Overview

Research-grade images of muscle cross-sections can be very large (often more than 5000 x 5000 pixels). This can lead to problems with running out of computer memory.

Python_MyoVision tries to overcome this problem using block processing - in essence, analyzing bits of the image in sequence when possible.

Instructions

1. Open a command prompt
   • If you don’t know how
     • type cmd in the Search field of your Start Menu
     • or Google it to find instructions that work for you
2. Change the directory to the Python_code folder of your repository
   • If you installed Python_MyoVision in c:\users\your_username_here\GitHub\Python_Myovision
     you can type
     cd c:\users\your_username_here\GitHub\Python_Myovision\Python_code
     and press enter

3. In the command window, type
   python py_vision.py analyze_images “..\demos\example_B\configuration_data\configuration_data.xml”
   and press enter

4. Within a few seconds, you should see status updates in the command window

5. Once the program has finished, look in the results folder specified in the configuration file.
   • By default that should be
     c:\users\your_username_here\GitHub\Python_Myovision\demos\example_B\results
6. You should see the following files:
   • final_results.xlsx
   • clean_overlay.png
   • annotated_overlay.png
   • processing.zip

7. Here is the original image

   

8. Here is the clean_overlay.png file

   

9. Here is the annotated_overlay.png file

   

10. You can see that the overlays are larger than the original - in other words, they were padded.

11. What happened, was that MyoVision detected that the original image (1509 x 1408 pixels) was larger than 1000 x 1000 pixels (the threshold value is defined by the block_size in <image_to_label_parameters> in the configuration file). It therefore split the original image into blocks of 1000 x 1000, padding the right-hand side and the bottom-edge with black areas as necessary during the image processing.
    • You can see the individual blocks by opening up processing.zip in the results folder and looking inside the blocks sub-folder

    • Here’s how the top-right block was processed