way to extend NIH Image is by using preexisting Adobe Photoshop compatible plug-ins or previously written macros. Computer-savvy users can write their own custom macros or even change the NIH Image source code, which is written in Pascal. Many macros already exist and are distributed with the program. Numerous additional, researcher-contributed macros are available through the NIH Image home page.

In addition to being completely customizable, this program compares well with many expensive, commercially available software packages, feature for feature.

Program Features

New Features

The most significant revisions in version 1.61 include bug fixes, the ability to open 24-bit color TIFF files in planar format, and support for QuickTime movies. Some of the eliminated bugs relate to video look-up table saving, exporting, and updating, as well as those that caused NIH Image to crash upon opening files in PICS format. For a complete listing of new features, see the release notes.

Basic Features

Image capturing can be done from within the program. Alternatively, one can open images in standard graphic formats such as MacPaint, TIFF, PICT, and PICS. NIH Image can analyze only 8-bit images, but it can convert 16- and 24-bit images to 8-bit. NIH Image uses Adobe Photoshop acquisition plug-ins to capture images from scanners or frame grabbers. These plug-ins also allow the program to read unsupported file formats. More information is available on Image-supported frame-grabber cards, the best CCD cameras and light boxes for image capture, and scanners.

Using Macros

Many of NIH Image's functions are performed by loading specialized macros to automate frequent or repetitive tasks such as measuring, stack analysis, or animation. One or more macros may be grouped together and saved as text files, but only one text file can be loaded at a time. Users will find that many of the existing macros are unexpectedly grouped together in an eclectic blend of features. Each macro in a file shows up as a new command in the Special menu. To access

figure 1

figure 2

figure 3

figure 4

figure 5

figure 6

features more quickly, one can assign keys to call particular macros in order. Users may find it useful to combine the macros found in different text files, or to add frequently used features to the program's source code.

Analysis and Presentation Features

Presentation and analysis of raw data for scientific publications is a grueling process requiring different equipment, photography departments, and a well-funded budget. NIH Image, in conjunction with a flatbed scanner, slide scanner, or CCD camera, enables researchers to avoid densitometers and printing from negatives. In addition, NIH Image can replace expensive software packages that specialize in analysis of each image type. To demonstrate this program's versatility for analyzing images generated from a variety of sources and for creating publication-quality figures, the following types of analyses were done: measurement of structures from TEM negatives of bacterial cells, quantitation of a chemiluminescent Western blot image from X-ray film, and assembly of images into a publication-quality composite or montage. These examples are simple, only scratching the surface of the tasks one can perform with NIH Image.

Size Calibration and Measurements

Measurement of structures on images can be accomplished if a scale bar is available for calibration. In the case of a TEM negative, a scale bar is stamped on the negative, which allows easy calibration for structural analysis (figure 1). To begin, the Line tool is used to draw a line against the stamped calibration bar (figure 2). Set Scale is selected from the Analyze menu, allowing entry of the "known distance" of the line (figure 3). If the units are not set prior to entering the known distance, the calibration will not work. Although this is not documented, a similar problem was recently mentioned in the NIH Image newsgroup. Once the image is calibrated, measurements are made by enclosing an area with a square, circle, or free drawing tool and selecting Measure from the Analyze menu (figure 4 and figure 5.) Measurements are recorded in a results file (figure 6) that may be imported into a spreadsheet and graphics program such as Microsoft Excel for further analysis or plotting of data. Unfortunately, the manual does not provide enough detail to determine easily all of the important parameters involved in calibrating the image.

Calibrating images for densitometric analysis also requires a standard density image for quantitative measurements. The most important aspect of densitometry is calibrating the image. If qualitative comparisons are desired, background subtraction can be performed and values from individual images can be compared to each other (figure 7, figure 8, and figure 9). For further reading, Using Image for Densitometric Analysis of 1-D Gels is a well-written tutorial maintained by the National Institutes of Health's Division of Computer Research and Technology.

