MRIcro System requirements: Windows 95 or laterLinux x86, Solaris x86 Hardware requirements: 16 or 24-bit colour Current version: 1.40 build 1 Price: Free (seeinstallationsection to download) Author: Chris Rorden Features: Converts medical images to SPM friendly Analyze format. View Analyze format images (big or little endian). Create Analyze format headers (big or little endian). Create 3D regions of interest (with computed volume & intensity). Overlap multiple regions of interest. Rotate images to match SPM template images. Export images to BMP, JPEG, PNG or TIF format. Yoked images: linked viewing of multiple images (e.g. view same coordinates of PET and MRI scans).

MRIcro is a small program that runs on Apple OSX for viewing the NIfTI format of medical images popular with scientists (medical DICOM images can be converted to NIfTI using tools like my free dcm2nii). Therefore, MRIcron remains useful for users of older computers or when connecting to remote supercomputer clusters. These users may also consider fsleyes which can be set up to only require OpenGL 1.4 and indirect rendering. The latest releases of MRIcron only include the MRIcron viewer and the dcm2niix image converter.

Index

IntroductionReturn to Index

MRIcro allows Windows and Linux computers view medical images. It is a standalone program, but includes tools to complement SPM (software that allows neuroimagers to analyse MRI, fMRI and PET images). MRIcro allows efficient viewing and exporting of brain images. In addition, it allows neuropsychologists to identify regions of interest (ROIs, e.g. lesions). MRIcro can create Analyze format headers for exporting brain images to other platforms.

Users familiar with other Windows programs will find that this software is fairly straightforward to use. Resting the mouse cursor over a button will cause a text hint to appear over the button. As a last resort, I have included this brief manual that describes the basic features.

A tutorial with a step by step guide of how to use MRIcro with SPM is available.

InstallationReturn to Index

Note: MRIcro supports the Analyze file format. I strongly suggest using MRIcron. MRIcron supports the NIfTI format and has many new features. The only reason to use MRIcro instead of MRIcron is if you are familiar with this software or if you have a computer with limited memory (<256 Mb).

This section describes how to install MRIcro on a computer with the Windows operating system. There is a separate web page that describes the installation of the Linux version of MRIcro.

1. With your computer connected to the web, download the 5.9 Mb installer program, choose a mirror that is close to you:
   • US zip file (shift+click here) http://www.nitrc.org/frs/download.php/414/mrizip.zip.
2. Double click on the 'mrinstall' icon. The installer will give you the option to install the manuals, a sample MRI image and other files. By default, the files will be installed in 'C:Program FilesMRIcro'. Note: with Windows 2000/NT, only adminstrators can copy files into the 'Program Files' folder. If you are using 2000 or NT, either log in as an administrator or choose a different folder to install the files (e.g. 'C:usernamemricro').
3. To run MRIcro, click on the 'Start' menu, select 'Programs', point to the 'MRIcro' folder and click on the 'MRIcro' icon.

Loading images and the header information panel Return to Index

MRIcro can view various medical image formats, including the Analyze format used by SPM. Analyze format images have two components: the image file (*.img) that contains the raw image data and a header file (*.hdr) that describes the image dimensions, data format and comments. MRIcro's header information panel displays the header file's information and includes a series of buttons that allow you to open and view headers.

Figure 1. The header information panel.

The 'open Analyze format hdr/img pair' button (it looks like an open folder) allows you to view an Analyze or NIfTI format image. When you press this button, a dialog will appear that allows you to select a header file to open. MRIcro will then attempt to open an image file of with the same name (e.g. if you select a header 'C:17.hdr', MRIcro will attempt to open the image file 'C:17.img').

Dcm2nii

The 'show extended header information' button (it looks like a 'i'nformation icon) will open a window showing additional information about the currently open header (e.g. Scan comments, patient ID, etc).

Slice viewer panel Return to Index

The slice viewer panel allows you to select the appearance of the loaded image file. The slider selects which slice is displayed (alternately, you can select the slice in the edit box located directly to the right of the slider, or pressing Fn1/Fn2 to view the next successively lower/higher slice; if your mouse has a scroll wheel you can roll the wheel to change slice).

Figure 2. The slice viewer panel. When viewing axial, sagittal or coronal views, a slider appears that allows you to set which slice is displayed (illustrated). When projection views are displayed, you can set the X, Y and Z coordinates independently by adjusting the three edit boxes.

The brightness and control the contrast of the image by changing the values located next to the sun (brightness) and contrast images. These numbers describe the window center and window width of the display. The window centre refers to the image intensity that will be displayed as a medium-gray and the window width describes the range between bright white and full black. Clicking the 'auto contrast' button (showing a black arrow on a mostly white circle) sets 1% of the image to be maximum black and 1% to be maximum white. The autobalance works well for MRI scans, but often is not appropriate for CT scans, where the bones appear much brighter than the background and brain tissue. For CT scans, you can often get good settings by choosing one of the 'Contrast Preset [CT]' in the 'View' window. For example, the 'Bone' preset sets the window center to be 400 and the width to 2000, and is often useful for highlighting bone in CT scans.

Figure 3. You can adjust the contrast for a specific region by dragging over the area of the image while depressing the RIGHT mouse button. The brightness (window center) and contrast (window width) values are adjusted to scale the selected area. Note that this will not work if you have selected any of the region-of-interest drawing tools (press Fn10 to deselect drawing tools).

Figure 4. This histogram shows the image intensity profile for the image 'IconT1'. Histograms can also be created for user defined regions of interest (see the ROI section for more details). Note the Threshold Mn value is listed, this can be used to set the image intensity scale of a target image to match an SPM template (see the tutorial for more details).

