[Imageworld] Announcing ImageCLEF Medical Image Annotation Task

[ttommasi at idiap.ch]

[ apologies for cross-postings; please forward widely.]


We are very happy to announce that the registration for CLEF 2009 -
Cross Language Evaluation Forum - is now open and we would like to
invite you to participate to the


------------Medical Image Annotation Task Challenge ------------



ImageCLEF will offer again this year, for the last time, a Medical Image
Annotation Task. It will be a survey on the last four years experience and
we warmly invite all the participants groups of the last editions
to take part to the challenge. Also new groups both from academic
institutions and industrial organization are very welcome.

Nowadays the use of digital images is rapidly expanding and it highly
influences most biomedical disciplines. In hospitals all the image
acquisition devices are connected to computerized equipments to save,
transfer and store medical imagery in digital format and there is an
increasing trend towards the creation of archives able to support
diagnostics and preventive medicine. The first step to work with such
an archive is to be able to recognize and classify images without relying
on the linked textual description. Visual based indexing can overcome
problems related to not-complete or erroneous textual labels and can
exploit pattern similarities and structural differences in the image
content.

The objective of the challenge is to compare algorithms able to
automatically annotate x-ray images after a training phase and find
out how well current techniques can identify image modality, body
orientation, body region, and biological system examined based on
the images. The results of the classification step can be used for
multilingual image annotations as well as for DICOM header corrections.

The competition will be conducted using images of the IRMA corpus
< http://libra.imib.rwth-aachen.de/irma/index_en.php > as in the previous
years and will contain about 7000 common images of the 2005-2008
training databases. They will be accompanied by three classification
label sets considering:


- 57 classes as in 2005
- 116 classes as in 2006 and 2007
- 197 classes as in 2008


based on different detail levels of the IRMA code, a hierarchical
multi-axial code for image annotation.

About 1500 radiographs for which classification labels are not available
to the participants have to be classified according to the three different
schemes.

We will ask each group to submit runs based on only one algorithm which
should be optimized to face the three different classification problems.
The aim is to understand how each algorithm answers to the increasing
number of classes and to the imbalancing. It would also be possible to
evaluate which is the best way to exploit the hierarchy.

More details on the tasks to be provided soon on the challenge website
< http://imageclef.org/2009/medanno >

All the results of the CLEF 2009 evaluation campaign will be presented
and discussed at the CLEF Workshop. It will be held from 30 September to
2 October in Corfu, Greece, immediately following the thirteenth European
Conference on Digital Libraries (ECDL 2009).

Moreover, we are going to organize workshop at MICCAI 2009
about content based annotation in medical image archieves. The goal is
to foster in-deph discussions and technical interchange among engeneers
and computer scientists who develop image processing algorithms as well
as medical experts who can utilize the porposed technology.

We are also going to contact an editor to prepare and publish a
book describing the most interesting and successful approaches developed
during the last years in this task. The grops participating to the
challenge will be invited to contribute a chapter of the book.


Organisers

Barbara Caputo, Idiap Research Institute, Martigny, Switzerland,
bcaputo at idiap.ch
Tatiana Tommasi, Idiap Research Institute, Martigny, Switzerland,
ttommasi at idiap.ch
Henning Mueller, University and University Hospitals of Geneva,
Switzerland, henning.mueller at sim.hcuge.ch
Thomas M. Deserno, RWTH Aachen University, Medical Informatics, IRMA
group, deserno at ieee.org
Jayashree Kalpathy-Cramer, Oregon Health & Science University, Portland,
OR, USA, kalpathy at ohs