Image Types and Formats DICOM(Digital Imaging and Communications in Medicine) is a standard popular in the medical imaging community. Support in ImageJ is limited to uncompressed DICOM files. DICOM files containing multiple images open as Stacks Use Imaged Show Info . to display the DICOM header information. A DICOM sequence can be opened using Fileb Import D Image Sequence .. or by dragging and dropping the folder on the 'ImageJ window. Imported sequences are sorted by image number instead of filename and the tags are preserved when DICOM images are saved in TIFF format. ImageJ supports custom DICOM dictionaries, such as theoneathttp://imagej.nihgov/ij/download/docs/dicom_dictionary.txtMore information can be found at the Center for Advanced Brain Imagin FITS(Flexible Image Transport System) image is the format adopted by the astronomical community for data interchange and archival storage. Use ImageD Show Info.. [to display the FITs header. More information here PGM(Portable Gray Map), PBM (Portable BitMap) and PPM(Portable PixMap)are simple image formats that use an ASCII header. More information here. AVi (Audio Video Interleave) is a container format which can contain data encoded in many different ways. ImageJ only supports uncompressed AVIs, various YUV 4: 2: 2 compressed formats, and PNG or JPEG-encoded individual frames. Note that most MJPG (motion-JPEG) formats are not read correctly. Attempts to open AVIs in other formats will fail SEE ALSO: Non-native Formats, II Image Types: Lossy Compression and Metadata, X War on JPEG Compression Non-native Formats When opening a file, Image J first checks whether it can natively handle the format. If Image does not recognize the type of file it calls for the appropriate reader plugin using HandleExtraFile Types, a plugin bundled with ImageJ. If that fails, it tries to open the file using the Ome Bio-Formats library (if present), a remarkable plugin that supports more than one hundred of the most common file formats used in microscopy. If nevertheless the file cannot be opened, an error Because both these plugins are under active development, it is important that you keep them updated. The OME Bio-Formats library can be updated using its self-updating plugin(Pluginsb LOCID Update LOCI Plugin. .)or Fijit's built-in updater(HelpD Update Fiji...). The following websites provide more information on the OmE Bio-Format http://loci.wisc.edu/bio-formats/image http://fiji.sc/bio-fOrmat http://loci.wisc.edu/bio-formats/using-bio-format In addition, the ImageJ web site lists more than sixty plugins that recognize more ' exotic'file formats. The ImageJ Documentation Portal also maintains a(somewhat outdated list of file formats that are supported by ImageJ SEE ALSO: Native Formats, FileD Import D, II Image Types: Lossy Compression and Metadata, X Warning on JPEG Compression, Acquisition plugins, Input/Output plugins Last updated: 2012/10/02
Image Types and Formats DICOM (Digital Imaging and Communications in Medicine) is a standard popular in the medical imaging community. Support in ImageJ is limited to uncompressed DICOM files. DICOM files containing multiple images open as Stacks. Use Image . Show Info. . . [i] to display the DICOM header information. A DICOM sequence can be opened using File . Import . Image Sequence. . . or by dragging and dropping the folder on the ‘ImageJ’ window. Imported sequences are sorted by image number instead of filename and the tags are preserved when DICOM images are saved in TIFF format. ImageJ supports custom DICOM dictionaries, such as the one at http://imagej.nih.gov/ij/download/docs/DICOM_Dictionary.txt. More information can be found at the Center for Advanced Brain Imaging. FITS (Flexible Image Transport System) image is the format adopted by the astronomical community for data interchange and archival storage. Use Image . Show Info. . . [i] to display the FITS header. More information here. PGM (Portable GrayMap), PBM (Portable BitMap) and PPM (Portable PixMap) are simple image formats that use an ASCII header. More information here. AVI (Audio Video Interleave) is a container format which can contain data encoded in many different ways. ImageJ only supports uncompressed AVIs, various YUV 4:2:2 compressed formats, and PNG or JPEG-encoded individual frames. Note that most MJPG (motion-JPEG) formats are not read correctly. Attempts to open AVIs in other formats will fail. See also: Non–native Formats, II Image Types: Lossy Compression and