Annotation of entities and relations in Spanish Radiology Reports.

ROLAND ROLLER; DARIO FILIPPO; FEIJU XU Y HANS USZKOREIT; VIVIANA COTIK

Conferencia; Recent Advances in Natural Language Processing; 2017

V. Cotik, R. Roller, D. Filippo, F. Xu, H Uszkoreit.

Radiology reports express the results of aradiology study and contain informationabout anatomical entities, findings, measuresand impressions of the medical doctor.The use of information extractiontechniques can help physicians to accessthis information in order to understanddata and to infer further knowledge.Supervised machine learning methods arevery popular to address information extraction,but are usually domain and languagedependent. To train new classifi-cation models, annotated data is required.Moreover, annotated data is also requiredas an evaluation resource of informationextraction algorithms. However, one majordrawback of processing clinical data isthe low availability of annotated datasets.For this reason we performed a manualannotation of radiology reports written inSpanish. This paper presents the corpus,the annotation schema, the annotationguidelines and further insight of the data.