sliceOmatic 5.0

sliceOmatic 6 - Previous versions
Critical updates New features an updates Minor changes


2024
rev-9d5

 25/Nov/2024
  • Major changes to AI TAG 2D:
    - We now specify the pixel dimensions instead of the image resolution.
    - Choices of 4 model normalization: "Off", "Batch", "Group" or "Instance".
    - You now have to install the Python library: "Tensorflow_Addons".
    - You need the Python library Tensorflow 2.10.0
    - You need the Python library Numpy 1.26.4
  • Add the AI TAG 3D module.
  • Add support for 3D GLI export.
  • Add "NIfTI" and "NRRD" to GLI export.
  • Fix a bug in DICOM "Multi-frame Grayscale Word Secondary Capture" where the first frame had the position of the last frame.
  • In AI "training" modules, we no longuer save the weights at all computed epochs, instead we try to keep the number of saved weights to arround 100.
  • In AI "training" modules after each weight save, we cleanup temporary "..._temp" folders created by python.
  • In AI "save" modules, the list of available weights is now sorted with the last weights firts.
  • In the Multi TAG tool "TAG Compare" we added support for 3D volumes.
Additional Modules
rev-9d5

 28/May/2024
  • Support 3D TAG files for 3D GLI files.
  • Support NIfTI as TAG files: In addition to the Config TAG option to create NIfTI files, if you simply rename a NIfTI with the same name as the GLI and add ".tag" extension, the program will read it as the associate TAG file and keep the NIfTI format when saving.
  • Slight improvements on the header of the trace files.
  • Added 2 new AI variables:
    $AI_NORMALIZATION (def 1) Add normalization steps in the AI models if > 0
    $AI_DROPOUT (def 0.3) Add dropout steps (of $AI_DROPOUT values) in the AI models if > 0
    These can be accessed through the AI Training interface.
  • Work on NIfTI input and output (flipped the vertical axis for exported TAG files).
Additional Modules
rev-9d3

 01/April/2024
  • Stop reading the "Python: interpreter ..." info from the sliceO_ini.scp files since that info is now either located directly by the program or specified in the "Python_Path.txt" file.
  • Added the "Flat/3D" option for the interface in the "Config" menu.
  • Fixed a mix-up in the labels for the AI 2D "Intensity Normalization" buttons.
Additional Modules
rev-9d2

 28/March/2024
  • Fix a crash with the Autosave.
rev-9d

 17/March/2024
  • Changed the Reslice module to reslice in any directions. The oblique angle can be set with sliders, manually or with 2 to 3 points.
  • Fix a crash with "TAG Highlight" when deleting files while the highlight is on.
rev-6c6
  • Complete re-design of the "Reslice" mode. It can now create a 3D volume out of almost anything.
  • Fix a problem where the last pixels on the right and at the bottom where not displayed.
  • Fix a small bug in side view that cause ray-tracing volumes to be computed twice when changing groups.
rev-6c5
  • Add image zoom on mouse wheel. The zoom factor can be changed with "$FRAME_ZOOM_FACTOR" (def: 1.1)
  • Add image contrast adjustment on "Ctrl" and mouse wheel. The contrast is centered on the pixel under the cursor.
  • Add a matching "Mouse Scrool Zoom Factor" control in the config menu.
  • Made the "Multi TAG Select" tool available in the 3D modes.
rev-6c4
  • Improve peak detection in the "Histogram Threshold" mode: I now filter the histogram until only 10 peaks remain and keep only the desired number of these peaks using their dynamic range. In the previous version I would filter the histogram until the desired number of peaks remain (based on the zero crossing of the derivative of the histogram), but this might discard some valid peaks and keep peaks that are not much more than noise on the curve.
  • Fix a bug in "Tool/Automation" that prevent mouse clicks to be recognized.
  • Add more traces to the AI augmentation.
  • Fix a problem in ABACS when the "Document" folder path is more than 80 char long.
  • Fix a problem with the browser when you had more than 1000 files having a ".tag" segmentation.
rev-6c3
  • Fix a problem in rev-9c2 where some of the Python files where not the correct versions
  • Add some checking on the CUDA toolkit and CuDNN library if you have installed the AI modules.
Additional Modules
2023
rev-9c2