Metadata, X Warning on JPEG Compression Non–native Formats When opening a file, ImageJ first checks whether it can natively handle the format. If ImageJ does not recognize the type of file it calls for the appropriate reader plugin using HandleExtraFileTypes, a plugin bundled with ImageJ. If that fails, it tries to open the file using the OME Bio-Formats library (if present), a remarkable plugin that supports more than one hundred of the most common file formats used in microscopy. If nevertheless the file cannot be opened, an error message is displayed. Because both these plugins are under active development, it is important that you keep them updated. The OME Bio-Formats library can be updated using its self-updating plugin (Plugins . LOCI .Update LOCI Plugin. . .) or Fiji↑’s built-in updater (Help .Update Fiji. . .). The following websites provide more information on the OME Bio-Formats: – http://loci.wisc.edu/bio-formats/imagej – http://fiji.sc/Bio-Formats – http://loci.wisc.edu/bio-formats/using-bio-formats In addition, the ImageJ web site lists more than sixty plugins that recognize more ‘exotic’ file formats. The ImageJ Documentation Portal also maintains a (somewhat outdated) list of file formats that are supported by ImageJ. See also: Native Formats, File . Import . , II Image Types: Lossy Compression and Metadata, X Warning on JPEG Compression, Acquisition plugins, Input/Output plugins 11 Last updated: 2012/10/02
Stacks, Virtual Stacks and Hyperstacks II IMAGE TYPES: LOSSY COMPRESSION AND METADATA Two critical aspects to keep in mind when converting images Lossy compression Transcoding an image into a format that uses lossy compression will alter the original data, introducing artifacts(see X Warning on JPEG Compression). This is the case, e. g, for JPEG formats(with the exception of some JPEG2000 images that use lossless compression). As such, these types of data are intended for human interpretation only and are not suitable for quantitative analyses Metadata In Image J, metadata associated with the image, such as scale, gray value calibration and user comments is only supported in tiff and zip(compressed tiff)images. In addition, selections and Overlays are also saved in the TIFF header (cf. File b Save [s). None of the above is saved in other formats(cf. Native Formats) 8 Stacks, Virtual Stacks and Hyperstacks Stacks ImageJ can display multiple spatially or temporally related images in a single window. These image sets are called stacks. The images that make up a stack are called slices. In stacks, a pixel (which represents 2D image data in a bitmap image) becomes a voxel(volumetric pixel),i.e,an intensity value on a regular grid in a three dimensional space. All the slices in a stack must be the same size and bit depth. A scrollbar provides the ability to move through the slices and the slider is preceded by a play /pause icon that can be used to start/stop stack animation. Right-clicking on this icon runs the Animation Options .. [Alt/ dialog box Most ImageJ filters will, as an option, process all the slices in a stack. Image opens multi-image TifFfilesasastackandsavesstacksasmulti-imageTifFs.theFilebiMportbRaw...command opens other multi-image, uncompressed files. A folder of images can be opened as a stack either by dragging and dropping the folder onto the 'ImageJ window or or by choosing FileD Import Image Sequence.. To create a new stack, simply choose Fileb New b Image..[n] and set the Slices field to a value greater than one. The Imaged Stacks b submenu contains commands for common stack operations SEE ALSO: Stacks Menu, Stack Manipulations on Fiji website, ImageSD Virtual stacks Virtual stacks are disk resident(as opposed to RAM resident) and are the only way to load image sequences that do not fit in RAM. The ere are sever al things to keep in mind when workin with virtual stacks Virtual stacks are read-only, so changes made to the pixel data are not saved when you switch to a different slice. You can work around this by using macros(e. g, Process Virtual Stack)ortheProcessbBatchbVirtualStack...command You can easily run out of memory using commands like Image b Crop X because any stack generated from commands that do not generate virtual stacks will be RAM resident Last updated: 2012/10/02
