NIH | National Cancer Institute | NCI Wiki  

Versions Compared

Old Version 1

changes.mady.by.user Klinger, Carolyn (NIH/NCI) [C]

Saved on

compared with

New Version Current

changes.mady.by.user Klinger, Carolyn (NIH/NCI) [C]

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Google Analytics Report

...

Google Analytics from August 1-31, 2022 for the MedICI websitefor Image RemovedGoogle Analytics from August 1-31, 2022 for the MedICI websitefor Image Added

Activities completed this month

  • MIDRC - Challenge1 planning is starting. We received the new data and are prepping the annotation app for receiving images. This is a COVID detection challenge and we are using chest X-RAYs. Further analysis is needed to determine the images we will use.
    • Lots of re-analysis is happening. There is concern that there is no “signal” in the images. Essentially the COVID (+) images have test results > 20days after the image was taken. This means we cannot be sure the images have COVID. We are filtering the images so that tests are between 1-20 days since the image was taken. We are doing a couple things:
      • Karen, Lubomir and Sam are going through the initial images to filter out poor quality images
      • They will take a second closer look at the images that they didn’t agree unanimously on.
      • Then Carol (an expert annotator) will go through the “best” images and further accept\reject them and point out what artifacts are in them.
    • CodaLab (Open Source Team):
      • We are in the process of getting a GPU VM setup to test GPU challenges.
      • There were issues with networking on the Paris-Saclay servers and since I don’t have access, I have to wait on them to deploy and fix. They are working on getting me VPN access. In the mean time, I will deploy some cpu\gpu workers charged to MedICI\Leidos in order to test if the code works as I have control to access\manage network settings in Azure.
      • There is a new challenge coming up and I intent to launch on our main server https://medici-codalab-main.eastus.cloudapp.azure.com/. In addition I will create a codabench instance or use the one I already have live to simultaneously launch this challenge on codabench to see if it could be ready for future challenges.
    • Federated learning challenge:

      • The challenge is finished. We are collecting finalist algorithms and running them on MGH’s internal data to see how well they performed. We will discuss results soon. First we need to organize MGH data somewhat.

...