Use of NIH-Image 1.59 for Post-acquisition Processing of Confocal Images.

Prepared by Harvey J. Karten


Dept. of Neurosciences
University of California at San Diego
La Jolla, CA 92093-0608
Email: hjkarten@ucsd.edu


With many thanks to Wayne Rasband for his invaluable contributions to scientists of all disciplines, and his advancement of image processing.

Addenda to Manual March 3, 1995:
1) New Macros to Merge Sides of Split Screen BioRAD image.
2) New Macro to Color Merge Z-Series of Split Screen BioRAD images.
3) Notes on Exporting/Importing PhotoShop format.
4) Comments on Scaling using Bilinear rather than Nearest Neighbor.

Addenda June, 1995
1) Transferring NIH Image stacks to Adobe Photoshop 3.0
2) New macro to automate Z-Projection - June 13, 1995

The work that led to this manual was supported by Grants to H.Karten from NIMH, NINDS and NEI. In our frenzied attempts to obtain Grant support, we often neglect to sufficiently thank the many dedicated scientists who work so hard to establish those programs that provide the necessary funds. This manual is dedicated to those many loyal supporters of all of us at NIH, NSF, DOD, NASA, and elsewhere.

I have adopted and adapted a number of Macros contributed by various NIH Image users. Their authorship has been obscured by the winds of time. I am grateful to them, and hope that they are not offended if not specifically credited. My thanks to Rusty Gage for providing me with unlimited access to his confocal microscope facility. Without his generosity, this small manual would not have happened.

Please let me know if this manual is helpful. Please inform me of any of errors in this manual. If you have any suggestions to improve this manual, additional macros that are helpful for processing confocal images, or other strategies for processing CLSM images, please let me know, and I will try to incorporate these changes.

Happy confocaling,
Harvey J. Karten, M.D.
HJKarten@UCSD.EDU

I. Introductory Comments:

NIH-Image is a useful tool for post-processing of Confocal Laser Scanning Microscope (CLSM) images. The experienced user of NIH Image may find many of these operations obvious. However, many users of CLSMs, previously unfamiliar with NIH Image will find it a useful tool for post-acquisition processing of images. This manual is intended to provide a relatively easy introduction to the use of NIH Image for this latter group of CLSM users.

Performing post-acquisition processing of images on the Mac will free the confocal instrument for image collection. Though not as extensive as VoxelView, Analyze, VolVis, SYNU or VoxBlast, particularly in applications requiring volume rendering and Voxel based calculations, the use of NIH Image and Photoshop 3.0 on the PowerPC will provide most of the functions needed for manipulating confocal images, and at much lower price.

This manual is intended for use with NIH Image, Version 1.59 or later. Several of the functions decribed in the manual will not operate correctly on earlier versions. Please obtain a recent copy of NIH Image from the ftp site: zippy.nimh.nih.gov or the associated Web Site.

Contents of this manual:


1. How to transfer files from the BioRAD, Leica and Zeiss CLSM to NIH-Image. It describes how to load both individual images and Z-series.

2. Image Processing and Image Analysis: NIH Image provides a wide range of functions for image analysis and processing, including changing contrast and brightness values, pseudocoloring images, rotating, cropping, scaling, various filtering operations, measurement of density, density slicing, measuring length and area, profile of a line, counting particles, etc. These are useful tools for modifying and analyzing confocal images. These functions fall into the more general category of image processing, and are discussed at greater length in the NIH-Image manual, About NIH Image. .

3. How to merge double and triple labeled sections, producing a three color images (RGB) with NIH-Image.

4. Using multiple windows, Z-series and stacks : The Mac allows you to open many windows, each containing a different confocal image. This facilitates comparison of different images. The various Stacks functions are amongst the most powerful of features of NIH Image. NIH-Image Stacks function allows you to rapidly step back-and-forth through a Z-series of sections. You can crop this stack to select specific features of interest. You can also use the Stacks function to rapidly and alternately compare different images.

5. How to "Project" all the plates (or only a selected number of plates) of a Z-series onto a single plane.

6. How to generate 3D images and 3D rotations : NIH-Image allows you to generate a 3-D series of images from the Z-series. You can display them in a montage window, fabricate stereo pairs or animate them to give the appearance of rotating a 3D object in space.

7. NIH Image can generate a double labeled 3D series for stereo and rotation.

  1. A.Some Hardware Considerations:

    Mac computers have proven particularly suitable for graphics applications. Their 32 bit memory model permits use of large quantities of RAM for rapid processing of image files. The graphics hardware and operating system have proven very suitable for image processing. All the procedures described in this manual can be achieved on a "low end" Mac, such as the Mac II ci, IIfx, to midlevel machines such as the 68040 Quadra series and, of course, will perform extremely well on the PowerPC series of machines. Separate versions of NIH Image, optimized for the 680X0 series as well as a single "fat" binary version that is optimized for both a 680X0 and a PowerPC, are available. Both versions as are now available from the FTP site. Much less expensive than many dedicated graphics workstations, such as a Silicon Graphics Indy^2, the cost of a fully equipped Mac will vary in price based on some of the variables listed below. NIH Image will perform adequately on any current Mac. The major differences in performance will present when generating a rotation series, resectioning a Z-series, or applying various filters. It is not necessary to have a 68040 with a floating point processor (FPU). However, for optimal performance, a PowerPC 7100, 8500, or the 9500, will prove most attractive. The 6100 does not have room for a frame grabber or a Video output card to video printers, VHS tape, etc. If you have the funds, the 9500/132Mhz unit is obviously preferable. The 8500 and 9500 have fast SCSI-2 ports for rapid data file transfers between external devices, built-in Ethernet for communication, and can be equipped with a high speed graphics display card.

    Operating System: Use System 7.1 or higher. System 7.5 is quite stable, and we now use it routinely.

    Graphic RAM and Monitors: NIH Image is an 8 bit program. However, beginning with version 1.56 NIH Image will run even if the monitor is in 24 bit mode. This permits rapid switching between PhotoShop 3.0 (24 bit program) and NIH Image. In order to obtain a full 24 bit image on a 20 inch monitor (at 1024x768 or higher) the graphics display card on an 8100 (or higher) PowerPC should be packed with 4 MB of VRAM to permit 24 bit color on a 20 inch monitor. The Apple Trinitron 20 inch multisynch monitor provides the best image, and also is the lowest priced 20 inch monitor for the size and quality of the image. However, all the operations outlined in this manual will work on a 14 inch monitor with 8 bits. The 7100 and 6100 only permit 2 MB of VRAM, thus you will be limited to using a 17 inch monitor with 24 bit color.