The histogram button (located at the bottom-left of the panel) displays the voxel brightness distribution in the currently open image (see figure, above).

Six icons appear along the top of the Slice viewer panel that allow you to select which view of the image is displayed: axial, sagittal, coronal or projection (axial, sagittal and coronal simultaneously), free rotate, multislice and 3D rendering. If the image does not correspond with the selected button, the image was not saved in axial format. In the projection view, three edit boxes appear at the top of the slice viewer, allowing you to independently set the X, Y and Z coordinates you wish to view (alternately, click on the image to jump to the mouse coordinates).

If you are viewing more than one image simultaneously (having launched MRIcro more than once) you can 'yoke' the views, so that the same view is mirrored for each image (for this feature to work, make sure that the 'yoke' check box is selected). Yoking will try to match the slices based on the origin and size coordinates. This allows you to compare normalisation between an image and its template, as well as localising structures on a PET scan by comparing the PET scan with its coregistered T1 MRI scan.

To see a series of transverse or coronal slices simultaneously press the 'multislice' button (it has an icon showing three axial slices next to a sagittal slice). In order to function, an image must be loaded. The slices can be selected in the 'options' window, as described in other commands section.

Selecting the free rotate button causes a new window titled 'Free Rotate' to appear (see the figure below). This window allows you to select which oblique section you wish to view. You can independently set the yaw, roll, and pitch of the scan. Furthermore, you can select which slice you wish to view. The 'pivot' settings allow you to set the axis for the image rotations. You can also 'nudge' the image (center the image in the frame). Finally, you can set new image dimensions. Custom views can be saved as graphic pictures (using the 'file/save as picture' command), printed, or press the 'Save' button at the bottom of the Free Rotate window to create a new 8-bit Analyze format image described by your rotations. Note that the free rotate tools mix viewer centered and object centered coordinates, which can become somewhat disorienting. Also, when free rotating multiple Regions of interest with the 'interpolate' box checked, you may see a slight white halo around the regions.

You can generate 3D volume renderings by pressing the '3D' button (showing a sagittal rendering of the cortex). This feature is described in more detail on my volume rendering page.

Figure 5. The free rotation view allows you to create a custom view.

Mricron Per Mac

The Mirror button (it displays the letter 'LR' in either standard or mirrored text, depending on whether mirroring is selected) allows you to flip the image's left and right. If you select mirroring, a message will appear reminding you that it is best not to edit regions of interest while in the mirror mode (as the ROI will be stored in a different orientation than the original image).

A drop-down box selects the color lookup table (LUT) used. By default, MRIcro can show images in the 'black & white' or 'hot metal' color schemes. However, MRIcro can also read Osiris format LUTs, which allow you to view images in different color schemes. MRIcro ships with several LUTs from the Osiris distribution, and you can create your own custom LUTs. If MRIcro finds any *.LUT files in its directory, the drop-down box will list the file names. Select the color scheme you want from the drop down box.

A drop down box allows you to select how MRIcro displays images with voxels of unequal dimensions (e.g. an image slice where voxels are 2mm in theX dimension but 1mm in the Y dimension). First, images can be viewed in the 'square' mode, where all voxels are displayed as being square, irregardless of interslice dimensions. Second, you can view images 'stretch'ed, where the size of voxels is proportional to the interslice dimensions, using a nearest neighbor approximation (though projection and multislice views are NOT scaled). Finally, you can view images as 'smooth'ed: voxel size is proportional and uses a nearest neighbor approximation (this creates 24-bit images, projection views are shown correctly scaled).

Use the zoom factor box (located at the lower right of the slice viewer panel) to set the image scale, MRIcro allows you to view images in x1-x6 scale. Depending on the size and scale of your images, you may want to adjust the size of the MRIcro window (by dragging the lower right corner of the viewer).

Region of interest panel Return to Index

MRIcro allows you to draw three-dimensional regions of interest (ROIs). This is useful for illustrating regions of the brain that have sustained damage. In addition, the volume of the ROI is computed. Moreover, ROIs from different individuals can be overlapped (on brain images that have been normalised to the same template) allowing neuropsychologists to assess common areas of damage.

ROIs can be saved to disk for future reference. All ROIs are drawn on top of the MRI image, rather than directly on it. This means that the brain images can be viewed with or without corresponding ROIs.

MRIcro provides a number of tools for creating and viewing ROIs. These tools are displayed as a series of buttons in the 'Region of interest' panel. These buttons will be described sequentially from left to right.

Figure 6. The Region of Interest panel contains tools for marking lesion locations or specific areas. When you load a ROI, the volume is displayed in the bottom left corner, here 2.6 cubic centimeters.

The 'open roi' button (it looks like an open folder with a ROI) allows you to select ROI[s] to open. If you want to open multiple ROIs simultaneously, depress the control key as you select the ROI names.

The ROI information button (it displays an 'i'nformation icon) displays the mean image intensity of a region of interest as well as generating an image intensity histogram (similar to the histogram depicted in the slice viewer section).

The 'save ROI' button (it has an icon of a disk with a ROI superimposed on it) allows you to save a ROI that you have created to disk.

Use the 'delete ROI on this slice' button (it shows a pencil erasing information) to remove the ROI only from the slice you are currently viewing. This button is useful if you make a mistake outlining or filling a ROI that you are drawing.

In order to remove all current ROI[s] from memory, click the 'delete entire ROI' button (it looks like a waste basket). If you have opened multiple ROIs, you will need to select this before creating a new ROI (you can only write to one ROI at a time).