Stacks, Virtual Stacks and Hyperstacks II Image Types: Lossy Compression and Metadata Two critical aspects to keep in mind when converting images: Lossy compression Transcoding an image into a format that uses lossy compression will alter the original data, introducing artifacts (see X Warning on JPEG Compression). This is the case, e.g., for JPEG formats (with the exception of some JPEG2000 images that use lossless compression). As such, these types of data are intended for human interpretation only and are not suitable for quantitative analyses Metadata In ImageJ, metadata associated with the image, such as scale, gray value calibration and user comments is only supported in tiff and zip (compressed tiff) images. In addition, selections and Overlays are also saved in the TIFF header (cf. File . Save [s]). None of the above is saved in other formats (cf. Native Formats). 8 Stacks, Virtual Stacks and Hyperstacks Stacks ImageJ can display multiple spatially or temporally related images in a single window. These image sets are called stacks. The images that make up a stack are called slices. In stacks, a pixel (which represents 2D image data in a bitmap image) becomes a voxel (volumetric pixel), i.e., an intensity value on a regular grid in a three dimensional space. All the slices in a stack must be the same size and bit depth. A scrollbar provides the ability to move through the slices and the slider is preceded by a play/pause icon that can be used to start/stop stack animation. Right-clicking on this icon runs the Animation Options. . . [Alt /] dialog box. Most ImageJ filters will, as an option, process all the slices in a stack. ImageJ opens multi-image TIFF files as a stack, and saves stacks as multi-image TIFFs. The File . Import . Raw. . . command opens other multi-image, uncompressed files. A folder of images can be opened as a stack either by dragging and dropping the folder onto the ‘ImageJ’ window or or by choosing File . Import . Image Sequence. . . To create a new stack, simply choose File .New. Image. . . [n] and set the Slices field to a value greater than one. The Image . Stacks . submenu contains commands for common stack operations. See also: Stacks Menu, Stack Manipulations on Fiji website, Image5D Virtual Stacks Virtual stacks are disk resident (as opposed to RAM resident) and are the only way to load image sequences that do not fit in RAM. There are several things to keep in mind when working with virtual stacks: – Virtual stacks are read-only, so changes made to the pixel data are not saved when you switch to a different slice. You can work around this by using macros (e.g., Process Virtual Stack) or the Process . Batch . Virtual Stack. . . command – You can easily run out of memory using commands like Image . Crop [X] because any stack generated from commands that do not generate virtual stacks will be RAM resident. 12 Last updated: 2012/10/02
Stacks, Virtual Stacks and Hyperstacks 000 MAX IAAA MAY mitosis 15x CUZt1950 Stacks and Hyperstacks in ImageJ: Fileb Open Samples Mitosis(26MB, 5D stack). Hyperstacks dimensionality can be reduced using Imag Tool. .[Z] The (V) on the window title denotes a virtual b Hyperstacks b Reduce Dimensionality., Image b Stacksb Z or ImageD Hyperstacks b Channels image(see Virtual Stacks) TIFF virtual stacks can usually be accessed faster than JPEG virtual stacks. A JPEG sequence can be converted to TiFF by opening the JPEG images as a virtual stack and ing Fileb Save Asb Image Sequence .. to save in TIFF format mageJ appends a(V), to the window title of virtual stacks and hyperstacks(see Hyperstacks) Several built-in Image J commands in the File b Import b submenu have the ability to open virtual stacks, namely: TIFF Virtual Stack., Image Sequence., Raw ., Stack From List ., AVI (cf. Virtual Stack Opener). In addition, TIFF stacks can be open as virtual stacks by drag and drop(cf. III Opening Virtual Stacks by Drag &e Drop) SEE ALSO: LOCI Bio-Formats and Register VirtualStackSlices plugins, Process Virtual Stack and virtualStack From List macros iii OPENING VIRTUAL STACKS BY drag drop TIFF stacks with a E. tif extension open as virtual stacks when dragged and dropped on the toolbar icon ImageJ 口Qa&图、AQ九三应 <<Open as Virtual Stack>> Hyperstacks bidimensional images, extending image to four(4D) or five(5D) dimensions: a(width), y(height), a(slices), c(channels or wavelengths) and t(time frames) Hyperstacks are displayed in a window with three labelled scrollbars(see Stacks and Hyperstacks) Similarly to the scrollbar in Stacks, the frame slider(t) has a play/pause icon SEE ALSO: Image b Hyperstacks b submenu Last updated: 2012/10/02