 28/August/2023
  • Add the possibility to read multiple 2D segmentation from another source in DICOM or NifTi format. (in 7b I added the possibility of reading external segmented volumes, now it also work with individual slices.)
  • Improved the speed of the TAG "Transfer" module.
  • Fix a problem with MPR Ortho and MPR Scout that did not recognize volumes as valid 3D.
  • Fix a problem with Side View for sagittal and coronal datasets.
  • Fix a problem where you could "right click" on multiple TAG buttons.
  • Better management of the Threshold values in Region Growing and ABACS.
    Note: A bug in the threshold values of "Region Growing" was introduced in rev-9c and corrected in 9c2.
  • Remove the "Modality" information and restrictions and changed the "Normalization" interface from the "AI 2D" module.
Additional Modules
rev-9b

 25/June/2023
  • Finished the "AI 1D Fct" module to the AI group by adding the "Predict" functionality.
  • Moved the modes "Class management (1D)" and "File management (1D)" from additional modes to the basic sliceO package.
  • Add a "Display (1D)" mode to display the 1D CSV database.
  • Add the possibility to load CSV files in sliceOmatic:
    - The modes "Display (1D)", "Class management (1D)", "File management (1D)" and class 1D "Index" now display the 1D CSV database.
    - Add a new "File" group "Menu 1D (CSV)" with 3 new menu DLLs": "CSV Browser", "Open CSV file" and "CSV save file".
    - Add a new 1D class: "Index" that enable to designate any of the 1D dataset column as an index column.
    - Modify the existing 1D classes. They now act as "Index".
    - The "Study 1D" class will cause column "Slice Info, Study ID" to be an index column.
    - The "Series 1D" class will cause column "Slice Info, Series num" to be an index column.
    - The "Position 1D" class will cause column "Slice Info, Slice position" to be an index column.
  • Moved the "Multi TAG" tools from additional tools to the basic sliceO package.
  • Rename and re-order some of the "File" menu groups and items.
  • Fix a bug that prevent volumes to be displayed or exported for all but the last TAG in a volume.
  • Fix a problem with shell re-construction of some spherical system geometries where the polygon's normals where inverted.
  • Revived the "-d" command line argument. You can use "-d path" to override the user's path (Documents).
  • Improved the finding of the Python installation directory.
  • Disable the AI "Directory of the Python Interpreter" config option. If needed, this path is now set with the "Python_Path.txt" file in the sliceO's install dir.
Additional Modules
rev-9

8/May/2023

Note: There has not been an "Official" release of rev-9
  • Add the first part of the "AI 1D Fct" module to the AI group.
    Note: The "Predict" portion is still unavailable.
  • Fix a problem in AI 2D TAG where files saved for "U-NET 1" (now U-NET Light) where read back as "U-NET".
  • Add some "BatchNormalization" to the "U-NET Dense" model to help it converge.
  • Add the "Rotation 180" to the AI 2D TAG augmentations.
  • Add version numbers to all Python AI files.
  • Change the AI 2D Models: Fix a problem that cause all previous computed weights to become incompatible. You will need to re-train your models.
  • Fix a problem in the AI_2D_TAG "Tversky" metrics. They are now controlled by 2 sliceO variables: $AI_TVERSKY_ALPHA and $AI_TVERSKY_BETA (default = 0.5).
  • Add the "Study-ID" to the pQCT Stratec format.
  • Fix a bug with the script browser.
rev-8

 8/March/2023
  • Add the Radiomics module.
  • Add a new option in the "TAG" tab of the "File/Config" menu that enable you to read/write TAG files in DICOM (.seg) and NIfTI (.nii) format in addition to the default ".tag" TomoVision format. However, this option is only applied to the "A" TAG files if you are using multi TAGs. The "B" to "D" TAGs still use the TomoVision format.
  • Fix a problem reading back ".tbg", ".tcg", ".tdg" TAG files if there was no ".tag".
  • Add better support for reading back 3D TAG files.
  • In "ultrasound" modes, when reading a Samsung MVL 4D file, the different 3D are not identified by the "Series" class, resulting in all 3D being seen as one volume. I add a pop-up and the Acquisistion class to solve this.
  • Fix a problem with some Cartesian MVL ultrasound files.
Additional Modules
rev-7b