The 'save ROI' button (it has an icon of a disk with a ROI superimposed on it) allows you to save a ROI that you have created to disk.

The 'Delete entire ROI' button (it has an icon of a wastebasket with a ROI superimposed on it) allows you to close a ROI.

Next, there is a small drop down box that defines the colour that single ROIs will be drawn in (when viewing multiple ROIs simultaneously a rainbow colour set is automatically used). The choices are Red, Green, Blue, White or Black. This allows you to choose a salient colour. You can set the ROI to black or white before printing the image to a black and white printer.

After loading multiple ROIs, you can select the 'ROI density colorbar' item from the 'ROI' menu to draw graphic of the overlap between ROIs. The colours indicate the number of overlapping ROIs. The leftmost (dark violet) colour indicates the index for a single ROI, while the rightmost (bright red) colour shows the index for all the ROIs overlapping. This feature is demonstrated in the figure below:

Figure 7. These figures show MRI multislices with four overlapping ROIs. The colour bar in the upper right corners indicates the ROI density (dark violet for one ROI, bright red for four overlapping ROIs). These figures also demonstrates the option window's 'show right hemisphere' function -where successive slices overlap each other. Multislices can be generated either for transverse slices (top panel) or coronal slices (bottom panel).

Select one of the pens for creating ROI outlines. There are two pens: the 'closed' pen (shortcut key Fn6) automatically closes any outline you have drawn. The 'open' pen (shortcut key Fn7) does not automatically close the outline. The 'fill button' (paint can icon, shortcut Fn8) can be used to fill a ROI that has been outlined by the pen (alternative: right click with the pen selected). Before beginning to create a new ROI you should load the image, clear any previous ROIs (using the 'save ROI' or 'delete entire ROI' buttons described above), and set the view to transverse, coronal or sagittal slices (you can not draw ROIs on the free rotate, projection and multiple views). Holding the shift key down while using a pen or fill will erase ROIs in the designated region.

In practice, ROIs can be drawn rapidly by selecting the closed pen (shortcut key Fn6) and then using Fn1 and Fn2 to move up and down to the desired slices. Using the pen with the shift key down is useful for trimming unwanted edges from a ROI. Use the left mouse button to outline the ROI, and then move to the centre of the ROI and right-click to fill the region. Once the ROI is drawn, use the 'Save ROI' button to store the ROI. Note that the ROI volume is listed in the Region of interest panel (computed in either cc or voxels, updated when you change slices or save the ROI).

The wizard's cap icon allows you to temporarily hide a region of interest (shortcut key Fn9). By rapidly hiding and showing the ROI, you can see whether the ROI you have drawn correctly maps the region you are interested in.

When the mouse moves over an image, two useful bits of information are displayed immediately below the region of interest panel. Text on the far left side shows the the mouse position in Talaraich space, followed by the image intensity immediately beneath the mouse.

Rotation and clipping panel Return to Index Enfield enforcer serial numbers.

The file menu contains a command labelled 'Save as..'. When this command is selected, a floating panel appears in the lower left corner of program's window. The rotation and clipping panel can be used to prepare image files for normalisation (coregistering an image to a standard template) with SPM.

Figure 8. The rotation and clipping panel.

The clip top and clip bottom fields allow you to choose the number of slices that you want to shave off of the top or bottom of a scan (for example, trimming scans that include too much spinal cord to avoid problems with SPM normalisations). The 'Format' list box allows you to convert coronal or sagittal Analyze format images to axial format (this is the natural preference for SPM). To determine the original format of your image, click on the 'axial [flat] view' button in the slice viewer panel, this will display the scan in its native format (e.g. if the flat format displays a sagittal image, the file is in sagittal format). The 'flip left/right' checkbox mirrors the image.

You can also set the 'data' type for the saved image. Usually it is wise to save the data in the same format as the source image. Down-sampling an image (e.g. saving a 16-bit integer file as an 8-bit data file) will save disk space (potentially at the expense of image intensity quality). Up-sampling an image (e.g. saving a 16-bit file as a 32-bit file) can be useful for when other programs require data of a specific type. If you have adjusted the contrast/brightness settings in the slice viewer panel, you will be asked if you wish to clip the brightest and/or darkest voxels, allowing you to customise the contrast of the output.

Micron Machine Works

Figure 9. MRIcro automatically displays the amount of clipping selected, in this example the top five and bottom ten slices are about to be clipped. This figure also shows the 'Hot Metal' color lookup table.

Before saving an image that you wish to rotate, you can check your settings by pressing the 'preview' button. The preview will show you two slices of how the image will appear after being rotated. If your settings are correct, the preview should show two transverse slices (with the left slice being more ventral than the right slice). Make sure to check that the left/right mirroring is correct.

When you have selected the desired clipping and check box options, press one of the three file save buttons that are located at the bottom of the window. The buttons 'Save [Sun]' and 'Save [Intel]' will save the files as Analyze format images. SPMwin, SPM99, SPM2 and MRIcro can all read either big or little-endian Analyze files, while SPM96 requires the images to be in the same format as the machine used. For more information, see the technical section.If the original image is a multi-volume image (i.e. the img file contains multiple MRI scans, which are all the same dimension), MRIcro will rotate each volume and save it as a separate Analyze format header/image (this is useful, as SPM can not read multiple volume files). The 'Save Dicom' button will save the image as a DICOM format image (as described in the DICOM section of this manual). Note that the DICOM format image will either be in 8-bit integer (if that is the data type of the source image) or 16-bit integer format irregardless of the data type seleted.