Stacks, Virtual Stacks and Hyperstacks Stacks and Hyperstacks in ImageJ: File . Open Samples . Mitosis (26MB, 5D stack). Hyperstacks dimensionality can be reduced using Image . Hyperstacks . Reduce Dimensionality. . . , Image . Stacks . Z Project. . . or Image . Hyperstacks . Channels Tool. . . [Z] The ‘(V)’ on the window title denotes a virtual image (see Virtual Stacks). – TIFF virtual stacks can usually be accessed faster than JPEG virtual stacks. A JPEG sequence can be converted to TIFF by opening the JPEG images as a virtual stack and using File . Save As . Image Sequence. . . to save in TIFF format ImageJ appends a ‘(V)’ to the window title of virtual stacks and hyperstacks (see Hyperstacks). Several built-in ImageJ commands in the File . Import . submenu have the ability to open virtual stacks, namely: TIFF Virtual Stack. . . , Image Sequence. . . , Raw. . . , Stack From List. . . , AVI. . . (cf. Virtual Stack Opener). In addition, TIFF stacks can be open as virtual stacks by drag and drop (cf. III Opening Virtual Stacks by Drag & Drop). See also: LOCI Bio-Formats and RegisterVirtualStackSlices plugins, Process Virtual Stack and VirtualStackFromList macros III Opening Virtual Stacks by Drag & Drop TIFF stacks with a .tif extension open as virtual stacks when dragged and dropped on the toolbar icon. Hyperstacks Hyperstacks are multidimensional images, extending image stacks to four (4D) or five (5D) dimensions: x (width), y (height), z (slices), c (channels or wavelengths) and t (time frames). Hyperstacks are displayed in a window with three labelled scrollbars (see Stacks and Hyperstacks). Similarly to the scrollbar in Stacks, the frame slider (t) has a play/pause icon. See also: Image .Hyperstacks . submenu 13 Last updated: 2012/10/02
Color Images 9 Color images ImageJ deals with color mainly in three ways: pseudocolor images, RGB images, RGB/ HSB acks, and composite images Pseudocolor images A pseudocolor (or indexed color) image is a single channel gray image(8, 16 or 32-bit)that has color assigned to it via a lookup table or LUT. A LUT is literally a predefined table of gray values with matching red, green and blue values so that shadows of gray are displayed as colorized pixels. Thus, differences in color in the pseudo-colored image reflect differences in ensity of the object rather than differences in color of the specimen that has been imaged 8-bit indexed color images(such as GIFs)are a special case of pseudocolor images as their lookup table is stored in the file with the image. These images are limited to 256 colors(24-bit RGB images allow 16.7 million of colors, see Image Types and Formats) and concomitantly smaller file sizes. Reduction of true color values to a 256 color palette is performed by color quantization algorithms. ImageJ uses the Heckbert's median-cut color quantization algorithm(see ImageD TypeD menu), which, in most cases, allows indexed color images to look nearly identical to their 24-bit originals SEE ALSO: Image b Lookup Tables b and LUT Menu True Color images As described in Image Types and Formats, true color images such as RGB images reflect gen colors, i.e., the green in an RGB image reflects green color in the specimen. Color images are typically produced by color CCD cameras, in which color filter arrays(Bayer masks)are placed over the image sensor Color Spaces and Color Separation Color spaces describe the gamut of colors that image-handling devices deal with. Because human on is trichromatic, most color models represent colors by three values. Mathematically, these values(color components) form a three-dimensional space such as the RGB, HSB, CIE Lab or YUV color space RGB (Red, Green, Blue)is the most commonly-used color space. However, other alternatives ach as HSB(Hue, Saturation, Brightness) provide significant advantages when processing color information. In the HSB color space, Hue describes the attribute of pure color, and therefore distinguishes between colors. Saturation(sometimes called"purity"or"vibrancy) characterizes the shade of color, i.e how much white is added to the pure color. Brightness(also know as Value- HSV system) describes the overall brightness of the color(see e. g, the color palette of corresponds to the grayscale version or e.g., since the Brigs s, using the HSB system over the Color Picker window). In terms of digital imaging processin traditional RGB is often advantageous tness component of an HSB image that image, processing only the brightness channel in hissectionispartiallyextractedfromtheMbfimAgeJonlinemanualathttp://www.macbiophotonics.ca mages/colour image processi. htm. Last updated: 2012/10/02