    RAM vs. VIRTUAL MEMORY: You will need large quantities of RAM for optimal performance. I suggest a minimum of 32 MB of RAM, and more if you can afford it. Avoid Virtual Memory if at all possible. (However, Apple does recommend that you assign 1 MB of Virtual RAM when using System 7.5+ on a PowerPC.) Graphics files are large, and confocal Z-Series are often huge. If you also have a motorized stage and make extended xy planes of Z-series, your files are likely to be 30-50 MB or larger and you will have to reconcile yourself to buying a minimum of 80-256 MB of RAM if you want to move faster than a glacier. A minimal rule of thumb is that you should have 2.5-3 times more RAM than the size of your largest files. You should also learn how to allocate this memory to NIH Image using the "Get Info" window. Otherwise the additional memory will be of no benefit to you. I suggest that you allocate a minimum of 24 MB of RAM to NIH Image, and another 24 MB to PhotoShop. Play around with the program. Many people are using RAM Doubler to compensate for limited amoounts of RAM. I have no experience with this program, and have heard mixed reports as to it value in working with large graphics files.

    DISKS and STORAGE MEDIA: You will need large storage media. But I assume that if you are working with confocal images, you have already had to deal with this problem. A hard disk of at least 500MB is required. For greater flexibility, a system disk of 1 GB and a magneto optical disk of 500 MB or larger for your data files is suggested. You may find it helpful to transfer your larger files from the slower mag-optical to your system hard disk when working on them. Then move them back to the magneto-optical for long term storage. Confocal Z-Series may readily exceed 25 MB. You should have enough storage space on your hard disk to allow you to save both the original file and any modifications you may make during a work session. We often find that we use more than 75 MB/Z-file in a single session.

    EXTENSIONS: A few words about clever System Extensions - all those great clever Extensions that make tea while you compute, or respond to voice commands (such as Apple's PlainTalk) chew up space, but worse, convert your PowerPC to molasses while continually polling for voice inputs, losing CPU cycles, etc. - Turn them all off. Stay only with essential basics. When you are doing image processing, turn off Sharing. If someone else in the lab decides to just glance at your directory across the Network, your machine will be distracted and slow things down. You will still be able to get back onto the Network with the click of key, but not be impeded by curious onlookers. Get used to dedicated computing, as in the days of yore, if you want to obtain your results as speedily as possible.

2. Loading NIH Image and the Confocal Macros:

This manual is written with the assumption that the reader is familiar with the Mac Operating System 7.1 or higher.

Make sure that you have allocated sufficient memory to NIH Image to perform many of the operations outlined. In addition to allocating memory to the Program, using the File Menu item, Get Info , you must also allocate sufficient memory to the NIH Image Undo & Clipboard Buffer Size, shown under the Options + Preferences Menu item . I suggest a minimum of 1400 KB. You must then Record Preferences under the File Menu. Close the Program and restart the Program.

Macros: The accompanying file, "Confocal Macros" should be copied into the Folder (Directory) containing the other NIH Image Macros. Select the Specials Menu, and the Load Macros item. Select the Confocal Macros from the resulting dialog box.

The Confocal Macros file contains a series of Macros that are particularly useful for processing double labeled pairs of sections, Z-series, generating stereo pairs, etc. The reader should also familiarize themselves with the use of Stacks in NIH Image, as described in About NIH Image.

A copy of the Macros file is included in the binhexed file "/pub/image/documents/confocals.hqx" provided via ftp. However, for those users who have obtained this file directly from the WebServer site, the confocal macros file is not readily available. In order to provide access to this file, "Confocal Macros " has been appended to the end of this manual. It should be copied to a "Simple Text" file, and saved as "Confocal Macros."

II. OPENING CONFOCAL IMAGES INTO NIH-IMAGE

The most commonly used CLSMs (Confocal Laser Scanning Microscopes) are the BioRAD, Zeiss, and Leica instruments. The software/hardware used to generate images in these instruments stores the files in a DOS based file format. Thus the user is confronted with two initial tasks: 1) Transferring the file from a DOS based computer to a Macintosh; 2) Converting the file from its original file structure to an NIH-Image file structure.

1. Transferring the file from a DOS, DOS/Windows or OS/2 based computer to a Macintosh:

The simplest means of doing this is to store the file on a removable disk medium of large capacity that be read by DOS, Mac and OS/2 systems. Since Confocal images generate large data files, most systems have magneto-optical disks of 500 MB to 1.3 GB. A popular medium for this is the 1.2 GB MO disk, sold by SONY, Verbatim and others. The most common drives are the Tahiti Max Optix 3, Pinnacle Sierra, various SONY, Ricoh, HP and NEC units. The cartridges are frequently (though not always) interchangeable. Many confocal scopes were supplied with Panasonic MO drives. These use a proprietary cartridge that cannot be read by the drives of the previously mentioned manufacturers. When confronted with a Panasonic drive, pursue one of the methods listed below.

I have encountered occasional problems when attempting to read standard 1.2 GB disks that were originally formatted on a DOS based Pinnacle or Tahiti drive. I suggest that you use "Multi-Driver" in your Mac to facilitate reading DOS-formatted optical disks. "Multi-Driver" is a Control Panel included in a software package sold by PC Access. Once you install this into your Control Panels folder you must "Restart " your computer ofr Multi-Driver to become effective.

Sneaker Net: Transfer the files from the storage location on the original disk to a floppy. The floppy disk can be formatted as either a DOS or Mac disk. The Mac PC-Exchange provided by Macintosh in System 7.1.2 and higher allows you to directly read PC formatted floppies. This is commonly referred to as "SneakerNet". The major limitation of this method is that it is slow and cumbersome if you have many files, and requires that you have many pre-formatted disks. If you have an extended Z-series on a BioRAD, the file may be 7-25 MB, and is not easily transported via floppy disks.

A much more efficient method is to have both the DOS and Mac based machines on a common Ethernet network, with a common local Server, or purchase a copy of MacLAN 5.0 for your PC. MacLAN operates within Windows 3.1. The current version of MacLAN does not work with Windows 95. MacLAN is distributed by Miramar Software and allows allows your Mac "Chooser" to see the PC as another client on an AppleTalk zone, and allows your PC to act as if it were another Mac on your AppleTalk Network. Assuming that you are familiar with using a Mac on a EtherNet network, this allows simple file transfers at high speed.

Transfer between computers that do not share a common Server or Network can be accomplished using Internet protocols such as FTP (File Transfer Protocol). In our experience, the most efficient program for doing this on the Mac side of things is Fetch 2.1.2 (or higher), a Public Domain Freeware Program available on the Zippy.nimh.nih.gov Server. Fetch 2.1.2 will facilitate all aspects of the transfer from a PC/DOS machine. If you do not know how to set up your PC as an FTP site, contact your local computer center for assistance. The latest version of Fetch 3.0b is still listed as a beta (test) version, but appears to operate without major problems.

when performing an FTP, make sure that you perform the transfer in Binary format, or the files will be unusable.

