Open Access Open Badges Research article

Natural language processing of radiology reports for the detection of thromboembolic diseases and clinically relevant incidental findings

Anne-Dominique Pham128*, Aurélie Névéol3, Thomas Lavergne3, Daisuke Yasunaga4, Olivier Clément56, Guy Meyer7, Rémy Morello1 and Anita Burgun28

Author Affiliations

1 Department of Biostatistics and Clinical Research, CHU de Caen, Caen F-14000, France

2 Biomedical Informatics and Public Health Department, University Hospital HEGP, AP-HP, Paris, France

3 LIMSI-CNRS, rue John von Neumann, Orsay F-91043, France

4 Department of Radiology, CHU de Caen, Caen F-14000, France

5 Radiology department, Assistance Publique- Hôpitaux de Paris, Hôpital Européen Georges-Pompidou, 20, rue Leblanc, Paris 75015, France

6 Université Paris-Descartes, INSERM UMR-S970 Paris Cardiovascuar Research center – PARCC, Paris, France

7 Pneumology department, Assistance Publique- Hôpitaux de Paris, Hôpital Européen Georges-Pompidou, 20, rue Leblanc, Paris 75015, France

8 INSERM UMR_S 872 Team 22: Information Sciences to support Personalized Medicine, Université Paris Descartes, Sorbonne Paris Cité, Faculté de Médecine, Paris, France

For all author emails, please log on.

BMC Bioinformatics 2014, 15:266  doi:10.1186/1471-2105-15-266

Published: 7 August 2014



Natural Language Processing (NLP) has been shown effective to analyze the content of radiology reports and identify diagnosis or patient characteristics. We evaluate the combination of NLP and machine learning to detect thromboembolic disease diagnosis and incidental clinically relevant findings from angiography and venography reports written in French. We model thromboembolic diagnosis and incidental findings as a set of concepts, modalities and relations between concepts that can be used as features by a supervised machine learning algorithm. A corpus of 573 radiology reports was de-identified and manually annotated with the support of NLP tools by a physician for relevant concepts, modalities and relations. A machine learning classifier was trained on the dataset interpreted by a physician for diagnosis of deep-vein thrombosis, pulmonary embolism and clinically relevant incidental findings. Decision models accounted for the imbalanced nature of the data and exploited the structure of the reports.


The best model achieved an F measure of 0.98 for pulmonary embolism identification, 1.00 for deep vein thrombosis, and 0.80 for incidental clinically relevant findings. The use of concepts, modalities and relations improved performances in all cases.


This study demonstrates the benefits of developing an automated method to identify medical concepts, modality and relations from radiology reports in French. An end-to-end automatic system for annotation and classification which could be applied to other radiology reports databases would be valuable for epidemiological surveillance, performance monitoring, and accreditation in French hospitals.

Natural language processing; Medical informatics; Embolism and thrombosis/diagnosis; Phlebography; Incidental findings; Human