Color Images 9 Color Images1 ImageJ deals with color mainly in three ways: pseudocolor images, RGB images, RGB/ HSB stacks, and composite images. Pseudocolor Images A pseudocolor (or indexed color) image is a single channel gray image (8, 16 or 32–bit) that has color assigned to it via a lookup table or LUT. A LUT is literally a predefined table of gray values with matching red, green and blue values so that shadows of gray are displayed as colorized pixels. Thus, differences in color in the pseudo-colored image reflect differences in intensity of the object rather than differences in color of the specimen that has been imaged. 8-bit indexed color images (such as GIFs) are a special case of pseudocolor images as their lookup table is stored in the file with the image. These images are limited to 256 colors (24–bit RGB images allow 16.7 million of colors, see Image Types and Formats) and concomitantly smaller file sizes. Reduction of true color values to a 256 color palette is performed by color quantization algorithms. ImageJ uses the Heckbert’s median-cut color quantization algorithm (see Image . Type . menu), which, in most cases, allows indexed color images to look nearly identical to their 24-bit originals. See also: Image . Lookup Tables . and LUT Menu True Color Images As described in Image Types and Formats, true color images such as RGB images reflect genuine colors, i.e., the green in an RGB image reflects green color in the specimen. Color images are typically produced by color CCD cameras, in which color filter arrays (Bayer masks) are placed over the image sensor. Color Spaces and Color Separation Color spaces describe the gamut of colors that image-handling devices deal with. Because human vision is trichromatic, most color models represent colors by three values. Mathematically, these values (color components) form a three-dimensional space such as the RGB, HSB, CIE Lab or YUV color space. RGB (Red, Green, Blue) is the most commonly-used color space. However, other alternatives such as HSB (Hue, Saturation, Brightness) provide significant advantages when processing color information. In the HSB color space, Hue describes the attribute of pure color, and therefore distinguishes between colors. Saturation (sometimes called “purity” or “vibrancy”) characterizes the shade of color, i.e., how much white is added to the pure color. Brightness (also know as Value – HSV system) describes the overall brightness of the color (see e.g., the color palette of Color Picker window). In terms of digital imaging processing, using the HSB system over the traditional RGB is often advantageous: e.g., since the Brightness component of an HSB image corresponds to the grayscale version of that image, processing only the brightness channel in 1This section is partially extracted from the MBF ImageJ online manual at http://www.macbiophotonics.ca/ imagej/colour_image_processi.htm. 14 Last updated: 2012/10/02
Color Images 心 Representation of an eight pixel color image in the RGB and HsB color spaces. The rGB color space maps the RGB color model to a cube with Red(r) values increasing along the x-axis, Green (G)along the y-axis and Blue(B)along the z-axis. In the HSB cylindrical coordinate system, the angle round the central vertical axis corresponds to Hue(H), the distance from the axis corresponds t Saturation(S), and the distance along the axis corresponds to Brightness(B). In both cases the origin holds the black color. The right panel shows the same image after brightness reduction, easily noted by the vertical displacement along the hsb cylinder. Images produced using Kai Uwe Barthels 3D Color routines that require grayscale images is a significant computational gain. You can read more about the hsb color model here In ImageJ, conversions between image types are performed using the Image b TypeD submenu Segmentation on the HSB, RGB, CIE Lab and YUV color spaces can be performed by the Image D Adjust b Color Threshold.. command (20. Segregation of color components( specially useful for quantification of histochemical staining) is also possible using Gabriel Landini,s Colour Deconvolution plugin. In addition, several other plugins related to color processing can be obtained from the image J website Conveying Color Information People see color with significant variations. Indeed, the popular phrase"One picture is worth