Overview of the ImageCLEF 2015 medical classification task

This articles describes the ImageCLEF 2015 Medical Clas-sification task. The task contains several subtasks that all use a dataset of figures from the biomedical open access literature (PubMed Cen-tral). Particularly compound figures are targeted that are frequent inthe literature. For more detailed information analysis and retrieval it isimportant to extract targeted information from the compound figures.The proposed tasks include compound figure detection (separating com-pound from other figures), multi–label classification (define all sub typespresent), figure separation (find boundaries of the subfigures) and modal-ity classification (detecting the figure type of each subfigure). The tasksare described with the participation of international research groups inthe tasks. The results of the participants are then described and analysedto identify promising techniques.