DO NOT ERASE THE ORIGINAL FILES UNTIL YOU ARE CERTAIN THAT YOU HAVE OBTAINED A SUCCESSFUL TRANSFER AND CONVERSION. We recommend that you always save copies of the original files in their original native (DOS) format.

In order to maintain a record of the content of each image or series of images as you collect them on the confocal scope, I have prepared a FileMaker Pro Template. The Template is also posted on the FTP Server for NIH Image within "confocals.hqx". When you download and decompress "confocals.hqx" the Template will be decompressed as "Confocal_FileMaker_Template". FileMaker Pro is a simple and inexpensive database program. It is available for both Mac and DOS/Windows, and files are interchangeable between the two versions.

2. Opening BioRAD Files into NIH Image.

a. BioRAD files consist of a 76 byte header that defines the size of the image in width and height, states if it is a Z-series, and, following the image data, provides information on parameters during collection of data, magnification scale, and Notes. Individual files can also be opened using the Import function of Image. However, this requires that the user know the width and height of the individual file (Usually 768x512, but this can vary with the option chosen in BioRAD). Unfortunately, I cannot figure out how to transfer a merged 8 bit BioRAD image to a Mac. The associated LUT is apparently stored in a manner or location that eludes me. If someone has solved this problem, please let me know. The Import function does not permit importing a Z-series. The BioRAD macro will correctly import a Z-series.

b. The macro "Import Biorad MRC 600 Z Series ", contained in the associated file "Confocal Macros", enables opening of either single sections or a Z-series. This works equally well on files generated with Comos and those generated with the recently released software, ThroughView 1024. ThroughView 1024 operates under OS/2. These files can be opened using the same Macro employed to open the DOS formatted files.

c. After loading the BioRAD Macro, you can "run" it by selecting it from the Special Menu. The dialog will ask you for the spacing of the Z-Series in Pixels. The default is 1.00. Accept that value for the moment. A BioRAD file containing only a single image will appear on the screen with the title of the original file (e.g. Axons03.PIC). We suggest that you immediately save the file, prior to making any modifications, as e.g., "Axons03.PIC.Img".

d. The associated calibration and Notes of a single image file will be placed in a separate NIH Image text window. If you wish to include this information on the image that you save as an NIH-Image file, you must paste it from the text window to the image window. If your notes contain information about Pixel size and dimension of the image, use that to calibrate the scale of the image. (See Set Scale in NIH Image Manual). Once you have calibrated the image, save the file once again. The current version of the BioRAD macro does not properly read the calibration or Notes file attached to a Z-series file.

e. BioRAD Z-Series can be stored as a single large file, or as a series of individual sections. The former configuration has the advantage that all related files are stored in a single locus. However, it also means that such files may be huge. Using the above Macro, a BioRAD Z-Series will be opened into a new Stack . Save this in a similar manner suggested above for a single image file.

f. If this is a Z-Series use Menu Stacks , and Select Options to enter the step size of the motorized focus used when originally recording the Z-Series. The magnification and slice spacing are stored with the original BioRAD file.

g. You may find it useful to "Add" a "Slice" to the beginning of a Stack to record specific comments about the series, what the data represents, etc.

h. Save the original BioRAD *.PIC files, -just in case. You can move all the original files into a separate subfolder to reduce clutter.

BioRAD Split Screen Images:
Simultaneous collection of double labeled sections on the BioRAD can be displayed and saved on a "split screen" image, with the two images from PMT 1 and 2 displayed "side by side." This has two advantages: 1) You can directly compare the location of different antigens simultaneously; 2) When collecting a Z-series with simultaneous split screen, you can be assured that the two images will be in the same focal plane. The Z-motor drive on the BioRAD has not proven as reliable as it should be, and is prone to slippage due to poor mechanical linkage. If the Z-series for two or more antigens is collected sequentially, you may not be able to rely on the accuracy of Z-position to return the section to the same vertical location in subsequent scans. The inaccuracy may also be due to the current use of sloppy focus mechanisms on most modern microscopes. The modern co-axial planetary gears do not reliably return the stage to the starting position when restored to the same posiiton on the fine focus knob.

The macro described in the previous paragraph will open and display the split screen image. If you want to separate the two halves of the screen into separate images, use the macro "Merge BioRAD Split". This will convert the original split image into a three slice stack (left image, right image and one blank black slice), then produce a merged color image. If you want to interchange the red and green planes, Select the window containing the Stack of three images, and run the Macro "Swap Red_Green".
The stack of three slices (left, right and blank), can then be saved as a 24 bit image in PhotoShop 3.0 format, in the following manner:
1) If you have run the preceding macro, and have generated an RGB image in NIH Image, NIH-Image will save the file as an RGB-TIFF file. This format is accepted as a PhotoShop 3.0 RGB image.
2) Save this file with a modified title in order to preserve your original file.
3) Open the file using Photoshop 3.0.

The quality of the RGB image produced within NIH Image is an 8-bit indexed image, is of marginal quality, and should be considered to be a crude "proof" image. You may find that the image can be greatly improved by changing contrast or brightness of the individual slices. Before you do this, make sure that you have saved a copy of the unmodified Stack. In general, you will find that such modifications are best accomplished in PhotoShop 3.0.

Merging Split Screen Z-Series:
The macro that operates on a Z-Series of a Split screen multi-labeled section will:
1) Place the Left image in one stack, and the right hand in a second stack.
2) Color Merge the two stacks, and allow you to animate the stacks. This is very helpful when you want to track individual processes in a complex field through a lengthy Z-series.


3. Leica Files

Leica file format is a TIFF format. The Leica operating system is a peculiar hybrid using a VME bus with a 68040 CPU operating under OS/9, but uses a Windows front-end that generates DOS types of files. These can be directly read by the Mac. They will show up on your Mac Desktop as PC files, with various optional icons, depending on how you set your system parameters in the Control Panel, PC Exchange . Each separate image file is accompanied by an "info.dat" file. Z-Series are stored as single files with sequential numbers, but with only a single info.dat file for the whole series. In order to reduce confusion in handling these Z-series with large numbers of files, move each Z-series set into a separate Folder on your Mac.

If you are running Image, you can "Open" these files directly, without having to Import them. However, you will not be able to double-click on the files to open them in Image . In order to do that, pursue the following procedure:

Obtain a copy of "CTC 1.4" (or later) to change file type and creator. CTC 1.4 is available from the NIH Image Server at zippy.nimh.nih.gov. Select all the Leica graphics files in each folder (not the information.dat files) and drag on top of the icon for CTC. Using the resulting dialog box, change the Creator to Imag , and the Type to TIFF . (Creator and Type are case sensitive, so enter exactly as spelled ) All the files will now have an NIH Image Icon, and will be treated as Image files by the program.

The info.dat files should be changed to Creator Imag and Type to TEXT . This file contains important information about scaling factors, size of image, size of pixel, spacing in a Z series, etc.

