Classification of radiology reports using neural attention models

• Type of Material: Conference Paper
• Authors: Falgun Chokshi, Emory University; Bonggun Shin, Emory University; Timothy Lee, Emory University; Jinho D. Choi, Emory University
• Language: English
• Date: 2017-06-30
• Publisher: Emory University Libraries
• Publication Version: Final Published Version
• Copyright Statement: © 2017 IEEE.
• Final Published Version (URL): http://dx.doi.org/10.1109/IJCNN.2017.7966408
• Title of Journal or Parent Work: Proceedings of the International Joint Conference on Neural Networks
• Volume: 2017-May
• Start Page: 4363
• End Page: 4370
• Grant/Funding Information: We gratefully acknowledge the support of the Foundation of the American Society of Neuroradiology (ASNR) Compar- ative Effectiveness Research (CER) Grant, the Association of University Radiologists (AUR) General Electric Radiology Research Academic Fellowship (GERRAF) Grant, and the Infosys Research Enhancement Grant.
• Keywords: cs.AI; cs.CL; cs.IR
• Research Categories: Computer Science; Engineering, Biomedical

Abstract

The electronic health record (EHR) contains a large amount of multi-dimensional and unstructured clinical data of significant operational and research value. Distinguished from previous studies, our approach embraces a double-annotated dataset and strays away from obscure 'black-box' models to comprehensive deep learning models. In this paper, we present a novel neural attention mechanism that not only classifies clinically important findings. Specifically, convolutional neural networks (CNN) with attention analysis are used to classify radiology head computed tomography reports based on five categories that radiologists would account for in assessing acute and communicable findings in daily practice. The experiments show that our CNN attention models outperform non-neural models, especially when trained on a larger dataset. Our attention analysis demonstrates the intuition behind the classifier's decision by generating a heatmap that highlights attended terms used by the CNN model; this is valuable when potential downstream medical decisions are to be performed by human experts or the classifier information is to be used in cohort construction such as for epidemiological studies.