 10/January/2023
  • Add support for Nifti files.
  • Add support for binary files. If you drop a binary file on sliceO, it will offer to use it as TAG values. The resolutin of the binary need to match the resolution (and number) of loaded frames. You can map the binary to any TAG value. You can replace/add or substract these TAG values to the existing TAG. These operation will take the "TAG Lock" values into account.
  • Fix a potential deadlock in the Geom Mask mode.
  • Small change to Toshiba CT where I had a problem on the FoV circle.
  • Fix bug in Snapshot of frames.
  • Fix a few crashes in Ultrasound modes.
Additional Modules
2022
rev-7

 12/November/2022
  • Add a "Reslice" mode to resample 3D volumes.
  • Fix a potential deadlock in the MPR classes.
  • Fix a potential deadlock in the Geom Mask mode.
  • Fix a cosmetic problem when displaying more than a 62 512x512 images full scale. Some icons in the "Mode All" view where displayed incorrectly when scrolling past the 62nd 512x512 image. After 62 images, the virtual position of the images is higher than 32,000 and since I was using a short... I now use int values, so the problem will only re-surface if you display more than 4 million images.
  • Add stats for "selected frames" in addition to the current frame in Noise FFT tool. Previously the Noise stats where only available for one frame, the current frame. Now you can have the mean stats (and histogram) for all the selected frames.
  • Fix a problem in AI where the ETA was greatly overestimated.
  • Fix a few bugs in the "File" page of the AI when you just read a few files from a dir at a time.
Additional Modules
rev-6

 26/October/2022
  • Add 3 new "Multi TAG" tools: "Select", "Copy" and "Compare". These tools are used to associate multiple (up to 4) TAG buffers to each frame. The 3 new TAG buffers will be saved/read as ".tbg", ".tcg" and ".tdg".
  • Fix a rounding error in Region Growing.
  • Fix a couple of problems in the scripts: a problem with negative values and another with variables as index of other variables.
  • You can now use the AI with images from different resolutions.
  • Add the option of doing a double click in AI "Test" page to copy the "true" and "Computed" TAG values to sliceO. These TAg files are also written to disk (as .tbg and .tcg).
  • Fix a problem with the "Transfer" mode where the lists where done from the root childrens, now the program look for "study" classes and use these. Add some pop-up and change the slice matching algo.
Additional Modules
rev-5

 5/October/2022
  • Add the first version of the AI Python module.
  • Additional Modes, Tools, Classes and Protocols are now part of the basic download. They are not selected by default, but you no longuer have to download them separatelly. You just need to select the modules you want at install time.
  • Add the "TAG Highlight" tool.
  • Fix a warning message for files that do not have a study ID.
  • Some improvements on the Plot windows.
  • Fix a bug with Results 2D scripts that have more than 1000 cells (Yes, it did happen!). The new limit is 4096.
  • Fix a bug with Results when TAG Nb = 256.
  • Fixed a problem when reading DICOM images that have a " " instead of "SS" as explicit VR value. (Seen on images created with Huahai systems).
  • Add the "TAG Transfer" mode. This mode enable to copy TAGs from one study to the slices of another, even if the target slices do not have the same dimension or resolution, or even orientation.
  • Add the flag "PEN_FLAG_ERASER" to the pen interface.
  • Fix a problem with the icons on the buttons of the Class management 2D mode.
  • Fix an installation script problem that did not install the Visual C DLL.
Additional Modules
rev-4

 25/July/2022
  • Fix a crash in File Management 2D when closing files.
  • Fix a problem with voxels outside the FoV in the ray mode of Side View.
  • Fix a problem with big dataset in Side View. Now I limit the size to 200 million voxels for Volume.
  • Fix a problem in closing 1D & 3D files.
  • Fix a crash when you "abort" an undo operation that did not fit in memory.
  • Change to the tools to allow colors and fonts in labels and names.
  • Fix a problem with the pop-up size of warning in results 2D and 3D.
  • Fix a problem with Results 2D and 3D script for TAG > 9.
  • Add a few skins choices in Config.
Additional Modules
rev-3