Opening a Leica Z-Series: Close all other NIH Image windows at this time. Using the File Menu, select Open . In the resulting dialog box, go to the desired folder and click "Open All", then "Open". This will open all the files in rapid sequence. Under Stacks , select Windows to Stack . Make sure that you set the magnification/calibration scales and slice spacing. Save the resulting stack with a new name.

4. Zeiss Files.

Zeiss files saved using the newer version of software with the LSM 310 and 410 are "standard" *.TIF files. Use the same procedure outlined above for the Leica files - i.e. use "CTC" to convert the File Creator and File Type to Imag and TIFF , respectively. The Files can then be directly opened by NIH Image. Make sure the LUT is correctly set. I have often found the Zeiss LUT map to be inverted, with a negative slope, and an inverted image. Click on the Icon in the lower left corner of the Mapping window to obtain the correct black/white relationship.

5. Molecular Dynamics/Sarastro

I have no information about the file format used in this software. If a user has information about this, please contact me via Email.

6. Noran Confocal Files -

There is a version of NIH Image that is used for direct data collection on the Noran. The resultant files are obviously compatible with NIH Image. The Noran is also sold with a version of Image-1, a DOS based program. A version of this program is now also available for DOS/Windows. I have no information about the file format used by Image-1.

III. BASIC IMAGE MANIPULATIONS

NIH Image is an highly effective teaching tool for students interested in Image Processing. NIH Image provides the user with an extensive range of image processing tools. You should become fully conversant with all these tools, including how to obtain a histogram, how to interpret the histogram, manipulating the Look-Up Table (LUT), substituting a pseudocolor LUT in place of the standard gray scale LUT, inverting the LUT, etc.

Keep in mind that most changes will only be made to the display buffer not to the file buffer. Thus, if you examine a histogram, then modify the brightness and/or contrast, the histogram will not be changed. If you now "Apply LUT" to the file buffer, the new histogram will reflect those changes. The original file, stored on your disk, will not be altered by this operation unless you now Save this modified file. The reader should use one of their own sample CLSM files to familiarize themselves with operations of NIH Image. All the operations in this section are fully dealt with in the NIH Image Manual, About NIH Image .

NIH Image has an extremely powerful Pascal-like Macro language. Sample macros are provided with the program from the FTP site. You can combine macros from different Macro files to generate a set of Macros suitable to your needs. I have collected a series of Macros that I have found useful for processing Confocal images. This is available from the FTP server and is contained in the file Confocal Macros contained in the file "confocals.hqx". I suggest that you routinely load the "Confocal Macros" file when running NIH Image. See the manual About NIH-Image for further information about Macros.

You now can open one of the files that you imported to the Mac. Until you are experienced with the program, make a backup copy of the file, and only work on the copy, not the original file. Using the "Save As..." function of NIH Image, change the name of the file so that you don't mistakenly modify your original data file.

1. Evaluating the quality of your original CLSM image:

The quality of the image that you obtain from the following manipulations will be directly dependent upon the quality of the image that you start with. Thus, if you start with a lousy image, you may be able to make it look presentable, but it is still going to be a lousy image.

Make sure that your original CLSM image uses the full range of 8 bit values (0-255). The most common flaw with confocal images, as with video images in general, is that the user has not utilized the full 8 bit range of gray values. Many of the manipulations that you do on the original CLSM may make your image appear to be satisfactory. However, many data files do not contain a full dynamic range of values (0-255), and the image was massaged by artificially spreading a narrow range of values (e.g. 50-120) by modifying the Look-Up Table (LUT).

In order to develop an appreciation for the gray scale content of your images in NIH Image, examine a histogram of the image.

Make sure that the image window is active by clicking anywhere within the window.

Now hit the Command+H key combination.

This will open a separate window with a histogram showing the distribution of gray scale values (from 0-255) on the x axis, and the number of pixels for each gray scale value on the y-axis. Ideally, the histogram should be spread out over the full range of values. The most common form of error is to have an extremely contrasty image with all the values clustered at one extreme end of the histogram.

Learn to use the adjustments to black level and gain on the CLSM to optimize the spread of gray values. If you have to trade off between using the "+1 to +3 LUT" position on the BioRAD versus a brighter setting of the laser, choose the brighter laser setting. It will burn out your specimen faster, but give you a better signal to noise ratio and a wider dynamic range in gray scale value. The Photon Counting mode, or the Accumulate mode on the BioRAD will often provide an image with the best dynamic range. On the BioRAD, I prefer to use the Accumulate Mode with Slow Scan for best results. It is not the purpose of this manual to teach the use of CLSMs, but I only wish to emphasize the importance of the quality of the original image.

Avoid high contrast images
If you get into the habit of checking the histogram of your images as you collect them on the CLSM, you will improve the quality of the original images, and find less need to twiddle with the LUT values.

Avoid noisy images - If possible, collect images using the F1 (Slow) setting (on the BioRAD) with a Kalman setting of at least 3, or Accumulate mode. When using accumulate mode, you can use a less intense laser source, and manually accumulate until the image appears satisfactory.

2. Editing Image

A. Cropping images, erasing, superimposing text, scale bars, rotating and scaling images, etc.

NIH Image provides a wide range of tools for editing your image. These are described in NIH Image manual About NIH Image .

Notes on Scaling: You may find that you want to enlarge a selected portion of an image to emphasize a particular point. For optimal images, do not use the magnifying tool for enlarging images prior to printing. Use the "Scale and Rotate" function of the Edit Menu. Select the option "Bilinear" Interpolation Method, rather than "Nearest Neighbor." The resultant image will appear much smoother, with less "pixellation". The Bilinear function will operate correctly on an Indexed Color image. In order to avoid pixellation in this instance, move the components of the RGB stack to individual Windows, Scale/Bilinear, then move them back to an RGB stack and convert to 8 bit color. The macro "Crop and Scale-Smooth" will simplify this operation on stacks.

3. Using LUTs:

The following operations are commonly used to enhance the image. These are standard operations on all confocal scopes, and are based on methods that are widely used for manipulating digital images. See appropriate section of the NIH-Image manual.

A. Modifying brightness and contrast

In the simplest operation involving lookup tables, you may choose to emphasize a selected range of index values (gray scale values), and minimize other values. The choice may be based on your interpretation of the information content of the image. Learning how to interpret the histogram, and modifying brightness and contrast are essential skills in image processing.

B. Linear and non-linear LUTs, including custom LUTs.

The standard LUT shown in the Map window (lower left side of your screen) is a linear LUT. You can modify the brightness and contrast values by either dragging the Brightness anad Contrast slider buttons, or by directly dragging the plotted line in the map window.