Assembly of Composite Images and QuickTime Movies

The most exciting presentation features of NIH Image include the automatic assembly of composite images from individual images, animation of images to make short movie sequences, and the ability to overlay two microscope fields by accurately using a registration mark. For example, this last feature makes it easy to line up images captured using phase-contrast and epifluorescence microscopy. To assemble a montage of individual images, all images must be of the same size. The easiest way to accomplish this is to use Adobe Photoshop. In Photoshop, rectangular area selections may be dragged from window to window, and selected areas can be duplicated and saved. This feature is lacking in NIH Image. A command in one of the stacks macros allows all images in a stack to be cropped to the same size. However, there is no control over where the cropping occurs in each image of the stack. Once all images of interest have been cropped to the same size, they must be assembled into a stack using the Windows to Stack command from the Stacks menu (figure 11). After the individual (figures 10a, 10b, 10c, and 10d) have been placed into a stack, select Make Montage from the Stacks menu. One may number the images automatically when the montage is assembled, or manually place letters or numbers on the image later with the Text tool (figure 12). For an excellent example of NIH Image's capabilities for handling stacked images, see the Multiple Images (Stacks) page of the program's tutorial.

Documentation and Online Resources

The most complete description of the program's features is available as an online manual. NIH Image's original features are better documented than recently added ones. Often, the manual only mentions a new feature, not effectively explaining its proper use. A downloadable version is available as a Word 5.1 document archived in HQX format. A frequently asked questions document (FAQ) provides many useful tips. A brief tutorial acquaints new users with the program's basic features.

For further assistance, users may subscribe to the NIH Image mailing list by sending email containing the line "subscribe nih-image [your real name]" in the message body to listproc@soils.umn.edu. Archives of past list traffic can be found at a gopher server where one archive allows one to search all past messages by keywords, and another to retrieve entire messages by the month in which they were sent. A Web archive also stores messages by months.

Limitations and Comments from Users

A list of known limitations in NIH Image can be found in the online manual. However, it addresses only image sizes, file sizes, and restrictions on clipboard exportation. The program itself has very few limitations, and modifications can be made to the source code to correct any problems ncountered. The most significant limitation is that of computer memory requirements, which increase dramatically when working with multiple images simultaneously or with very large images. A few users have reported difficulties in exporting measurements from NIH Image into Microsoft Excel 5, and have had problems with reliable particle analysis using TIFF images under Mac OS 8.

Future Prospects

NIH Image is continuously improved and expanded, in part by users who are encouraged to share new macros or source code. Version 1.6 was released in December 1996, and a beta version, 1.6.2b20, in December 1997, fixing a few program bugs and increasing image handling capacity. NIH Image runs under DOS, Linux, or NextStep by using the Executor Mac emulator from ARDI; however, other versions of NIH Image are currently available for other platforms. A port of NIH Image called Scion Image for Windows is available from the Scion Corporation Web site or from the NIH Image Spin-offs directory. It is a 32-bit application that requires either Windows 95 or Windows NT 4.0. It is worth noting that Scion also distributes Scion Image for Mac OS, an extended version of NIH Image. Modifications include more advanced image capturing capabilities, video export features, and 24-bit color image display. A platform-independent version of NIH Image called Image/J, written in the Java programming language, is currently under development. A beta version of Image/J is available.

NIH Image is an extremely versatile image analysis program with nearly limitless potential for expansion. It is perfect for the research lab that cannot afford comparable commercial programs, or for those that may customize a program for specialized techniques. NIH Image lacks some amenities offered by commercially available image analysis packages. However, when used in combination with Adobe Photoshop, many of these features can be duplicated or streamlined to produce similar results.

NIH image requires a color-capable Macintosh with at least 2 Mb of free RAM running System 7.0 or later. If working with 3-D images, 24-bit color, or animation sequences, a Macintosh with 16 Mb or more RAM is recommended. NIH Image directly supports or is compatible with large monitors, flatbed scanners, film recorders, graphic tablets, PostScript laser printers, phototypesetters, and color printers.

The NIH Image binaries, available macros, source code, and documentation may be downloaded for free from the National Institutes of Health's Image Web site, or by anonymous FTP from zippy.nimh.nih.gov in the /pub/nih-image directory (username "anonymous" and the password one's email address). Users can also choose to download a variety of user-contributed macros, sample images, NIH Image stacks, spin-off programs, and Photoshop-compatible plug-ins that work with NIH Image.