The Language of Brain Signals: Natural Language Processing of Electroencephalography Reports

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 2268–2275 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 2268 The Language of Brain Signals: Natural Language Processing of Electroencephalography Reports Ramon Maldonado, Sanda M. Harabagiu The University of Texas at Dallas 800 W. Campbell Rd., Richardson, TX 75080 {ramon,sanda}@utdallas.edu Abstract Brain signals are captured by clinical electroencephalography (EEG) which is an excellent tool for probing neural function. When EEG tests are performed, a textual EEG report is generated by the neurologist to document the findings, thus using language that describes the brain signals and their clinical correlations. Even with the impetus provided by the BRAIN initiative (brainitititive.nih.gov), there are no annotations available in texts that use natural language describing the brain activities and their correlations with various pathologies. In this paper we describe an annotation effort carried out on a large corpus of EEG reports, providing examples of EEG-specific and clinically relevant concepts. In addition, we detail our annotation schema for brain signal attributes. We also discuss the resulting annotation of long-distance relations between concepts in EEG reports. By exemplifying a self-attention joint-learning method used to predict concept, attribute and relation annotations in the EEG report corpus, we discuss the promising results of automatic an- notations, hoping that our effort will inform the design of novel knowledge capture techniques that will include the language of brain signals. Keywords: language of the brain, medical concepts, long-distance relations. 1. Introduction The diagnosis and clinical management of epilepsy as well as the evaluation of other types of brain disorders (Smith, 2005), including encephalopathies, neurological infections, Creutzfeldt-Jacob disease, and even in the evaluation of the progression of Alzheimer’s disease are informed by clinical electroencephalographies (EEG). An EEG records electri- cal activity along the scalp and measures spontaneous elec- trical activity of the brain. The signals measured along the scalp can be correlated with brain activity. But, as noted in (Beniczky et al., 2013), the complexity of the EEG sig- nal, interpreted and documented in EEG reports, produces inter-observer agreement known to be moderate. As more clinical EEG signals and reports become available, the in- terpretation of EEG signals can be improved by providing neurologists with results of search for patients that exhibit similar EEG characteristics. Neurologists that are interested in patients that exhibit cer- tain EEG characteristics can use in their their queries med- ical concepts that characterize the EEG tests, e.g. EEG activities of the brain, EEG events, or the clinical corre- lations between brain signals and pathologies of the brain. In order to discover relevant EEG reports, there is a need to automatically identify EEG-specific concepts as well as relations they hold to the clinical correlations of the brain signals. The ability to identify such concepts is hindered by the absence of annotations in medical texts covering EEG- specific information. In this paper we describe an annotation effort performed on a corpus of 23,000 EEG reports from over 15,000 patients collected over 12 …

Content truncated to 500 words