Article Preview
TopIntroduction
The enormous amount of information that is acquired, stored and managed within heterogeneous healthcare environments, supported by complex healthcare information networks, is increasingly used to continuous improvement of healthcare provision (Asche, Seal, Kahler, Oehrlein, & Baumgartner, 2017).
The constant technological evolution has given rise to automatic tools to access, process and analyse clinical information. In particular, concerning medical imaging, these tools allow remote and safe analyses of clinical information related to millions of people, comparisons of clinical effectiveness, efficiency, and efficacy, dissemination of good practices, and, therefore, the improvement of the quality of healthcare provision (DeVoe et al., 2011; Klaiman, Pracilio, Kimberly, Cecil, & Legnini, 2013). In this context, advanced processing methods (e.g. artificial neural networks) might optimise the analysis of large amount of healthcare data (Moreira, Rodrigues, Kumar, Al-Muhtadi, & Korotaev, 2017), which might be useful to promote the quality of healthcare provision (Gholipour, Rahim, Fakhree, & Ziapour, 2015).
The population aging and, consequently, the increasing prevalence of chronic conditions, among other factors (Van Oostrom et al., 2016), including technological developments, defensive medicine, and the need to perform a large number of medical imaging studies and to achieve results in short periods (Amis Jr. et al., 2007), requires the improvement of the medical imaging procedures, which is reinforced by the fact that many of these procedures involve the use of ionizing radiation (Thrall, 2009; Sachs & Long, 2016).
In addition to the optimization of exposure factors to increase the performance of the medical imaging practice, another aspect that should be considered is the patients’ exposure to X-radiation, namely the number of medical imaging studies conducted during their lifetime (Fabritius et al., 2016). Therefore, initiatives should be taken to identify the causes of duplicated and repeated radiation exposures within medical imaging studies (Nol, Isouard, & Mirecki, 2006; Prieto et al., 2009; Hendee et al., 2010; Yuan et al., 2013), namely taking advantage of the information that is part of the Digital Imaging and Communication in Medicine (DICOM) metadata (Tsalafoutas & Metallidis, 2011; Brady & Kaufman, 2015).
The DICOM metadata data produced by different medical imaging modalities and distinct equipment manufacturers can be used in multiple scenarios, namely in efficiency metrics for imaging device productivity (Hu et al., 2011), or to provide data for different initiatives such as, for example, development of medical imaging performance indicators (Prieto et al., 2009); Jahnen, Kohler, Hermen, Tack, & Back, 2011; Dave & Gingold, 2013) or biomedical research (Freymann, Kirby, Perry, Clunie, & Jaffe, 2012).
Important parameters such as image processing and image acquisition parameters, patient dose and geometric information are stored in the Picture Archiving and Communication Systems (PACS). Despite these data being stored on the healthcare repositories, the traditional information systems are not able to access them. These difficulties can be overcome using the Dicoogle (Valente, Silva, Godinho, & Costa, 2016). Dicoogle is an open source solution that allows extracting textual information from the PACS and performing flexible queries over DICOM metadata. The flexibility to index different Information Object Definition (IOD) (i.e. DICOM objects types with different data elements) is due to the replacement the traditional database by an indexation engine. Following this approach, it is possible to index all existent DICOM data elements (text-based) without need to create new fields, new tables, and new relations that would be necessary for a non-proprietary relational database approach (Santos et al., 2015).