Overlays [Statistical Maps] Return to Index

The 'Overlay' menu allows you to select an image which is superimposed on top of another image. This is useful for displaying functional statistical maps (generated by SPM from PET, fMRI or SPECT data). Overlays can also be used to check the coregistration of two images. To display an overlay, you should first choose the primary image you want to use (e.g. use the 'Open' command from the file menu to select your anatomical image). Next, use the 'Load overlay' command in the Overlay menu to select the image you wish to superimpose. The overlay does NOT need to be the resliced to the same dimensions as the anatomical image: MRIcro will correctly reslice the data (as long as the image dimensions and origin are correctly specified for each image). The overlay menu also allows you to select the color of the overlay and whether the overlay is shown as opaque or transparent. The images below show overlays of functional data (left) and the use of the overlay function to check normalization. For more details, visit my overlay and volume rendering pages.

Left: functional results can be overlayed. Above: you can use overlays to check alignment after normalization.

Selecting coordinates Return to Index

The 'Select MNI/Talairach Coordinates' command in the 'View' menu allows you to specify which region of the brain you want to view. With the slice viewer panel, you choose which slice in the image you wish to view. In contrast, the coordinates command allows you to type in the steotactic coordinates used by neuroimagers. This command only makes sense if your image has been normalized (stretched, rotated and centered to match a standard neurological frame of reference). There are two popular frames of reference for brain imaging: MNI space (used by SPM) and Talairach space (used in the atlases of Talairach and Torneaux). A check box in the coordinates window allows you to set which frame of reference you wish to use (the Talairach coordinates are estimated used Matthew Brett's handy mni2tal and tal2mni routines). These two methods are illustrated below. When looking at an image from the Talairach and Tournoux atlas (left) you should make sure the 'tal2mni' box is checked (shown in red). On the other hand, when looking at SPM results, make sure this feature is switched off.

Other commands Return to Index

The 'File' menu contains the command 'save as picture', which allows you to save the currently displayed image as a 2D graphic image. Supported formats include the popular BMP, JPEG, PNG and TIF formats. The File menu's 'Print' item allows you to print the currently displayed image.

The ROI menu contains commands that allow you to export ROIs as big-endian (Sun) or little-endian (Intel) Analyze format images. These commands save the currently displayed region of interest as an 8-bit Analyze format image, which can be used in SPM as a mask (allowing sensible normalization of brain images that have large lesions). Note that the only values on the mask image are 0 (unmasked) and 100 (masked), and that the mask's image intensity scale value is set to 0.01 (so from SPM's point of view, the Mask has voxels of the values 0 and 1).

The ROI menu's 'Export Image as ROI' converts an Analyze format image in MRIcro's custom ROI format. Is ttw worth it. Note that the ROI format is binary -the ROI does not store intensity information. Load the image you wish to convert before selecting this command. When you select this item, a window appears that allows you to set the intensity ROI's threshold. As you adjust the threshold, the image will will preview the portions of the image will be included in the ROI. When you are happy with the selection, press the 'Save as ROI' button to create the new ROI.

In addition, you can transfer a region of interest between two images that have different dimensions. One example of the utility of this command is identifying cortical regions on a PET scan. Using the 'File/Transfer ROI' command you can create a ROI on the individuals T1 anatomical MRI scan and then copy the ROI to the corresponding PET scan(s).

The 'Etc' menu's command 'Options' allows you to adjust a number of MRIcro's settings. First of all, you can select which slices will appear when the 'multislice' button is pressed. There are twelve data fields, allowing up to twelve slices to be viewed simultaneously. Fields that are set to zero will not be presented, allowing you to show fewer than twelve slices. You can also select whether the multislice button will generate coronal or transverse images. The 'defaults' button at the bottom of the window will select a common set of slices useful for viewing images that have been normalised to SPM's T1 template. The 'include sagittal' check box selects whether or not a sagittal view will appear when the 'multislice' button is pressed. The option window also includes a number of additional check boxes that describe how the multislice view will be presented. A few of the less intuitive options are shown in the figures below. Once you have selected the desired options, you can press 'OK' to save your selection, or 'Cancel' to ignore your changes. A number of these options are described in the tutorial, and some are illustrated in the figures below.

Figure 10. If the option window's 'Cumulative ROI boxed [CT]' check is selected, the sagittal view will extrapolate regions from the selected slices (left panel). In this case, ROIs only need to be drawn on the slices that will be viewed. Otherwise, the sagittal view will show ROIs for each slice independently.

Figure 11. Right clicking the multislice button allows you to superimpose hotspots onto an image. The hotspots are defined in a text file, which includes the X Y Z coordinates and the colour/shape of the figure to be superimposed. This can be useful for comparing a neurological patient's lesion to hotspots from fMRI/PET studies. In this example, a number of hotspots have been drawn on top of a frontal lobe ROI. For more details, see the MRIcro FAQ. This picture also illustrates the 'translusent ROI' feature.

Figure 12. The left panel shows axial slices of two ROIs -one is primarily subcortical and the other is primariily cortical. When the option window's 'Cumulative only true overlap' is selected, the sagittal view will show ROI densities based on the maximum true density of ROIs (middle panel). If this option is not selected, the sagittal view ROI will sum all ROIs, independent of their location on the X axis (right panel). Note that this will sum ROIs irrespective of whether they are both cortical/subcortical or even in the same hemisphere.

