Anatomical Entity Recognition with a Hierarchical Framework Augmented by External Resources

Abstract

References to anatomical entities in medical records consist not only of explicit references to anatomical locations, but also other diverse types of expressions, such as specific diseases, clinical tests, clinical treatments, which constitute implicit references to anatomical entities. In order to identify these implicit anatomical entities, we propose a hierarchical framework, in which two layers of named entity recognizers (NERs) work in a cooperative manner. Each of the NERs is implemented using the Conditional Random Fields (CRF) model, which use a range of external resources to generate features. We constructed a dictionary of anatomical entity expressions by exploiting four existing resources, i.e., UMLS, MeSH, RadLex and BodyPart3D, and supplemented information from two external knowledge bases, i.e., Wikipedia and WordNet, to improve inference of anatomical entities from implicit expressions. Experiments conducted on 300 discharge summaries showed a micro-averaged performance of 0.8509 Precision, 0.7796 Recall and 0.8137 F1 for explicit anatomical entity recognition, and 0.8695 Precision, 0.6893 Recall and 0.7690 F1 for implicit anatomical entity recognition. The use of the hierarchical framework, which combines the recognition of named entities of various types (diseases, clinical tests, treatments) with information embedded in external knowledge bases, resulted in a 5.08% increment in F1. The resources constructed for this research will be made publicly available.

Mapping anatomical related entities to human body parts based on wikipedia in discharge summaries

“…50 discharge summaries from the i2b2 Challenge [51] are used in our experiments, from which 2224 anatomical named entities are extracted [5]. F1-measure is used to evaluate the quality of each method on two series of experiments.…”

“…Our annotation dataset comes from the output of our previous work [5], which used a CRF model to recognize anatomical named entities from 300 discharge summaries in the 2010 i2b2 challenge corpus [51]. The discharge summaries are provided by Partners Healthcare, Beth Israel Deaconess Medical Center, and the University of Pittsburgh Medical Center.…”

“…Name entities that do not refer to but are highly related to human body parts are defined as implicit anatomical related entities (e.g., colonoscopy). In our previous research [5], a Conditional Random Fields (CRF) model was presented for anatomical entity recognition. Entities from discharge summaries are recognized, extracted, and labeled as explicit or implicit entities.…”

Essential Elements of Natural Language Processing: What the Radiologist Should Know

“…Other examples of medical ontologies include the Unified Medical Language System and Medical Subject Headings. Ontologies have been successfully used in NER algorithms explicitly designed for the medical domain (29,30).…”

EXTRACT: interactive extraction of environment metadata and term suggestion for metagenomic sample annotation

“…automatic identification of terms mentioned in text, can assist the standards-compliant reporting of sampling environments in two primary ways. Although most work on biomedical NER has focused on recognition of gene and protein names ( 8 ), methods also exist for recognizing terms more relevant to sample annotation, namely organisms ( 9 , 10 ), tissues ( 11 , 12 ), diseases/phenotypes ( 13–17 ) and environments ( 18 ). NER can be used to suggest terms based on existing free-text fields in sequence/sample repositories or the literature associated with the samples.…”