Event-based clinical findings extraction from radiology reports with pre-trained language model
This work addresses the need for comprehensive semantic representations of radiological findings to support diagnosis and research in healthcare, representing an incremental improvement with domain-specific application.
The paper tackled the problem of extracting detailed clinical findings from radiology reports by creating a new annotated corpus and using a BERT-based model, achieving F1 scores of 90.9%-93.4% for finding triggers and 72.0%-85.6% for argument roles on internal data, with generalization to external data showing 95.6% for triggers and 79.1%-89.7% for roles.
Radiology reports contain a diverse and rich set of clinical abnormalities documented by radiologists during their interpretation of the images. Comprehensive semantic representations of radiological findings would enable a wide range of secondary use applications to support diagnosis, triage, outcomes prediction, and clinical research. In this paper, we present a new corpus of radiology reports annotated with clinical findings. Our annotation schema captures detailed representations of pathologic findings that are observable on imaging ("lesions") and other types of clinical problems ("medical problems"). The schema used an event-based representation to capture fine-grained details, including assertion, anatomy, characteristics, size, count, etc. Our gold standard corpus contained a total of 500 annotated computed tomography (CT) reports. We extracted triggers and argument entities using two state-of-the-art deep learning architectures, including BERT. We then predicted the linkages between trigger and argument entities (referred to as argument roles) using a BERT-based relation extraction model. We achieved the best extraction performance using a BERT model pre-trained on 3 million radiology reports from our institution: 90.9%-93.4% F1 for finding triggers 72.0%-85.6% F1 for arguments roles. To assess model generalizability, we used an external validation set randomly sampled from the MIMIC Chest X-ray (MIMIC-CXR) database. The extraction performance on this validation set was 95.6% for finding triggers and 79.1%-89.7% for argument roles, demonstrating that the model generalized well to the cross-institutional data with a different imaging modality. We extracted the finding events from all the radiology reports in the MIMIC-CXR database and provided the extractions to the research community.