%0 Journal Article
%A Dada, Amin
%A Ufer, Tim Leon
%A Kim, Moon
%A Hasin, Max
%A Spieker, Nicola
%A Forsting, Michael
%A Nensa, Felix
%A Egger, Jan
%A Kleesiek, Jens
%T Information extraction from weakly structured radiological reports with natural language queries.
%J European radiology
%V 34
%N 1
%@ 0938-7994
%C Heidelberg
%I Springer
%M DKFZ-2023-01528
%P 330-337
%D 2024
%Z 2024 Jan;34(1):330-337
%X Provide physicians and researchers an efficient way to extract information from weakly structured radiology reports with natural language processing (NLP) machine learning models.We evaluate seven different German bidirectional encoder representations from transformers (BERT) models on a dataset of 857,783 unlabeled radiology reports and an annotated reading comprehension dataset in the format of SQuAD 2.0 based on 1223 additional reports.Continued pre-training of a BERT model on the radiology dataset and a medical online encyclopedia resulted in the most accurate model with an F1-score of 83.97
%K Information extraction (Other)
%K Machine learning (Other)
%K Natural language processing (Other)
%F PUB:(DE-HGF)16
%9 Journal Article
%$ pmid:37505252
%R 10.1007/s00330-023-09977-3
%U https://inrepo02.dkfz.de/record/277860