A DICOM viewer with flexible image retrieval to support diagnosis and treatment of scoliosis Juan Miguel Medina ⇑ , Sergio Jaime-Castillo, Esther Jiménez Department of Computer Science and Artificial Intelligence, University of Granada, C/ Periodista Daniel Saucedo Aranda s/n, 18071 Granada, Spain article info Keywords: DICOM viewer Medical CBIR Fuzzy databases abstract This paper presents a medical image viewer implemented in Java whose innovative features are: on the one hand, its capability for visual edition and storage of measurements involved in diagnosis and treatment of scoliosis (a medical condition in which the patient’s spine is curved) and performed on dig- ital X-rays; on the other hand, its capability for retrieving images in a flexible way from medical image databases on the basis of those measurements, which are the standard method for diagnosing this pathol- ogy. Hence, the viewer is intended to be a useful tool for physicians in diagnosis and treatment of scoliosis. Ó 2012 Elsevier Ltd. All rights reserved. 1. Introduction Images are a fundamental tool in health care for diagnosis, clin- ical studies, research and learning. Currently, there are multiple techniques to capture images from patients to help diagnostic tasks such as X-ray images, Computed Tomography (CT), Magnetic Resonance Imaging (MRI), Positron Emission Tomography (PET), and Ultrasonography. The diagnostic task generates a large amount of images which must be archived for future evaluations. Fortu- nately, most of these techniques produce digital images, which are more efficiently archived and handled, by means of computer systems, than physical ones. In medical imaging these computer systems are called Picture Archiving and Communication Systems (PACS). PACS (Huang, 2010) consist of a network of computers devoted to the storage, retrieval, distribution and presentation of images. The main parts of PACS are: PACS servers, that store and send the images, and PACS workstations, that can use local periph- erals for scanning image films into the system, printing them out and interactively displaying digital images. PACS workstations offer means of manipulating the images (cropping, rotating, zoom- ing, windowing and others) through software referred to as DICOM viewers, since the most common format for image storage is DI- COM (Digital Imaging and Communications in Medicine). PACS systems solve the problem of storing digital images but do not pro- vide mechanisms to retrieve them based on the pathology’s infor- mation items they contain. Instead, the single available mechanism is querying the server by patient’s name or patient’s file number. There are many implementations of DICOM viewers: I Do Imaging (n.d.) provides a tool to find free DICOM and PACS applications. NEMA (n.d.) also contains information about the DICOM format and related resources. Content-based image retrieval (CBIR) (Lew, Sebe, Lifl, & Jain, 2006) is the application of computer vision techniques to the prob- lem of digital image search in large databases. In Jaime-Castillo, Medina, and Sánchez (2009) we proposed a CBIR system that allows automatically extracting measures of the spine curvature starting from X-rays of patients with scoliosis. These measures are repre- sented and stored into the Fuzzy Object-Relational Database System (FORDBMS) proposed in Cubero, Marín, Medina, Pons, and Vila (2004) and Barranco, Campaña, and Medina (2008b), which pro- vides a flexible way of querying X-rays on the basis of the spine curve measurements stored into it (Medina, Jaime-Castillo, Barranco, & Campaña, 2009). This paper describes a prototype of DICOM viewer, an impor- tant element of the proposed CBIR system. This viewer provides physicians with a tool to visually edit and measure curves and to perform queries to retrieve DICOM images of patients based on the parameters of the spine deformity. The client DICOM viewer has been developed starting from Tudor DICOM viewer (SANTEC, 2009). Because of this, our viewer also includes all features pro- vided by the Tudor DICOM viewer (see Section 4.1). The rest of the paper is organized as follows. Section 2 intro- duces the idiopathic scoliosis pathology, its diagnosis and how to perform measurements on X-rays to evaluate the spinal deformity. Section 3 describes the most important modules of the CBIR sys- tem to which the viewer belongs. Section 4 describes the viewer, illustrates the use and functioning of the novel features that it pro- vides, and analyzes some aspects that affect the performance of the 0957-4174/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.eswa.2012.02.012 ⇑ Corresponding author. E-mail addresses: medina@decsai.ugr.es (J.M. Medina), sjaime@decsai.ugr.es (S. Jaime-Castillo), estherj@correo.ugr.es (E. Jiménez). Expert Systems with Applications 39 (2012) 8799–8808 Contents lists available at SciVerse ScienceDirect Expert Systems with Applications journal homepage: www.elsevier.com/locate/eswa