ten thousand words"may not apply to certain color images, specially those that do not follow he basic principles of Color Universal Design. Citing Masataka Okabe and Kei Ito Colorblind people can recognize a wide ranges of colors. But certain ranges of colors are hard to distinguish. The frequency of colorblindness is fairly high. One in 12 Caucasian(8%), one in 20 Asian(5%), and one in 25 African(4%)males are so-called'red-green'colorblind. There are always colorblind people among the audience and readers. There should be more than TEN colorblind in a room with 250 people(assuming 50% male and 50% female) I. There is a good chance that the paper you submit may go to colorblind reviewers. Supposing that your paper will be reviewed by three white males(which is not unlikely considering the current population in science), the probability that at least one of them is colorblind is whopping 22%! One practical point defined by the Color Universal Design is the use of magenta in red-green overlays(see also 66). Magenta is the equal mixture of red and blue. Colorblind people that i See Wootton R, Springall DR, Polak JM. Image Analysis in Histology: Conventional and Confocal Microscopy Cambridge University Press, 1995, ISBN 0521434823 This section is partially extracted from Masataka Okabe and Kei Ito, Color Universal Design(CUD)-How tomakefiguresandpresentationsthatarefriendlytoColorblindpeoplehttp://jfly.iamu-tokyo.ac.jp/color/, accessed 2009. 01.15 Last updated: 2012/10/02
Color Images Representation of an eight pixel color image in the RGB and HSB color spaces. The RGB color space maps the RGB color model to a cube with Red (R) values increasing along the x-axis, Green (G) along the y-axis and Blue (B) along the z-axis. In the HSB cylindrical coordinate system, the angle around the central vertical axis corresponds to Hue (H), the distance from the axis corresponds to Saturation (S), and the distance along the axis corresponds to Brightness (B). In both cases the origin holds the black color. The right panel shows the same image after brightness reduction, easily noted by the vertical displacement along the HSB cylinder. Images produced using Kai Uwe Barthel’s 3D Color Inspector plugin. routines that require grayscale images is a significant computational gain1 . You can read more about the HSB color model here. In ImageJ, conversions between image types are performed using the Image .Type . submenu. Segmentation on the HSB, RGB, CIE Lab and YUV color spaces can be performed by the Image .Adjust . Color Threshold. . . command [20]. Segregation of color components (specially useful for quantification of histochemical staining) is also possible using Gabriel Landini’s Colour Deconvolution plugin. In addition, several other plugins related to color processing can be obtained from the ImageJ website. Conveying Color Information2 People see color with significant variations. Indeed, the popular phrase “One picture is worth ten thousand words” may not apply to certain color images, specially those that do not follow the basic principles of Color Universal Design. Citing Masataka Okabe and Kei Ito: Colorblind people can recognize a wide ranges of colors. But certain ranges of colors are hard to distinguish. The frequency of colorblindness is fairly high. One in 12 Caucasian (8%), one in 20 Asian (5%), and one in 25 African (4%) males are so-called ‘red–green’ colorblind. There are always colorblind people among the audience and readers. There should be more than ten colorblind in a room with 250 people (assuming 50% male and 50% female). [ . . . ] There is a good chance that the paper you submit may go to colorblind reviewers. Supposing that your paper will be reviewed by three white males (which is not unlikely considering the current population in science), the probability that at least one of them is colorblind is whopping 22%! One practical point defined by the Color Universal Design is the use of magenta in red–green overlays (see also [66]). Magenta is the equal mixture of red and blue. Colorblind people that 1See Wootton R, Springall DR, Polak JM. Image Analysis in Histology: Conventional and Confocal Microscopy. Cambridge University Press, 1995, ISBN 0521434823 2This section is partially extracted from Masataka Okabe and Kei Ito, Color Universal Design (CUD) — How to make figures and presentations that are friendly to Colorblind people, http://jfly.iam.u-tokyo.ac.jp/color/, accessed 2009.01.15 15 Last updated: 2012/10/02