Viewing medical images Return to Index

In addition to Analyze format files, MRIcro can also view a wide range of standard medical image formats. A full list of these formats is listed in the Technical Details section of this manual. MRIcro will automatically detect the format of your image regardless of whether it is Analyze, DICOM, etc. format. The easiest way to open a medical image is to simplay drag and drop its icon onto MRIcro. Alternatively, select the 'Open foreign..' command from the 'Import' menu and select your image. One exception to this rule is ECAT images: MRIcro will not detect ECAT6 files and can only automatically view some ECAT7 files. For ECAT files, one should first convert the file to Analyze format (as described in the next section).

Converting medical images to Analyze format Return to Index

SPM, Analyze, mri3dX and MRIcro all use the Analyze format as their native image format. However, many scanners save images to DICOM or proprietary file formats. The technical section describes the formats that MRIcro and other programs can convert to Analyze format.

When you convert medical images to Analyze format, you can select a series of 2D images that will be stacked and saved as a single unified 3D Analyze format file. Selecting 'Convert foreign to Analyze' from the 'Import' menu will create a new window that allows you to describe your images. My tutorial describes this process in detail. ECAT images can be converted selecting the 'Convert ECAT to Analyze' command from the import menu. ECAT images can be saved as the raw data or as scaled data (taking into account the calibration and scaling factors) - the conversion format can be selected in by choosing 'Etc/Options' and setting the 'ECAT convert' values. Finally, you can convert SPMwin headers to SPM headers using the 'Import/SPMwin VHD to Analyze' command.

MRIcro can also convert Analyze format images to the DICOM format. The command 'Save as..' allows you to save an convert an image to DICOM. The *.img file that is created when you press 'Save DICOM' will be in the DICOM format. The image may not strictly obey the guidelines of the DICOM standard. However, the images should be compatible with most DICOM image viewers. The DICOM file will be either 8-bit (if the source file uses 8-bits per voxel) or 16-bit (if the source file uses 16-bits or greater per voxel). The DICOM file that is created is 'anonymized' -and will contain no private details from the Analyze header (e.g. patient name, scan date, etc).

Uninstalling Return to Index

To uninstall MRIcro for Windows, select the 'Add/Remove Programs' control panel (to access the control panel, select StartMenu/Settings/ControlPanel). Simply select 'MRIcro [remove only]' from the 'Add/Remove Programs' control panel and press 'Add/Remove'.

In order to remove MRIcro for Linux from your computer, delete the folder '/usr/local/bin/mricro'. You will need root permissions to do this.

MRIcro normally remembers the recent files you used, your preferred views, and other information. If you want to reset MRIcro to the default settings, run the program and select the 'Uninstall' command from the 'Etc' menu. Then quit MRIcro and relaunching the program.

Technical Details Return to Index

• Referencing. Scientific publications can refer to Rorden, C., Brett, M. (2000). Stereotaxic display of brain lesions. Behavioural Neurology, 12, 191-200.
• License. The compiled MRIcro application is provided as freeware under the BSD license. Portions of the source code are available from the ezDICOM project at sourceforge.
• Chris Rorden's MRIcro, copyright 1999-2005, all rights reserved.
• Redistribution and use in binary forms, with or without modification, are permitted provided inclusion of the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
• Neither the name of the copyright owner nor the name of this project (MRIcro) may be used to endorse or promote products derived from this software without specific prior written permission.
• THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER 'AS IS' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
• Trouble shooting.
  • If MRIcro initially shows an image correctly but does not update the display when the slice is changed, turn off the 'graphics acceleration' option found in the 'Etc' menu's 'option' window (graphics acceleration greatly speeds image display, but is not compatible with all graphics cards).
  • When loading multiple regions of interest (in order to show mutual regions of overlap) some Windows NT users may find that not all of the ROIs are loaded. Some versions of NT will only report the first 255 characters of file names. A quick workaround is to shorten the length of the file names (e.g. instead of 'roiforfile1.roi', name the file '1.roi'). To check the number of ROIs that are loaded, choose 'ROI density colorbar' ('ROI' menu), the number of colours displayed shows the number of ROIs loaded.
  • In addition to this manual, the MRIcro FAQ, volume rendering page and the tutorial resolve most questions. For interactive support, you can also visit the JISCmail MRIcro List. Please realise that this software is downloaded more than 35,000 times per year. I am happy to help, but it is easy to feel overwhelmed by questions. You can get help from the MRIcro users list, either using the JISCmail web site.
• Foreign language manuals:Tsutomu Soma has kindly created a Japanese translation of this manual.
• Usage. MRIcro is designed for research use, rather than clinical use. The author accepts no liability for the use of this software. Please use this tool sensibly.
• Memory. MRIcro requires a large amount of Random Access Memory (RAM) in order to view Analyze format images. Images with a data type of 8-bit integer will require one byte per image voxel. Images of 16-bit integer data type require 3 bytes per voxel (one byte for the screen buffer and two bytes for the data). Images with any other data type (e.g. 32-bit integer) require five bytes per voxel (one byte for the screen image buffer and four bytes for a real-number array that allows contrast optimization). Editing or viewing ROIs will require an additional byte per voxel. For example, viewing the SPM template (91x109x91 voxels, 8 bit integer data type) will require .86 Mb of RAM (1.72Mb when viewing or editing a ROI). The table below lists the bytes per voxel (bpv) for images of different data types. In addition, the table lists the RAM required to open a 256x256x124 voxel image of each data type (in megabytes). The 'Etc' menu's 'About' command will display the RAM currently being used.

