[visionlist] Call-for-Participation: 1st GANs Task @ ImageCLEF 2023 (Synthetic Medical Images)

[Bogdan Ionescu]

[Apologies for multiple postings]

ImageCLEFmedicalGANs (1st edition)
Registration: https://www.imageclef.org/2023/medical/gans
Run submission: May 10, 2023
Working notes submission: June 5, 2023
CLEF 2023 conference: September 18-21, Thessaloniki, Greece


*** CALL FOR PARTICIPATION ***
The task is focused on examining the existing hypothesis that GANs are
generating medical images that contain the "fingerprints" of the real
images used for generative network training. If the hypothesis is
correct, artificial biomedical images may be subject to the same
sharing and usage limitations as real sensitive medical data. On the
other hand, if the hypothesis is wrong, GANs may be potentially used
to create rich datasets of biomedical images that are free of ethical
and privacy regulations. The participants will test the hypothesis by
solving one or several tasks related to the detection of relations
between real and artificial biomedical image datasets.


*** TASK ***
Given a set of real-world medical images comprising 2D axial CT image
slices of the heart (including the middle sections and adjacent
slices) of patients afflicted with lung #tuberculosis, the task
challenges participants to develop #machinelearning solutions to
automatically determine which real images were used in training the
generator of realistic synthetic examples.


*** DATA SET ***
The image datasets comprise 2D axial CT image slices of the heart,
including the middle sections of the heart and adjacent slices. These
images are obtained from patients afflicted with Lung Tuberculosis and
are stored in the form of 8 bit/pixel PNG images with dimensions of
256x256 pixels. The development dataset comprises three distinct sets
of images. One set contains images that were generated using a GAN,
while the other two sets are comprised of real images. The first of
these real image sets contains images that were used during the
algorithm's training process. The second set consists of real images
that were not used during the training process. Test dataset is a
collection of two image sets. The first set contains 10,000 images
that have been generated, while the second set is made up of a
combination of 200 real images that were either used or unused during
the training process.


*** IMPORTANT DATES ***
- Run submission: May 10, 2023
- Working notes submission: June 5, 2023
- CLEF 2023 conference: September 18-21, Thessaloniki, Greece
(https://clef2023.clef-initiative.eu/)


*** OVERALL COORDINATION ***
Serge Kozlovski, Belarusian Academy of Sciences, Belarus
Vassili Kovalev, Belarusian Academy of Sciences, Belarus
Ihar Filipovich, Belarus State University, Belarus
Alexandra Andrei, Politehnica University of Bucharest, Romania
Ioan Coman, Politehnica University of Bucharest, Romania
Bogdan Ionescu, Politehnica University of Bucharest, Romania
Henning Müller, University of Applied Sciences Western Switzerland, Switzerland


*** ACKNOWLEDGEMENT ***
Alexandra Andrei, Ioan Coman, Bogdan Ionescu, and Henning Müller
contribution is supported under the H2020 AI4Media "A European
Excellence Centre for Media, Society and Democracy"  project, contract
#951911 https://www.ai4media.eu/.


On behalf of the Organizers,

Bogdan Ionescu
https://www.AIMultimediaLab.ro/