The "Confocal Macros" file included with this manual provides a number of alternate, non-linear, LUTs based on sampling the values within your display buffer. These include Log, Parabolic, Square, Square Root and Gamma transforms. Play with each of these and observe the effects on your images. You will probably find the Gamma Transforms most useful when used with a setting of 1.5 to 2.0. Many of the resultant changes produce results similar to those obtained on the BioRAD with +1, +2 and +3 LUT settings. When you open BioRAD files with NIH Image, the results may not match what you saw on the BioRAD. Most commonly, the image will appear much darker, and lacking the detail you so clearly remember seeing on the screen when collecting the original image. NIH Image has not corrupted your files. This is most frequently due to the fact that the image you viewed on the confocal microscope may have had a non-linear Output-LUT attached to it. The original data is imported intact, but the Output-LUT is not attached to the new NIH Image file, and a new Linear LUT is appended to the file. See NIH Image manual for modifications of the LUT.

C. Enhance Contrast Operator in NIH-Image

The Enhance Menu item in NIH-Image has a specific function entitled "Enhance Contrast". This produces a custom linear LUT effect that is as good as any result I can obtain in my attempts to manually modify the LUT curves. However, I often find that this operation produces an excessively contrasty image, with too steep a slope in the LUT mapping window. If so, then manually change the LUT in the mapping window to give a slope of the LUT about halfway between the original value and that produced by Enhance Contrast . If that is satisfactory, then Apply LUT (Command+L). If you feel that you want still more contrast in the image, repeat the above sequence.

In order to appreciate the effects of this operation on the original image, examine an histogram of the image before and after this procedure.

As in all alterations to the LUT, the Enhance Contrast operation only modifies the display buffer, not the image in the file buffer or the original file on the disk. In order to alter the file buffer, you must perform Apply LUT . This will not alter your original disk file. If you wish to do so, Save the file. Once you have done that, you cannot go back to the previous image. You may, therefore, prefer to save the modified file under an alternate name.

Enhance Contrast is very different from the Equalize operation, which is likely to result in excessively splotchy images. Compare the effect of each of these operations on the appearance of the Mapping, Histogram and LUT.

D. Thresholding and Density Slicing

See NIH Image manual.

E. Pseudocoloring images.

Pseudocoloring confocal images assists the viewer in displaying double labeled sections, detecting major differences in concentration, or changing concentrations, as in Calcium ratio imaging. See appropriate menu item, and NIH-Image manual for use of this operation. All the color applications describedin this manual, although they may resemble the original colored fluorescent image, are pseudocolor.
F. Exporting to Photoshop
See preceding comments.

4. Enhancing/Filtering image

NIH-Image provides a wide range of tools for enhancing and filtering images, including filters to sharpen, smooth, shadow, detect edges, etc. You can write your own kernels, run median filters, Sobel operators, etc. Some of these operations are multistage operations and alter both the display buffer and the file buffer (but not the disk file), and cannot be Undo ne. You will have to reopen the original file to restore the image.

For further details, see NIH Image manual.

5. Quantitative Measurements:

NIH Image was originally developed for quantitative measurements. There are many useful functions described in the About NIH Image manual. Various functions include: Length, Area, Density, Particle counting, Profile of a Line, etc. About NIH Image will provide guidance in the use of these operations.

6. NIH Macro Language

One of the most powerful features of NIH Image is the Macro language. This allows the user to write simple Pascal-like scripts. The distribution kit provided from the FTP site contains many sample macros that can be readily modified to your particular needs.

IV. Advanced topics:

1. Merging pairs of double labeled sections

One of the most valuable and commonly used features of confocal microscopy is the ease of obtaining images of double and triply labeled histological sections. There are two major obstacles that may cause difficulties when attempting to merge two images: 1) The contrast and brightness values of one section may be markedly different from the other(s). This is best dealt with by careful evaluation of the images at the time of original data collection, and modifying your means of collection; and 2) The two series of images of such a pair may not be in perfect register with each other. This may be due to various factors, including misalignment of the PMTs, the mirrors, filter blocks, etc. Most of the errors appear to occur in translation (X and Y axes) and not in rotation. A shift of a few pixels may not be noticed, but occasionally the error results in a marked shift from one color plane to the second. Simple translation errors can be corrected by shifting the images one or more pixels at a time. Rotational errors are more difficult to correct, take longer, and frequently result in image warping. If you find that you have marked rotational errors, the Service Personnel from the manufacturer of your confocal should deal with this.

NIH Image provides an alignment operation, Register. This can be used on sections in a Stack (see below).

A. NIH-Image and Adobe Photoshop

Two or three gray scale images of different fluorophores can be combined into a single colored image, with each fluorophore represented by a different color. The color chosen to represent each fluorophore is arbitrary, and can differ from the original one. There are two different programs that can be used successfully for this purpose, NIH-Image and Adobe Photoshop. There are benefits and disadvantages in the use of each program. Photoshop is relatively expensive, but a superb commercial program. It supports 24 bit images and allows almost instantaneous adjustment of the individual R, G and B planes of a merged image. The NIH-Image produces an 8-bit custom palette of the merged image. It takes 5-10 seconds to produce this image on a IIfx, and is correspondingly faster on a Quadra 950 and a PPC. Although this is an 8 bit color image, the result is often satisfactory, and occasionally even comparable to that obtainable with Photoshop. However, NIH Image may produce excessive dithering of the resultant image, and you cannot make small adjustments in brightness or contrast of the final color image obtained with NIH-Image. Instead, you have to go back to the original gray scale images, modify them, and then once again Merge the images.

Since each Indexed Color image produced with NIH-Image has its own unique LUT, you cannot directly do a side-by-side comparison of two different color images if you Merged using "Custom Colors", as the color values shift markedly as you change windows. If you selected "System Colors", the quality of the color is more limited, but you will be able to compare results with other windows merged using the same System LUT. This is a major disadvantage in relying exclusively on NIH-Image. However, for routine operations, NIH-Image is satisfactory.

In comparison, Photoshop, using a full 24 bit window, allows you to compare multiple colored images simultaneously on a single screen. Photoshop also provides an excellent range of filters, convolutions, etc. The more immediate advantages of NIH-Image are manifest in measurement capability, generating stacks, Z-Series projections, and 3D projections and rotations. Photoshop does not provide such facilities.

Until recently (prior to Version 1.56), NIH Image could only run under an 8 bit monitor setting. If you wanted to shift back and forth from NIH Image to Photoshop, you had to reset the monitor to 24 bit. Beginning with version 1.56, NIH Image operates satisfactorily with the monitor set to 24 bits.

NIH Image Version 1.59 now directly allows the user to save an RGB stack of three sections in a format that can directly be read by Adobe PhotoShop 3.0. If the file is modified in PhotoShop, then saved as an Adobe TIFF file, it can be re-opened by NIH Image as a three slice stack. However, if you add additional Layers or Channels in PhotoShop, it forces you to Save the file in Adobe PhotoShop format. This cannot be read by NIH Image. An NIH Image Z-Series containing Merged Color Slices cannot be read by Adobe PhotoShop.