  Data Type	RAM when loading	RAM when viewing	RAM with ROI
  8-bit	1 bpv (7.75 Mb)	1 bpv (7.75 Mb)	2 bpv (15.5 Mb)
  16-bit	3 bpv (23.25 Mb)	3 bpv (23.25 Mb)	4 bpv (31 Mb)
  Other	32-bit: 5 bpv (38.75 Mb) 64-bit: 12 bpv (93.00 Mb)	5 bpv (38.75 Mb)	6 bpv (46.5 Mb)

• Y2K. MRIcro is year 2000 compliant.
• Monitor depth. MRIcro uses a palette of 256 colours for drawing brain images. Because Windows reserves a number of colours, your display should be set to a 16-bit (high colour) 24-bit or 32-bit (true colour) colour palette. If your computer can not support more than 256 colours, you should switch off the 'graphics acceleration' check box in the options window (select 'options' from the Etc menu). To find out the current setting, open the 'display ' control panel and select the 'settings' tab. Using a colour palette of less than 16 bits (e.g. 256 colours) will limit the levels of gray that MRIcro can display. MRIcro renders images in 255 grays (reserving one colour for the region of interest). Fewer gray levels are used when viewing multiple regions of interest.
• Big-endian and little-endian data. Different computers use different formats for ordering numeric values that require more than one byte. With little-endian processors such as Intel and compatibles the bytes with a higher address are most significant. In contrast, big-endian Sun, Silicon Graphics, Motorola and HP processors use the opposite byte order. Alpha and PowerPC processors can use either depending on a compiler switch. This means that Analyze format headers and some Analyze format data files (those with more than one byte per voxel, e.g. data type greater than 8-bit integer) may be different between different computers. MRIcro, SPMwin, SPM99, and SPM2 automatically detect the endian-ness of the headers files. When using MRIcro to create header files, you should save them to the same format as the image file.
• Region of Interest. ROIs are saved as 'filename.ROI'. This is a proprietary binary format that uses run-length encoding (a simple compression algorithm) to store a relatively compact image of the region of interest.
• Credits. MRIcro was programmed using Delphi (object-oriented Pascal). The sophisticated sliders and real-number edit boxes are components from the free RXlib component collection. SPM guru Matthew Brett suggested and helped test a number of useful enhancements (Matthew has written a number of great web pages that describe how SPM works, which are available at www.mrc- cbu.cam.ac.uk/Imaging/). Tom Womack added the SSE support for faster viewing of 32 and 64-bit images and improved the rendering techniques. Chris Rorden wrote MRIcro while working on a project grant from The Wellcome Trust.
• Header information used by SPM. During loading and saving of header files, MRIcro will show an alert message if it detects an error that might cause SPM difficulty. SPM requires the header to accurately describe the image file: the image dimensions, data type and byte offset all need to be correct. In addition, SPM will use the origin information to select the voxel deemed to be the 'centre' of the image, which in normalised images is the centre of the anterior commissure. If the origin is set to 0,0,0, then SPM will assume that the centre has not been set, which usually works pretty well. SPM also uses the 'scale' factor (the last item displayed in the extended header information window) to compute the true image intensity. For more information on these topics, search the SPM archives. To learn more about the Analyze file format, view the Mayo Clinic's extensive guide (www.mayo.edu/bir) or read Medical Image Format FAQ . For further details on SPM's specific and nonstandard usage of the Analyze format, visit the SPM home or www.mrc-cbu.cam.ac.uk/Imaging/.
• Supported 2D picture formats. MRIcro can open, save and print BMP, JPEG, PNG and TIF images. Free viewers such as IrfanView can batch convert images to other formats such as GIF. JPG images are very compact, but some of the image quality is lost. BMP and TIF retain the image quality, but the files are large. PNG (portable network graphics) format is both lossless and compressed, combining the best features of TIF and JPG. A number of PNG viewers are available to support virtually every major operating system, see www.libpng.org/pub/png/pngapvw.html for details. My web graphics page describes the relative merits of these formats.
• Converting images to Analyze format.Note: the programs listed below support the Analyze file format. I strongly suggest converting images to the NifTI file format - this extension of the Analyze format stores inportant information. For NIfTI converters please see my dcm2nii webpage:
  