 30/June/2022
  • Fix a bug in 2D/2D results that prevent the program from saving config that contain min, max and mean CT info.
  • Fix a bug in 2D/3D resuts where the units prevent the selection to be recognized from the script.
  • Fix a bug in script saving for the ultrasound VR module.
  • Fix a problem in "2D File management/TAG". The changes to the tag path where not used.
  • Increase the "Side View" Volume time-out from 30 sec to 5 min.
  • Fix a problem with the drag&drop of TAG files.
  • Add wide char string support in the rules files.
Additional Modules
rev-2

 15/June/2022
  • Fix a bug preventing ABACS to work.
  • Add the "Ultrasound" modes, tools, menu and config options. This is the equivalent of "Baby_SliceO"
    Note: the Ultrasound modes are only enabled on system with a Cuda compatible GPU.
  • Fix a bug in Side_View if you do not have enough GPU mem to show the model in "Ray" mode.
Additional Modules
rev-1

 15/May/2022
  • First release of SliceO 6.
  • Complete change of the internal character representation. I now use Unicode to enable to work with foreign characters in the file/dir names.
    - All script files are now in UTF-8 for the same reason.
    - CSV files are now saved in UTF-16LE instead of the old ASCII format.
  • Change the default path for a lot of files:
    - The User directory has been moved to "Documents"\TomoVision (c:\Users\user_name\Documents\TomoVision) instead of c:\Users\user name\AppData\Roaming\Tomovision. This should help the users reach it since AppData was hidden and hard to reach.
    - The scratch dir has been moved to the same dir: "Documents"\TomoVision instead of c:\temp since c: is protected on the most paranmoid systems and I could not create the Temp folder there.
    - The snapshot will send its images to "Pictures" (c:\Users\user name\Pictures)
    - The traces will be sent to the "Desktop" (c:\Users\user name\Desktop)
  • Placed all the interface texts in the "SliceO_Interface" sub directory.
  • Add the possibility to collapse/expand groups in the File/Tools and Mode menus.
  • Completely re-design the debug/trace modes (accessible with "Ctrl-T")
  • Add a bunch of "Status" info options (including "License Info") in the File menu.
  • Add a "Filter FFT" class in "Clases" that enable to filter images from the FFT transform.
  • Add a "Noise FFT" tool in "More tools" that evaluate the noise in the images from the FFT transform.
  • Remove the ABACS+ module at the request of Voronoi.
  • In the Side View Tool:
    - Fix an orientation problems, in side View, with non-DICOM images that have Z going from head to toe (old Siemens).
    - Add a 90/360 deg option for the ray-tracing fct of side_view (only available if you have a CUDA card).
    - Add an arrow representing the viewer in addition to the clip plane line when using Side View.
  • Fix a problem with the script commands: "color: black val" and "color: white val".
  • In the Geometrical Mask Mode:
    - Add a "Line" tool. This line has a thickness defined with:
    "mask: thickness value".
    - Add a "Square" tool. This square, like the "Surface" tool, has a fixed surface. The parameter "side" is used to define the size of the square. The edges of the square are side in length, the surface is (side)**2.
    "mask: side value"
    - Slightly decreased the height of the "Mask" mode interface.
  • In the Mixer Class:
    - Fix a crash in Mixer if the files names where to long.
    - The values of "P" and "T" are now defined for both GLI and TAG eq in the mixers. (It use to be that the GLI pixel values "P" was only defined for the GLI eq, and the tag value "T" only for the TAG eq.) By having both defined in all equations, we can now do things like "(T1=2)?0:P1" to mask pixels of the GLI images based on their TAG values.
    - Add the "Key Words" option in the Mixer to enable the module to discriminate/associate based only on specific key words in the specified DICOM tags.
  • Fix a problem with 2D and 3D results where the default save dir was c:\temp even if the scratch dir has been changed to something else. I also added a new option to the save directory, the choices are now: fixed (you select the dir), source dir (the dir where the source files are) or scratch dir (now "Documents"/TomoVision).
  • Fix a problem that cause the points to be invisible in 3D.
  • Fix a problem with the protractor in 2D Measures.
Additional Modules

Scroll
to Top