Difference between revisions of "TubeTK/Design Discussions"

From KitwarePublic
Jump to navigationJump to search
Line 21: Line 21:
Examples:  
Examples:  


-------------------
MIDAS_ADD_TEST( ${PROJECT_NAME}MIDAS-Test4
MIDAS_ADD_TEST( ${PROJECT_NAME}MIDAS-Test4
             ${PROJ_EXE}
             ${PROJ_EXE}
             MIDAS{SyntheticTubePlusSurfaceImage_oriented.mha.md5}
             MIDAS{SyntheticTubePlusSurfaceImage_oriented.mha.md5}
             ${TEMP}/SyntheticTubePlusSurfaceEEDFilterResultImage_oriented.mha )
             ${TEMP}/SyntheticTubePlusSurfaceEEDFilterResultImage_oriented.mha )
-----------------
Test with a directory as input argument: DICOMSeries


-----------------
-----------------
Line 38: Line 43:
------------------
------------------


tubeEdgeEnhancingAnisotropicDiffusionMIDAS-Test1
Test that demonstrates fetch only: tubeEdgeEnhancingAnisotropicDiffusionMIDAS-Test1


-----------------
MIDAS_ADD_TEST( ${PROJECT_NAME}MIDAS-Test1
MIDAS_ADD_TEST( ${PROJECT_NAME}MIDAS-Test1
             ${PROJ_EXE}
             ${PROJ_EXE}
Line 46: Line 52:
               MIDAS_FETCH_ONLY{CroppedWholeLungCTScan.raw.md5}
               MIDAS_FETCH_ONLY{CroppedWholeLungCTScan.raw.md5}
               )
               )
----------------


= MIDAS macro =
= MIDAS macro =

Revision as of 19:59, 27 April 2012

Introduction

Tubetk has adopted MIDAS to store testing and algorithm validation dataset. For this purpose, we have created a dedicated tubetk MIDAS community

In the tubetk community, we have two main types of folders

  1. Input data folder: Where all input data for testing will be stored
  2. Baseline dataset folder: These are individual folders that will be used to store the baseline images for regression tests in tubetk

How to add a new tubetk test

  1. If the input data is not already in Midas,
    1. Upload the data to the "Data" folder
    2. Download the key file (*.md5) and store it in the MIDAS_KEYS directory at the top level of tubetk source code tree
  2. Add baseline dataset for the new test ( if your test is a regression testing )
    1. Upload the dataset to the appropriate CLI or module folder. If your test is part of a new "module" or CLI, create a new a folder in MIDAS and then upload the dataset
    2. Download the key file (*.md5) for the baseline image and store it in the MIDAS_KEYS directory.
  3. Modify cmake list files in the Testing directory to "add" unit tests to use the data from MIDAS.
    1. Use MIDAS_FETCH_ONLY macro, if you would like to fetch the dataset but not use it directly in the test
    2. Use MIDAS_DIRECTORY, if you would like to specify a whole directory as an input command line argument to your test

Examples:


MIDAS_ADD_TEST( ${PROJECT_NAME}MIDAS-Test4

           ${PROJ_EXE}
            MIDAS{SyntheticTubePlusSurfaceImage_oriented.mha.md5}
            ${TEMP}/SyntheticTubePlusSurfaceEEDFilterResultImage_oriented.mha )

Test with a directory as input argument: DICOMSeries


MIDAS_ADD_TEST( ${PROJECT_NAME}MIDAS-Test1

            ${PROJ_EXE}
              -i
              -r
              -a Test1 001
              -w
              -o ${TEMP}
              -b DSC_
              MIDAS_DIRECTORY{DICOMSeries} )

Test that demonstrates fetch only: tubeEdgeEnhancingAnisotropicDiffusionMIDAS-Test1


MIDAS_ADD_TEST( ${PROJECT_NAME}MIDAS-Test1

           ${PROJ_EXE}
              MIDAS{CroppedWholeLungCTScan.mhd.md5}
              ${TEMP}/itkAnisotropicEdgeEnhancingDiffusionImageFilterTest.mha
              MIDAS_FETCH_ONLY{CroppedWholeLungCTScan.raw.md5}
             )

MIDAS macro

The following are the main MIDAS cmake macro that will be used when adding a test

  1. MIDAS_ADD_TEST
  2. MIDAS_TGZ
  3. MIDAS_FETCH ONLY
  4. MIDAS_DIRECTORY