  Viewer	Formats, Notes
  MRIcro [Windows, Linux]	DICOM (uncompressed and compressed), AFNI (.head), 8bit BMP (.bmp), BioRad PIC, CTI ECAT6/7, Elscint, Freesurfer, GE (LX, Genesis, 4.X, 5.X, compressed), Interfile, NEMA, PGM, PPM, Picker CT, Philips (.PAR/.REC), PovRay density (.DF3), Siemens (Magnetom Vision, Somatom, Somatom Plus), Space (.vol), SPMwin (.vhd), 8bit TIFF (uncompressed), Vista (.v), VFF (.vff), VoxBo (.cub), Zeiss LSM 510, raw see my conversion tutorial.
  Bru2Anz [Windows]	Bruker Paravision
  ImageJ with Analyze plugin installed [Java: Windows, Mac, Unix]	DICOM (uncompressed), BMP, TIFF, JPEG, GIF, Raw Choose File/Open to open your images (if you want to stack multiple images, place them alone in a folder and use File/Import/AllAsStack). If your image is 24-bit RGB (e.g. most JPEGs), you need to convert the image to grayscale. First, you may want to optimize the contrast using the functions in the Process menu. Then use the Image/Type/8bitGreyscale function. Select Plugins/AnalyzeWriter to convert your images to Analyze format. If the color-intensity of the new Analyze image is inverted, use Plugins/Inverter before using Plugins/AnalyzeWriter.
  XMedCon [Windows, Unix]	DICOM (uncompressed), GIF, ECAT6, Interfile, NEMA
  Mosaic to Analyze [Matlab: Windows, Unix, Mac]	Siemens Magnetom Vision
  Syngo Converter [Matlab: Windows, Unix, Mac]	Siemens DICOM (uncompressed)
  exp2ana3d.m (for 256x256 images), exp2ana3dt.m (for 64x64 EPI)[Matlab+SPM: Windows, Unix]	Giuseppe Pagnoni's Matlab script for converting Philips .PAR/.REC export files.
  ge2spm [Matlab: Windows, Unix]	GE LX, GE 4.X, GE 5.X format
  stim2analyze volume_mri_convert [Unix]	GE Genesis
  Dicom_Toolkit [IDL: Windows, Unix, Mac]	DICOM
  VisionToSPM [Matlab: Windows, Unix, Mac]	Siemens Vision
  ge2spm [Unix]	GE Genesis
  ana2mnc [ Perl script: Windows, Unix, Macintosh]	MINC, Bruker Paravision, GE 4.x, GE 5.x, DICOM (uncompressed), Scanditronix, ECAT, Siemens Magnetom, Phillips
  MINC to Analyze [Matlab: Windows, Unix, Mac]	MINC
  Image Converter [Windows]	Siemens System 7, Shimadzu HeadTome IV, Hamamatsu Photonics SHR2000, GEMS 2048-15B
  SPM2 [Matlab: Windows, Unix, Mac]	DICOM [Requires Matlab 6.0 or later, only some DICOM variations, select 'DICOM' from the 'Toolboxes' menu]
  IMGCON [Windows, Unix]	BMP, GIF, PNG, TIF, JPEG, PGM, MIT, EPS, raw
  MRIConvert [Windows]	DICOM (uncompressed)
  FormatConvert [Unix]	Pittsburgh 1.0, Afni, GE (I*.nnn files, E*.MR files), BrainVoyager (VMR, VTC, STC, VMP)
  scanSTAT [Macintosh]	GE (.MR), ANMR (.irp/.img), MGH

• Unsupported 3D formats. Please see my dedicated page describing importing unsupported image formats using MRIcro. If you a working with raw data, or a medical format which is not listed in the table above, you can try this desperate measure to convert the image to Analyze format:
  • First, give the raw data file name the extension '.img'.
  • Next, create an Analyze format header using MRIcro's header information panel. The bare minimum is to specify the number of voxels in each dimension (the X, Y, and Z dimensions), the number of bytes per voxel (the type drop down menu, e.g. 32-bit real type specifies 4 bytes [a 'byte' is 8 bits] per voxel), the image offset in bytes (usually zero). If you are unsure of your image dimensions, read Dave Clunie's Quick and Dirty Tricks for viewing medical images.
  • If your raw data file contains multiple MRI scans in the same file, you need to specifiy the number of 'Volumes'. The size of the raw data in bytes should be at least X*Y*Z*N*V+O. Where X,Y and Z are the image dimensions, N is the number of bytes per voxel, V is the number of volumes and O is the offset (if the offset is set to a negative value, each indvidual volume will have its own offset, so the image size will be X*Y*Z*N*V*-1*O).
  • If you are dealing with raw data, set the 'offset' value in the header information panel to zero. If your file is in an unknown filetype instead of raw data, you can usually assume that the file contains a header at the start of the image. For example, if your image is a 256x256x1 voxel scan, with 2 bytes per voxel, the raw image data should be 131072 bytes long - if your file is 131224 bytes long, set the header offset to 152 (131224 - 131072).
  • Save this header by pressing the floppy disk icon in the header information panel. Give the header the same name as your image, but using the extension '.hdr' for your header.
  • Now open your image by pressing on the 'open image with displayed header' button (it is in the header information panel, and shows a folder with the letters 'img').
  • Note that most image formats store their data starting from the upper left corner, storing data in the same way we read English (left to right, top to bottom). However, the Analyze format stores the data starting with the bottom right (going right to left, bottom to top). Therefore, you may need to use MRIcro's 'Save as..' function to find the correct orientation.
• History. Version 0.99 was the first release. Version 1.0 includes projection views. Version 1.01 can export regions of interest as 8-bit Analyze format images. Version 1.02 allows you to create regions of interest on images with up to 1024x1023 voxels in the XxY dimension. Version 1.1 added the ability to show proportionally scaled images, reports the mean intensity of ROIs. Version 1.11 previews image rotations, and allows you to navigate projection views by clicking on the desired location. Version 1.12 can display slice Z-values in multislice views. Version 1.13 includes free rotation (oblique slices), data type conversion, and the ability to transfer regions of interest to images with different coordinates. Version 1.14 can convert Analyze images into MRIcro's ROI format. Version 1.15 implements the ability to yoke projection views. Version 1.16 introduces intensity histograms, the 'hot metal' colours, the image intensity colourbar and image intensity autobalancing contrast. Release 1.17 can write to universal naming convention servers, adds the ability to turn off graphics acceleration, adds additional features to the multislice options, and uses less RAM for 16-bit images. Version 1.18 added XBars and watermarks to all views, allows adjustment of fonts, allows customized ROI colours, optionally creates translucent ROIs on multislices. In version 1.19, the multislice view can show overlapping brain slices to reveal only the left or right hemisphere, smoothed ROIs can be saved in Analyze format. Version 1.20 introduces coronal multislices, the ability to draw ROIs on sagittal and coronal as well as transverse slices, and the ability to read and extract multiple volume Analyze format files. Version 1.21 can read and write DICOM format images. MRIcro 1.22 can read Genesis format files, the 'File/Export ROI as Analyze image' will now save multiple overlapping ROIs, right-clicking the ROI information button will give verbose information about ROI intensities and Talaraich coordinates, a DICOM-style precise contrast control has been added, floating point Not-A-Number data is preserved, and a routine to convert a series of 2D Genesis or DICOM format scans to a single 3D Analyze format image. Version 1.23 adds support for ECAT and Interfile formats, all standard views can be yoked, ROIs can be intensity filtered (e.g. so that only white or gray matter is included in the ROI). Version 1.24 added the ability to open Picker CT and Siemens Magnetom Vision format, the ability to see the same slice from multiple images (useful for looking at motion and activity in fMRI scans) and supports Osiris format color schemes. Version 1.25 can read and convert SPMwin vhd headers, supports raw binary colour schemes (768 byte color tables with 256 bytes sequentially for red, green, and blue), and has improved support for 32-bit images. Release 1.26 adds a magnifier tool and 'smooth' image scaling option. ROI tools have been added (union, intersection, masking and subtraction) as described in the tutorial. Version 1.27 allows ECAT image data to be scaled using the 'scaling factor' and 'calibration' values. When drawing ROIs, there is now an 'undo' function. Individual slices and regions can have their ROIs segmented based on image intensity (as described in the tutorial). Release 1.28 can convert JPEG-compressed (lossless, lossful and XA) DICOM images to Analyze format, can invert colors, and can save images as JPEG format. Version 1.29 views/converts Elscint, Par/Rec and Somatom images. Version 1.30: New installer/uninstaller, Linux version, SSE support. Release 1.31 adds PNG support and can save a 2D bitmap graphic for each slice of a 3D image, hotspots can be added to coronal or axial slices. Version 1.32 can display volume renderings, 'Select MNI/Talairach Coordinates' command in 'View' menu and supports image overlays (e.g. statistical maps of functional data). Volume 1.33 adds improved quality surface rendering, improved statistical overlays and includes support for VoxBo CUB1 format images. Version 1.34 includes a better sample MRI scan, improved BET, and tweaks to the rendering and foreign image conversion. Version 1.35: includes anatomical templates , improved importing of unsupported images , can convert anisotropic images to be isotropic, opens gzipped Analyze .img files. Version 1.36 includes image despeckling and an arcball tool for setting the viewpoint of rendered images; improves Siemens image conversion [thanks to Siemens] and speeds up rendering [thanks to Tom Womack]; improves the quality of surface renderings; opens confocal images (BioRad and Zeiss TIFF), cluster size can be set for overlays, basic 3D ROI drawing, holding the CTRL key down while drawing ROIs implements a thicker brush, in projection views, SHIFT+F1/SHIFT+F2 moves up/down coronal slices, CTL+F1/CTL+F2 moves up/down sagittal slices. Version 1.37 includes a 'Conversion Wizard' ['Import' menu] that can convert a series of DICOM images to Analyze format in a single, automated step. Version 1.38 improves contrast and brigthness settings, using the medical imaging standard 'window centre' and 'window width' values. Version 1.39 introduces support for NIfTI (.nii, .nii.gz and .hdr/.img files) and support for FSL 3.2. Version 1.40 improves image conversion for newer scanners.
• Additional color schemes: MRIcro can display images in the 'black and white' as well as 'hot metal' color schemes. Additional Osiris-format color lookup tables (LUTs) can also be displayed. The Osiris format is a text format, so you can create your own custom LUTs. In addition, MRIcro can support 768-byte binary data LUTs (NIH Image, XMedcon format). MRIcro expects the *.LUT look-up tables to be placed in the same directory as the MRIcro.exe program. Additional color schemes are available by downloading the free XMedcon 'extra stuff' software.
• Platforms, limitations, alternatives: MRIcro supports the Windows and Linux platforms. Here is a short list of some freewareAnalyze format image viewers that are currently available for the Windows PC (for a list of Linux software, see the MRIcro for Linux web page. A more extensive list of DICOM viewers for Unix, Macintosh and PCs is available at my DICOM page):

  Viewer	Platforms	Formats [Notes]
  SPM	Unix/Windows NT	Analyze [requires Matlab]
  Slice Overlay	Unix/Windows NT	Analyze [requires Matlab]
  SPMwin	Windows	Analyze
  ACTIV 2000	Windows	Analyze/DICOM/GE/GIS/PAR/Siemens
  Medal	Windows	Analyze/DICOM
  etdips	Windows	Analyze/DICOM/TIFF
  Spamalize	Unix/Windows/Macintosh	Analyze/GE/TIFF [requires IDL]
  AMIDE	Unix/Windows	Analyze/DICOM/ECAT6/Interfile
  XMedCon	Unix/Windows	Analyze/DICOM/ECAT6/Interfile
  ezDICOM	Windows	Analyze/ECAT/Interfile/Siemens/Picker/GE/DICOM/VoxBo
  ImageJ w. Analyze plugin	Unix/Windows/Macintosh	Analyze/DICOM
  Space	Windows	Analyze/VOL [beta release]

• Linux. A native Linux-native version of MRIcro is available (MRIcro for Windows can also be used in Linux by using WINE). Alternatively, try the new MRIcron for Linux.
• Macintosh. There are a couple techniques for running the Windows version of MRIcro on a Macintosh. Alternatively, try the new MRIcron for Macintosh.
• Converting ROIs. MRIcro stores Regions of Interest using a simple compression technique.I have written a web page that describes tools for converting between the ROI and Analyze or NIfTI formats.
• Updates. The latest versions of the MRIcro software and help manual are available from www.mricro.com .