Automatic image annotation or image classification can be an important step when searching for images from a database. Base on the IRMA project a database of 11,000 fully classified radiographs taken randomly from medical routine is made available and can be used to train a classification system. 1,000 radiographs for which classification labels are not available to the participants have to be classified. The aim is to 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 classification task this year will consider the complete IRMA code and it will be up to the groups to what level of detail the images will be annotated. Therefore, errors in the annotation will be counted depending on the depth in the tree, and the difficulty of the choice.
Training set and extended Training set now available:
The training data and the extended training data (dev set for tuning the parameters) is now available.
Downloads (you will need the username/password that is provided when registering for ImageCLEF 2007):
Note: if you are not able to download the data using the links directly you have to go to the "Information on the datasets"-page, login using the credentials that are provided upon registration for ImageCLEF and then you can retry the links or navigate to on the IRMA page to the downloads.
In ImageCLEF 2007, the medical automatic annotation task considers the complete IRMA code and penalizes misclassifications at different levels of the code differently.
A detailed description of the error counting scheme is here
The submission format will be
<imageno> <imagecode> <imageno> <imagecode> ...e.g.
2034 1121-127-700-500 2229 1121-110-411-700 2630 1121-120-942-700 2633 1121-120-951-700 2711 1121-120-921-700 ...
For assistance, we offer some tools:
The most interesting and successful submissions to this task are invited to submit a paper to a special issue in Pattern Recognition Letters :
Automatic annotation of medical images for image retrieval -- ImageCLEF 2007
Guest Editors: Thomas Deselaers, Henning Müller, and Thomas M. Lehmann