Here is a demo application (in constant development) that allows analysis of the textures obtained regions of interest (ROI's) magnetic resonance imaging (MRI) of hams from Iberian pigs. This application uses segmentation techniques using active contours (SCA) for automatic detection of the contours of the ham muscles (biceps and semimembranosus) from which we obtain the ROI's above. Texture algorithms used are: Gray Level Coocurrence Matrix (GLMC), Neighboring Gray Level Dependence Matrix (NGLDS) and Gray Level Run Length Matrix (GLRLM).
Below is a gallery of images we use as an example of the application.
To download the images (which must be decompressed prior to the implementation of Ham), click here Download .
Here are the files and packages must be installed before running the application:
- JVM (Java Virtual Machine). (Download here).
- Packages .NET FRAMEWORK:
- .NET Framework Redistributable Package version 2.0 (x86). (Download here)
- .NET Framework Software Development Kit (SDK) version 2.0. (Download here)
Below are additional files that the user must save to disk in your system for the proper functioning of the application:
- File. RAR with images of iberian ham (which previously had to be unpacked). ( Download )
- File. RAR with template of the biceps (which previously had to be unpacked). ( Download )
Make sure you have the packages and files (mentioned above) installed on your system before you start the application.
Click on the link below to run the application Download and then select "Open with: Java (TM) Web Start Launcher". You may need to verify and check the box to trust the content as shown in the picture below.
The sample application allows the extraction of Iberian Ham automatic ROIs from the image sequence of a ham in particular and further analysis of these ROIs using different algorithms textures (GLMC, NGLDS and GLRLM). The study of these ROIs is focused on two specific muscles of the ham biceps and semimembranosus.
The steps for the consideration of these ROIs are:
- Load the image list of iberian ham.
- Setting certain parameters before and correcting the position of the ham if necessary.
- Select a subsequence of images relating to the muscle (biceps and semimembranosus).
- Selection of the initial template from which to perform the extraction process of ROIs and their adjustment to the muscle in question.
- Selection of ROIs extracted from discarding those that are not considered valid for analysis of textures. Besides saving disk are text files that store the results of texture analysis.
These are, broadly speaking, the purpose and stages making up the sample application.
Main Screen Application
As shown in Figure 1 shows the main screen the user will be located just run the application. In it, there are 2 menus, (File and Operations) an icon (equivalent to option Open Image File menu) and an area reserved for the display of images of the ham (Input Image). Is complemented with 5 buttons that allow users to advance the entire sequence of images of ham automatically or step by step.
Figure 1.- Main screen of the sample application of Iberian Ham.
- File Menu
- Menu Operations
- Active Contour Segmentation: This is the most important and which allows automatic extraction of ROIs and the subsequent texture analysis thereof.
- Otsu Thresholding: Option that, given an ROI implements the Otsu thresholding algorithm and see the histograms of both the original image Threshold image.
- Textures: This option is subdivided into more than one option is explained in more detail. Basically since the application can view the results of applying the various algorithms of texture on a ROI image previously uploaded by the user.
File Menu (up)
In this menu allows the opening and closing image of the application.
Figure 2.- File Menu.
Open Image (up)
With this option, a dialog box appears (Figure 3) where you can select the directory where the images are located for the ham. Once you have selected will appear in the main screen of the application the first image in the sequence. With the buttons at the bottom (see Figure 4), the user can view / advance the slideshow.
Figure 3.- Dialog to open a directory with the sequence of images of a ham.
At this point, note that the format of the images of the leg in question must be. PNG. If located in another different format (. BMP,. JPG, etc.). Will not be possible to load the image sequence of the selected directory correctly. As a possible solution, you can use a format converter available on the Web and then run the application.
Figure 4 .- buttons to view the image sequence of the Iberian Ham.
Menu Operations (up)
This menu has 3 options (see Figure 5).
Figure 5 .- Operations of the application menu
Active Contour Segmentation (up)
You can only access this option if you previously were loaded images of ham from the Open Image File menu. Otherwise, when you try to access it, that action will have no effect.
As shown in Figure 8, it is necessary to specify what will be considered muscle (biceps and semimembranosus, now the choice for the semimembranosus is temporarily disabled), what kind of ham it (fresh or aged) and Finally, check that the ham is in the position indicated by the application. The position of the ham should be the one shown in Figure 7:
Figure 7 .- The ham should be positioned so that the hoof is towards the north and the muscle is oriented to the west.
Figure 8. - Dialog for setting the configuration parameters.
