IT Infrastructure to Support the Secondary Use of

Authors: Kai Yan Eugene Leung Fedde van der Lijn Henri A Vrooman Miriam C J M Sturkenboom Wiro J Niessen

Publish Date: 2014/08/17

Volume: 13, Issue: 1, Pages: 65-81

PDF Link

Abstract

We propose an infrastructure for the automated anonymization extraction and processing of image data stored in clinical data repositories to make routinely acquired imaging data available for research purposes The automated system which was tested in the context of analyzing routinely acquired MR brain imaging data consists of four modules subject selection using PACS query anonymization of privacy sensitive information and removal of facial features quality assurance on DICOM header and image information and quantitative imaging biomarker extraction In total 1616 examinations were selected based on the following MRI scanning protocols dementia protocol 246 multiple sclerosis protocol 446 and open question protocol 924 We evaluated the effectiveness of the infrastructure in accessing and successfully extracting biomarkers from routinely acquired clinical imaging data To examine the validity we compared brain volumes between patient groups with positive and negative diagnosis according to the patient reports Overall success rates of image data retrieval and automatic processing were 825  823  and 662  for the three protocol groups respectively indicating that a large percentage of routinely acquired clinical imaging data can be used for brain volumetry research despite image heterogeneity In line with the literature brain volumes were found to be significantly smaller pvalue 0001 in patients with a positive diagnosis of dementia 915 ml compared to patients with a negative diagnosis 939 ml This study demonstrates that quantitative image biomarkers such as intracranial and brain volume can be extracted from routinely acquired clinical imaging data This enables secondary use of clinical images for research into quantitative biomarkers at a hitherto unprecedented scaleIn the last decades the use of medical imaging in routine clinical practice has increased both in quantity and diversity As advances in imaging hard and software and in imaging tracers uncover new ways to visualize disease processes medical imaging will continue to fulfill an important role in the diagnosis and treatment of patients in routine clinical care The inclusion of imaging protocols in the diagnostic guidelines and criteria is a testimony of this importance In the field of brain imaging for example imaging has become critical for the diagnosis and/or operative planning for the treatment of acute traumas tumors and diseases such as epilepsy and multiple sclerosis to name only a few More recently imaging has become supportive for the diagnosis of Alzheimer’s disease Jack et al 2011 Frisoni et al 2010 Owing to these developments the amount of imaging data stored in Picture Archiving and Communication System PACS databases across hospitals continues to growTraditionally the interpretation of medical imaging data in clinical practice is performed qualitatively by trained radiologists However in clinical research and population studies image information is increasingly condensed into a set of quantitative measures This research is aimed at the development of ‘quantitative imaging biomarkers’ which objectively can determine the presence and stage of a disease or the response to a treatment For example the degree of carotid artery stenosis has been shown to be a quantitative imaging biomarker for stroke NASCET 1991 ECST 1998 The Agatston score quantifies the amount of coronary calcification to predict the presence of obstructive coronary artery disease Agatston et al 1990 Kondos et al 2003 Detrano et al 2008 Also a low hippocampal volume in magnetic resonance images has been used as quantitative imaging biomarker for dementia den Heijer et al 2006 Bobinski et al 2000 Jack et al 1992 Furthermore computer tools are being developed to automate these measurements Adame et al 2004 Tang et al 2012 Isgum et al 2007 Shahzad et al 2013 Fischl et al 2002 van der Lijn et al 2008Following both the trends of the increasing quantity of imaging data in routine clinical practice and the rising importance of quantitative imaging biomarkers in research we see the natural opportunity for the secondary use of routinely acquired clinical imaging data to support medical imaging research on hitherto unprecedented scales The principle of using legacy medical data for research has eg been successfully applied using electronic patient records from hospital databases wwwi2b2org or general practitioners databases wwwipcinlIn recent years the interest in using legacy imaging data for systematic medical knowledge discovery has increased For example information technology systems for unlocking clinical imaging data for research have been proposed based on the Biomedical Informatics Research Network BIRN Chervenak et al 2012 BIRN 2013 Extensible Neuroimaging Archive Toolkit XNAT Doran et al 2012 Marcus et al 2007 2011 Informatics for Integrating Biology the Bedside I2B2 Mi2b2 2012 Murphy et al 2010 or webbased infrastructures Baltasar Sánchez and GonzálezSistal 2011 Bland et al 2007 Furthermore the feasibility of using these infrastructures for quantitative imaging research has been investigated In the work by Hoogenboom et al 2012 2013 the feasibility of using biomarkers extracted from legacy imaging data has been investigated with brain morphology markers from structural magnetic resonance imaging MRI and white matter microstructure markers from diffusion tensor imaging DTI Also FennemaNotestine et al 2007 demonstrated the feasibility of pooling legacy multicenter data into one analysis to investigate hippocampal changes in normal aging In principle a vast amount of information is contained in imaging data that are acquired in routine clinical care but this information is not often used for research The main advantages of using legacy data are the avoidance of additional acquisition costs and the potential to increase the scale of research This could for example lead to the ability to detect subtle effects that may otherwise remain hidden and to capture rare cases Moreover it would allow for subdivisions within groups However there are serious challenges logistical challenges related to secure data access retrieval and anonymization and data analysis challenges owing to heterogeneity in imaging data due to differences in scanner and acquisition protocolsIn view of this the aim of this work is 1 to design and implement a system which can extract imaging data from clinical repositories and process them such that they are available for research purposes and 2 to show the feasibility to apply automated image processing to these data for supporting quantitative imaging biomarker research on routinely acquired clinical imaging data The system should satisfy the following requirementsAnonymization The first priority is to ensure the confidentiality of sensitive health care data Secondary use of health care data requires consent unless this is infeasible If requesting consent is infeasible data can be used for research if they are anonymized and proper safeguards are in place In line with Directive 95/46/EC the law for Protection of Personal data Dutch Wet Bescherming Persoonsgegevens the Health Insurance Portability and Accountability Act and the World Medical Association Declaration of Helsinki Gezondheidsraad 1993 KNAW 2003 HIPAA 2013 WMA 2002 proper deidentification mechanisms must be constructed and approved Moreover the infrastructure must be secure and guarded by a trusted and independent third party privacy officer This privacy officer provides the contact for patients with respect to their right to privacy and withdrawalCope with clinical workflow It is of utmost importance that the system does not interfere with the clinical workflow The current infrastructure to access clinical images stored in hospital PACS databases is not suited for large scale image retrieval and processing Available retrieval mechanisms could severely interfere with clinical workflow by overloading available resources The system must also be able to cope with variation in scanning protocols logistical issues and patient diversity This requires a system which is robust to the input data and includes proper quality assurance algorithms

Keywords: