Competition

MedNIST

Organized by challenge-organizer - Current server time: March 20, 2023, 3:22 p.m. UTC

Previous

Training
Aug. 1, 2020, 6:53 p.m. UTC

Current

Final
Aug. 20, 2020, 6:53 p.m. UTC

End

Competition Ends
Never

MedNIST

Brought to you by MedICI

Image Classification Tutorial with the MedNIST Dataset

Introduction In this tutorial, we introduce an end-to-end training and evaluation example based on the MedNIST dataset.

You'll go through the following steps:

  • Get the dataset
  • Train the model with a PyTorch program (on your own).
  • Evaluate on test dataset during the Test Phase

The MedNIST dataset was gathered from several sets from TCIA, the RSNA Bone Age Challenge, and the NIH Chest X-ray dataset.

The dataset is kindly made available by Dr. Bradley J. Erickson M.D., Ph.D. (Department of Radiology, Mayo Clinic) under the Creative Commons CC BY-SA 4.0 license. If you use the MedNIST dataset, please acknowledge the source, e.g.

mednist_tutorial.ipynb The following commands download and unzip the dataset (~60MB): wget -q https://www.dropbox.com/s/5wwskxctvcxiuea/MedNIST.tar.gz

MedICI: Evaluation

Phase 1 is a place holder phase allowing you time to gather and look at the data as well as train your model.

At some point you will use the Upload Docker Submission tab under the Participate Tab to upload an image with your model inside. You will need to setup you docker image to accept a folder of images at '/mnt/in/ and post your prediction to '/mnt/out/'. '/mnt/in/' will house the test data set you will perform inference on. This upload process will accept a *.tar or *.tar.gz docker image created with:

$ docker save repo/image:tag | gzip > image_name.tar.gz or $ docker save repo/image:tag > image_name.tar

Upon completion of the upload you will see a list of available image tags. These are images in a private repo and we will provide you the tag for each of your images.

Phase 2 will accept the docker tags you upload as a submission. We will then run your docker image with the afore mentioned mounts so that your model can infer the correct classes from the test input.

MedNIST: Rules

Use training phase to traing your model.

Use docker image submission for test phase.

Training

Start: Aug. 1, 2020, 6:53 p.m.

Description: Training phase: create models and upload your docker images in preparation for the test phase.

Final

Start: Aug. 20, 2020, 6:53 p.m.

Description: Test phase

Competition Ends

Never

You must be logged in to participate in competitions.

Sign In
# Username Score
1 challenge-organizer 0.154