Mistos

Microscopy Image Storing and Processing System

Requirements

• Windows 10
• 2020.11 64Bit Individual Edition (https://repo.anaconda.com/archive/Anaconda3-2020.11-Windows-x86_64.exe)
• Chrome browser (recommended)
• Oracle JDK > 1.6
• microsoft visual c++ 14.0
• (CUDA enabled GPU for faster processing)
  • List of Cuda enabled GPU's: https://developer.nvidia.com/cuda-gpus
  • Guide to setup GPU support: https://www.tensorflow.org/install/gpu

Setup

Git

• If you already have the latest version of this repository stored locally, skip this section
• If you have not used git before, download and install it
  • https://git-scm.com/downloads
  • Make sure to check "Git from the commandline and also from 3rd party software" option during installation
• Get latest Mistos build by opening a commandline in a folder where you want to save it
• run "git clone https://github.com/Maddonix/mistos_2.git"

Oracle JDK

• If already installed (check in system contro -> programs) you may skip this section
• Download and install:
  • https://www.oracle.com/de/java/technologies/javase-jdk15-downloads.html
  • choose appropriate version (e.g. Windows 64 bit)

Visual Studio

• If you already have a Visual C++ Redistributable ( >14.0 ) you may skip this step
• Otherwise visit https://visualstudio.microsoft.com/downloads/, download and install (check the C++ build tools box)

Anaconda

• If you already have an up-to-date version of anaconda installed, you may skip this section (make sure conda is accessable from commandline!)
• Download appropriate Anaconda Version for your system. (Tested with: Windows, Python 3.8, 64-Bit)
  • https://www.anaconda.com/products/individual
• Install Anaconda
  • add conda to path during installation
• Activate conda in the commandline
  • open "anaconda prompt"
  • run "conda init cmd.exe"

Anaconda Mistos Environment

• Double-click "setup.bat"
  • Warning: Since Mistos implements neuronal networks based on pytorch and tensorflow, both are installed. This takes some time, so you may want to grab a coffee or something to eat in the meantime.

Filepaths

• By default, a Mistos directory will be created in your home directory (eg. "c:\users\tlux\mistos")
• Optionally you may change the filepaths:
  • Open the config.json file
  • Enter a valid file path to an existing folder for
    • "EXPORT_DIRECTORY" (all exports will be stored here)
    • "WORKING_DIRECTORY" (internal storage for the app)

Start App for the first time:

• for python terminals activate conda: -- d:\programming\venv\mistos\scripts\activate.ps1 (powershell) -- d:\programming\venv\mistos\scripts\activate (regular terminal)

Frontend

• open a commandline or powershell in "mistos_2" directory
• run "conda activate mistos"
• run "python frontend_server.py"
• allow network access for private networks if requested

Fileserver

• open a commandline or powershell in "mistos_2/mistos-backend/src" directory
• run "conda activate mistos"
• run "python fileserver/server.py"
• allow network access for private networks if requested

Backend

• open a commandline or powershell in "mistos_2/mistos-backend/src" directory
• run "conda activate mistos"
• run "python python main.py"
• allow network access for private networks if requested

Start App After the First Time

• Double-click "start_mistos.bat"

Update Environment

• Double-click "update_mistos.bat"

Troubleshooting

• Anaconda not added to PATH correctly

  • Solution: Add Anaconda to path
    • Find Anaconda path (default is: C:\users\{username}\Anaconda3)
    • Open the Windows menu "Systemumgebungsvariablen bearbeiten"
    • Go to "Umgebungsvariablen..."
    • double-click on "Path" in the upper list ("Benutzervariablen für {username}")
    • click "New" and paste the Anaconda path
    • click okay and close the menus
    • open the "anaconda prompt"
    • run "conda init cmd.exe"
    • a reboot may be necessary for changes to take effect

• napari errors:

  • get latest napari build by opening a commandline in a folder where you want to save it
  • run "git clone https://github.com/napari/napari.git"
  • run "cd napari"
  • run "conda activate mistos"
  • run "pip install ."

• Deepflash errors:

  • get forked deepflash2 by opening a commandline in a folder where you want to save it

  • run "git clone https://github.com/Maddonix/deepflash2.git"

  • run "cd deepflash2"

  • run "conda activate mistos"

  • run "pip install -e ."

  • hard coded shape error:

    • deepflash2/models.py line341: m.conv1 = nn.Conv2d(1, self.inplanes, kernel_size=7, stride=2, bias=False) -> m.conv1 = nn.Conv2d(in_channels, self.inplanes, kernel_size=7, stride=2, bias=False)

• You changed the filepaths under config before deleting all entries:

  • The database will still hold references to the old paths, the app will not work anymore
  • Option 1:
    • change the paths back, the app should work again
    • delete all experiments and images
    • now you may change the path
  • Option 2:
    • This option will delete all old database references, only do it if you don't need them anymore
    • go to the mistos app folder
    • go to mistos-backend/src
    • delete the file "sql_app.db"

• Conda environment broken:

  • Solution: Rebuild the environment
    • open a commandline in the mistos project folder
    • run "conda remove --name mistos --all"
    • run "conda env create -f environment.yml"

Main Views:

• Dashboard (default)

• File Viewer

• Experiment Viewer

• Maybe: MachineLearning Screen?

Tutorial

For a detailed tutorial, open the Mistos/tutorial folder and follow the instructions.