All images must be of the same whether guidance on how to cure ham on fresh. This orientation is to place the claw ham on the top of the image (orientation north) and muscle facing the left side of stone image (orientation west). If otherwise, you must reset the guidance.
In addition, the sequence of images of ham must start by images of semimembranosus. If the image sequence begins with images of biceps will require investment order in order to sort the sequence to start with images of muscle cited (guidance reset button).
If the image sequence and ham is incorrectly oriented and the order of the sequence begins with images of semimembranosus is correct, will not make any necessary adjustment and operation will the election of the image for subsequent contour segmentation process assets (see figure 12).
In both cases, the user must click on the Reset button Guidance for carrying out the relocation operation and management of images of the ham (see Figure 9).
Figure 9. - The case where it is necessary to readjust the position of the images of the ham.
Therefore, the next window that will meet the user shall be as shown in Figure 10 after opening the directory where the images of ham from the icon (see Figure 10). It is offered the option of selecting the correct orientation and indicate what the original image sequence (taking into consideration that must begin with the images of semimembranosus).
Figure 10. - Initial window of the process of adjusting the position of the images of ham.
Once you're finished applying the necessary changes (as seen in the progress bar located at the bottom of the window of Figure 10), the user can only click on the button Next to continue the process of readjustment. In the next window that appears (Figure 11) the user can choose the orientation of the muscle (taking into account that the muscle should be placed to the left of the bone). When the progress bar located at the bottom indicates that it has processed all the images, they will terminate the process of readjustment of the images of ham).
Figure 11 .- Adjustment of the orientation of the muscle.
After implementing the realignment process, the following window (Figure 12). Here, we show the original images, center and end of the subset of images refers to the muscle by the application proposal. If the user does not agree with this subsequence may modify any of the 3 images (click on the Edit button) and move the buttons or backward in the sequence of images of ham to find the image you replaced (Replace button) to the previously chosen. In case of mistake, you can always cancel this action by the Cancel button.
Figure 12. - Choosing a subsequence of images, in this case relating to the biceps muscle.
If the user is satisfied with the subsequence chosen, simply press the mouse on the OK button to continue the process, showing a new window (Figure 13) which is about choosing the initial template from which we will try perform automatic detection of muscle in question and its limits within the central image.
First, it will be necessary to load a list of templates (files. SF) to be found in the directory where you unzipped the file the user. RAR concerned. This will be carried by pressing on the Open Folder button templates that display a dialog box in which the user will indicate where the directory containing the list of templates. Once done, you'll see the templates loaded on the left side of the window, choosing the template that best fits the muscle and defined in the scroll bars around the image (it is not necessary for users to plot the area excessive muscle with precision).
Figure 13. - Choice of the initial template and muscle definition on the central image of the subsequence.
Next, there will be only press the OK button to begin the process of automatic detection of the muscle on the subsequence chosen and wait a few seconds until the results are displayed (Figure 14). The user can preview the ROI's calculated traveling in a to one every one of the resulting images by selecting those in which the ROI is within the limits of the biceps. If so the user must dial the Present while, otherwise, you must dial the Deleted (Figure 14).
Figure 14.-Selection of ROIs calculated by the application.
Following the selection of the best ROIs, the user will click on the OK button will save text files the results of texture analysis on these ROIs that the user previously marked as valid (Present)
Finally, the results (both images and files of textures) can be seen in the folder SalidaCA created within the directory where the images of the ham.
Umbralización Otsu (UP)
If the user wants to access the menu option Otsu Thresholding Operations, you'll see the following window (Figure 15).
Figure 15. - Option menu Otsu Thresholding Operations.
The user only has to open the image of a ROI using the icon after which will display both images the original ROI Threshold its image. Appear at the bottom histograms of both images (Figure 16).
Figure 16. - Results of Otsu thresholding algorithm on a ROI.
Pressing the OK button, the user can return to the main menu of the application.
This option consists, in turn, a call option ROIs Textures that looks like that shown in Figure 17 and is intended to display the results in the application of texture analysis of an ROI chosen by the user.
Figure 17. - A window that is displayed when accessing the option Textures ROI.
As noted, there is an area reserved to display the ROI (analyzed image) located on the left side of the window and another area (right) reserved for observing the results of the 3 algorithms applied textures on the ROI that the user has selected previously by selecting Upload ROI.
With the image loaded, simply press the button Get Features to get the results (see Figure 18) in which, for each algorithm, show the values associated with each of the features discussed.
Figure 18. - Results of texture analysis of the selected ROI.
When leaving this window, the user returns to the main screen of the application.