B. Double Labeled Sections: Building a Stack (Best done using a Macro)

a) Color Merge in Image :

Close other Windows. Once you become more adept at handling and opening new stacks (using a Macro), it will not be necessary to close other windows.

Two color plate: If you wish to combine only two plates, open the two files.

Use the Macro "Color Merge Two Images" contained in the sample Confocal Macros that is provided with this manual. If you examine the sequence of commands in the Macro, you will find the following operations:

  1. Open a fresh stack;

  2. paste the "red" image into the first slice

  3. add an additional slice

  4. paste the "green" image into the second slice

  5. add a black (empty) third slice

  6. Merge from RGB to 8-bit color using a custom LUT.

  7. Open a new window containing the merged color image.

    Alternately, you can do this operation manually to familiarize yourself with the procedure. The first file opened will become the Red Plane, the second file the Green Plane. Under Menu item Stack choose Windows to Stack . This will put the two plates into a stack labeled RGB . Any further operations will not alter your original data files, so you can always go back and start again.

    Save a copy of the NIH Image Stack. The following section describes how to use this Stack with Adobe Photoshop 3.0.

    If you want to alter the contrast or brightness of either of the color planes, you have to modify either the original images or those in the RGB stack. I suggest that you limit your initial attempts to the slices in the RGB stack. When you change the Brightness or Contrast of any single image, you must then "Apply LUT" (Under the Enhance Menu) to that single image. Do not use the "Apply LUT to Stack " macro as this will modify all the images in the Stack.

    Using Menu Stacks select "RGB to 8-bit Color ". In the resulting Dialog box, select Custom Colors . This will generate an Indexed Color window from the Stack, but will not alter the stack itself.

    Since each original file may have a different distribution of values (see Histogram), the saturation of each color plane may differ markedly. Be creative, try different ways of doing it. e.g. Use Enhance Contrast operation on one slice at a time. Try nonlinear LUTs, various filters, etc..

    The resulting Indexed Color Image can be saved with a unique name.

    Once you have mastered this sequence, you will have a clearer understanding of the operation of the Macro "Color Merge Two Images" contained in the sample Confocal Macros that is provided with this manual.

    If you now Save the RGB stack (RGB TIFF) in NIH Image, the resulting file can be viewed with PhotoShop as a 24 bit file.

    b) Merging Files with PhotoShop

    The recently released version of PhotoShop 3.0.1 has excellent layer management.

    Version 1.58/1.59 now permits NIH Image stacks to be directly saved in Photoshop 3.0 TIFF format.

  8. The stack must consist of 3 slices.

  9. Before saving the stack, open the Slice Info menu item in the Stacks menu. Confirm that RGB is selected.

  10. Save the file. It will write the file as an RGB TIFF file.

    You can then directly open this NIH Image Stack in PhotoShop 3.0.

    An alternate means of converting the stack to an RGB TIFF format is to call the RGB to Color function. This will automatically change the Slice Info window to RGB. The resulting Indexed Color image will give an approximate (8 bit) preview of the results to be obtained with Photoshop (24 bit).

    In the event that the stack was not saved in the preceding manner, you still can open an NIH Image stack within Photoshop in the following manner

1) Under the Photoshop File Menu, select Open.
2) Using the PhotoShop Open dialog, go to the desired directory containing the NIH Image Stack file of interest.
3) Using the same Dialog Box, pull down list at the bottom of dialog box, choose "Raw" file format.
4) This will show all files in the chosen Folder.
5) Select the NIH Image Stack. The selected stack much contain at least 3 images. You have to know the dimensions of the images (When using a single field BioRAD image, this may be 768x512. If you used a split field screen when originally collecting images on the BioRAD, and you then split the screen to a stack with the Macro provided with this package, then the image size is frequently 386x512).
6) The resulting dialog box requests that you fill in
a) Width (e.g. 386 pixels);
b) Height (e.g. 512 pixels);
c) Number of channels (e.g., 3, for the number of slices in the stack);
d) Select "not interleaved";
e) Offset =768 (3 X 256 = 768).
7) The originally NIH Image stack will now open as a three channel color stack in Photoshop. However, the colors in the image are inverted. Invert (Command+I) the image to obtain an appropriately colored image.
8) Save As... Save the new Photoshop image using the "Save As..." dialog box. Do Not save as a Raw image. Select TIFF format, and choose a new name for the file, otherwise the simple Save command will overwrite the original NIH Image file and store the data in Raw format.
9) You now will be able to use the various Photoshop tools to modify the resulting 24 bit image. You should explore the use of the Levels command (Command+L) as well as the Brightness/Contrast command (Command+B).
You can modify each color channel separately using several different methods, as described in the PhotoShop 3.0 manual.
The quality of the resulting color image will generally be much better than that provided by NIH Image, as it can utilize a full 24 bit lookup table, and does not require dithering of the image.
The Photoshop manual and tutorial will provide further guidance in modifying the images.
If you have a quadruple (or more) labeled section (e.g. four fluorophores, or three fluorophores and a Nomarski/DIC image), and have stored them in a four slice stack in NIH Image, you can Open them in PhotoShop using the method described above.


c) Preferred method
It is obviously easier to use the new procedure provided in Version 1.58.

The first slice of the NIH Image Stack forms the Red layer, Green the second layer, and the third slice forms the Blue layer.

The gamma, brightness and contrast of each layer can be individually modified with immediately evident results. For further details, see the PhotoShop 3.0 manual.

C. Compare results of NIH Image 8-Bit merge with Photoshop 24 bit merge

The quality of the color image (assuming your monitor is set for 24 bit color) is generally much better in Photoshop than the optimized 8 bit image obtained in NIH Image.

D. Adjusting color contrast on sections in a stack (See Macros)

(Do Enhance Contrast operation separately on each section in the stack. Failure to do so will result in merging of saturated and unsaturated images.)

E. Merging a doubly labeled pair of Z-series using a Macro.

Open the two stacks. The first stack will be assumed to be the "red" stack. The second stack will be the "green" stack.

Close all other windows.

Select the Macro Color Merge Two Stacks.

Lean back and watch the fun.

When completed, select the Merged stack and Animate it, or step through it with the "<" and ">" keys.

2. Projection of a Z-Series and 3D Rotations

Many of the most valuable qualities of confocal Z-series are that the data can be manipulated to obtain "through views" of the stack, the stack can be resliced at various angles converting a transverse view into a sagittal or horizontal plane, it can be used to generate 3D images, stereo pairs, and other operations. Indeed, NIH Image is able to perform operations as well as much of the software provided by the manufacturers of Confocal microscopes, and equivalent to many of the basic operations provided by very expensive image processing programs such as VoxelView, VolVis and VoxBlast running on a Silicon Graphics Workstation. For more elaborate operations using true Voxels, complex shading and raytracing in 24 bits, these latter programs are excellent choices. For the most common operations, however, NIH Image is quite adequate. Optimal performance on these tasks will be achieved using a PowerPC with sufficient memory to handle your typical datasets.

A. Stepping through a Z-series using Stacks.

Open a Z-series in a Stack.

Use < and > to step through the plates in stack, one at a time. The reader is urged to read the section in the NIH-Image manual regarding Stacks.

B. Animating a Z-series

Set [Special]-[Video]-for oscillating movies. This will produce smooth back-and forth motion of the Z-series animation.

Command+= to animate a series. Number keys 1-9 control the speed of the animation.

C. Projecting a Z-Series onto a single plane

This is an extremely useful benefit of the confocal microscope. A stack of Z-series sections are all in optimal focus. If projected onto a single image plane, all object throughout the thickness of the imaged section will now be in sharp focus and spatial relationships may be more evident.

Version 1.58 provides macro calls to perform the Z--Projection. A sample macro to accomplish this operation is included in the accompanying macro files.

Selection of area for Z-Projection:

  1. Choose an extended Z-series (e.g. more than 10-20 images) with well defined profiles of objects.

  2. Run the macro "Project Z Series".

    You will gain a better unbderstanding of this procedure if you examine the macro, and will find that it merely automates the following procedures.

    Use Project command.

    Select Brightest Point.

    Set Initial Angle to 0, Increment and Total Rotation also to 0.

    Set Transparency values, as desired. Play with different values. Start with the Default Values. Once you are more familiar with the Thresholding tool (Magic Wand), you will be able to use that to set the range of values desired for your Z-series.

    This will generate a single Z-Projection of all the plates in the Stack, viewed from directly in front of the stack.

    Experiment with different view angles by using different values for the Initial Angle , while leaving Total Rotation at 0.

    D. Re-Slicing the Z-Series along alternate planes: (XZ, YZ and theta-Z)

    One of the most useful features of NIH Image is the ability to rapidly re-slice a Z-series data stack along alternate planes. In addition to simple orthogonal planes (i.e. XZ and YZ), you can reslice at any arbitrary angle between X and Y to generate a single Z plane cross section. This is very useful for evaluating the extent of penetration of antibodies into tissue, evaluating cell spacing, etc.

    The closer the images in the original Z- series, and the greater their number, the more natural appearing the final results. In order to provide a seemingly continuous image in the resliced plane, the software interpolates gray values. For improved quality of re-sliced images, your original Z-series should be as close as possible (.e.g 0.25 micrometers).

    Have you set the calibration for magnification and slice spacing, as described above? (Also see Image Manual).

    Your original data file may contain the needed information about slice spacing. In many of our files, the step size was 0.38 m (on a BioRAD), and the Pixel size 0.105 m (9.5 pixels per m). However, this depends upon the Z-axis step size, the objective and the zoom factor.

    Pull down the Stacks Menu, and select "Options" from the list. Enter the appropriate value in the slice spacing dialog.

    Select an image from the series that best shows the features of interest. Using the Calibration/Measurement tool (Dashed Line in right hand column of Tool Palette). Draw a line across the section along the desired plane of resectioning. If you wish to constrain the plane to either X or Y, press the Shift key and hold down as you draw the line.

    Under Stacks Menu, choose "Re-Slice" ("also Command /"). If circular profiles appear excessively flattened in either the X or Y axis, experiment with different values of "Slice Spacing" in the Options dialog.

    E. Re-slicing to make a Z-series in an alternate plane:

The preceding instructions describe how to obtain a single re-slice in the XZ or YZ plane. If you would like to obtain a new stack of Z-series section, not just a single section, use the Macro provided, Re-Slice Horizontally or Re-Slice Vertically . The reslicing is limited to orthogonal planes of section. Thus, if you wish to relice along an oblique angle, you will first have to rotate the stack. In order to do this, use the 'angle measurement' tool from the tool palette. Measure the degrees variance of the object in the image from the angle you finally desire.

Use the Macro Scale and Rotate Stack, using the angle value established in the previous paragraph.

Now select a rectangular area that encloses the region you want to reslice in either horizontal or vertical plane.

Use the Macro Reslice Horizontally or Reslice Vertically .
Rapid Dynamic 3D Reslicing:
Norbert Vischer of the Netherlands has recently contributed an extremely useful routine, "3D Slicer" for reslicing a stack of sections along the X, Y and Z axes. This routine is provided in the program "Object-Image 1.59", available from the NIH Image FTP site. Object-Image 1.59 provides extremely rapid reslicing in real time. The individual resliced sections cannot be saved, for unlike the routines described above, they are not placed in a new stack. Reslicing of sections is accomplished by dragging the mouse along the X, Y or Z axes of a master stack, and results in rapid generation of resliced images in the two other planes. Reslicing can only be done along orthogonal planes.
1) Open a Z-stack of sections, such as the sample MRI Images of a human skull and brain.
2) Enter the correct magnification scale and slice spacing (5.0 mm), as described above.
3) Under Stacks Menu, select 3D Slicer. This will open a new window containing an angled perspective view of the original stack, with 3D projection planes along the two alternate planes, as well. In the upper part of the window will be an image of the resliced section parallel to the X axis. On the right side of the window will be the resliced section parallel to the Y-axis.
4) Place the mouse over the X, Y or Z margin. The cursor changes to a two headed arrow indicating the direction of movement. Drag the X, Y or Z axis on the main central image, and observe the rapid resectioning in the alternate planes.
5) Place the mouse over the intersection of the X and Y axes and the cursor becomes a four-pointed arrow. You can now simultaneously reslice parallel to both the X and the Y axes.
6)Double clicking the mouse on the X or Y axes will turn off the resliciing tool for that axis. Double clicking on the dashed lines on the X or Y axis will restore the reslicing planes.

F. Generating a 3D series from a Z-series.

This is one of the slowest of procedures, and a fast PowerPC will prove very desirable when doing this operation. When you are still learning the basics of this procedure, use the "Selection" Tool from the Palette and outline a small portion of the image when generating a 3D series. It will complete the task much more rapidly.

Use care in settings. Remember to correct for thickness and spacing of individual sections.

Select "Project" from the Stacks Menu.

The Dialog box is one of the most complex that you will encounter when using NIH Image. Initially, follow these instructions:

Set the "Initial Angle" to 0.

Set the "Total Rotation" to 180.

Set the "Incremental Angle" to 15.

Leave the remaining values in their default configuration.

At the bottom of this dialog box are two columns of options:

In the left column, select "Y-Axis"" as the axis of rotation.

In the right column, select "Brightest Point".

Now click OK and wait. The program will generate a new stack of images of the "rotating" objects in the original stack.

Once you have had further experience with this procedure, Project using a full 360 degree rotation at closer intervals, changing transparency values, axes of rotation and the various other options available.

A macro to facilitate this operation is provided.

G. Animating a 3D series (Producing "Apparent Rotation")

Under Special Menu, select Video Options. Check "Oscillate Movies". Close the dialog box.

To generate the impression of rotation with your newly generated 3D stack, use Stacks Menu and select "Animate" (Command+=). This is heavily memory dependent, so keep your initial efforts small.

You can control the speed of rotation with keys 1-9. You can step through single sections using the < and the > keys.

H. Make a Stereo pair or series of stereos?

There are several methods of generating stereo images from a stack of sections collected in the Z-Plane. One simple method, the "pixel shift" method, is that used in the original BioRAD software. A second method is by calculating the perspective from different angles of view by ray-tracing:

Pixel Shift Method: Start with a stack of a Z-series. You will generate two Z-projections from this original stack. One will be "pixel shifted" to the left and a second "pixel shifted" to the right. This is achieved by shifting each successive section in the stack by one additional pixel (or more, if desired) prior to performing the Z projection. The two resultant images are then placed side by side. The Pixel Shift method is generally limited to stereo images centered around the original Plane of image collection.

Projecting and Ray-Tracing: A second method is that of projecting (ray-tracing) the stack onto an imaginary view plane from different angles of view and generating a 3D rotation series. The second method is computationally more complex, but provides the possibility of generating stereo views from any angle around a central point.

Start with the original Z-series Stack. Select the Stacks Menu and Project . Set Initial Angle to 354 , and Total Rotation to 12 . Set Increment to 6 degree increments. This will produce a new Stack containing three slices (354 , 0 and 6 ).

a) Stereo series in black and white

Once you have generated a 3D series, make a Montage series using the appropriate function from the Stacks Menu. The Montage function is non-destructive, i.e. it does not erase the 3D stack, however, I suggest that you Save your work as you go along.

Stacks Menu: Montage

The resulting dialog box will show a series of values that are dependent upon the number of sections in the stack. If you made a simple stack with only 3 sections, then choose 1 row and three columns, and increment of 1. If you generated a full 360 rotation series, you can select any range of slices (e.g. slices #4-8 out of 16 slices), or incremental slices (every 2nd or 3rd slice in the stack), and define the numbers of rows and columns you wish to see displayed.

This will generate a series of images in a new window on the screen. If the resulting images are too large to line up next to each other, use a scaling factor when generating the Montage.

Start with two side-by-side images. To facilitate seeing the stereo effect without special viewers, choose a set of images with a prominent object to provide an alignment cue in the center of each image. Set the image size and separation with the alignment cues no more than 50 mm apart on the screen. Gradually work your way up to greater separations and then slightly larger images. Initially, you may find that you require a stereo viewer to visualize a stereo image from side-by-side images. With practice, you will not require a stereo viewer and should be able to scan across pairs of sections and jump from one stereo image to the next. (The typical interpupillary distance of most people is about 65 mm. You can also practice learning to fuse stereo images by using some of the recent popular books with "random dot stereograms", such as "The Magic Eye.")

b) Stereo Pair of Single Labeled Section in color

An alternate manner of presenting stereo images is to merge a stereo pair into a single image plane, assigning one image to red and the other to green.

Generate a 3D rotation series around the y-axes. Limit the number of image to two images at a separation of approximately 6-12 . For first trials, set the Initial Angle to 354 , the Total Rotation to 12 , and the increment to 6 . This will generate three plates in a new stack. Animate the new stack of images, and if the animation produces the desired effect, delete either the middle or one of the end plates. Add a black plate as slice 3/3. (See section above on Merging Slices to generate 8 bit color images). Using the Stacks Menu item RGB to 8-bit color, you will obtain a single merged image which can be viewed with a pair of Red/Green stereo glasses.

The greater the separation between plates (i.e. 12 rather than 6 ) the more exaggerated the stereo effect. Play with different values. Some people find that angles greater than 15-18 are excessive.

c) Color Stereo images of a double labeled section:

Although this is a more complex set of operations, it is a logical extension of the methods described above.

First generate a 3D rotation series of each Z-series of a double labeled section..

Save the results.

Close all windows except those of the two 3D rotation series.

Now use the macro Color Merge of Two Stacks. You will now have a rotation series (3D) of two simultaneous different fluorophores.

Animate the series.

Now generate stereo pairs, using the Montage method described above.

How do you like them Apples?

V. Printing on Video Printer, Slide Maker, VCR

You can obtain an instant print of the Image window using a Video Printer, such as the Mitsubishi 67U or Sony thermal video printer. This provides a gray scale print. The Mitsubishi produces a print of approximately 3x4 inches, and provides a running record of your results. Scion markets a board (TV-3) for the NuBus that will directly send the contents of a selected window to an NTSC device. The driver for the TV-3 board must be placed in the NIH Image Plug-Ins folder. The drawback of the TV-3 is that it can only send out an image of maximum dimension of 640x480 pixels. Since the standard BioRAD image is 768x512, the image must first be scaled to a 640x480 size, or select a portion of the image corresponding to this size. Images smaller than 640x480 will only fill a portion of the printed field. It would be useful if the driver provided an autoscaling function. A simple macro, however, can be written to accomplish a similar result. The new PowerPC 7500 and 8500 provide direct NTSC and S-Video output. I have not had a chance to test them with the Mitsubishi printer, but based on published specifications, these should provide results similar to those obtained with the Scion TV-3 board.

Prints of much better quality, including black and white or color prints, can be obtained using the Kodak 8600 Dye Subimation Printer. The printer is provided with a Photoshop Plug-In driver. Place this Plug-In in the folder of the same name in the main NIH-Image directory. When you want to print the contents of a window, select the desired image. Under the File Menu, choose Export , and select the Kodak 8600 printer driver. The typical image will print in approximately one minute. This works best with monochrome images printed on a Black ribbon. Color prints are best produced using PhotoShop. A major limitation in the use of the Kodak printer is the cost of each print ($2-$3), as well as the cost of the printer itself (ca. $8,000).

The wider availability of Color Laser printers may provide satisfactory color prints at lower cost than those obtained with the Kodak 8600 printer.

Slides : There are a number of slidemakers, including the LFR Lasergraphics, the GCC/Polaroid slidemaker, Agfa and others.

If you are using PowerPoint to send images to a slidemaker, Scale the images to the full size of the screen (e.g. 768x512) prior to saving them. Use the Bilinear Interpolation, rather than Nearest Neighbor, for best results. You can also use PowerPoint to combine different images on a single slide, add text, or other graphics elements. Paste the NIH Images onto a Black Background to avoid having a white border around the image.

VI. Use of NIH-Image for Image collection, and instrument control

The majority of CLSMs have their own dedicated software for image collection. However, several instruments, such as direct viewing slit scanners, and the AOD real-time scanner have also been used directly with NIH-Image for data collection.