Uncertainty modeling for ontology-based mammography annotation with intelligent BI-RADS scoring Hakan Bulu a,n , Adil Alpkocak b , Pinar Balci c a Dokuz Eylul University, Department of Computer Engineering, The Graduate School of Natural and Applied Science, Izmir, Turkey b Dokuz Eylul University, Department of Computer Engineering, Engineering Faculty, Izmir, Turkey c Dokuz Eylul University, Department of Radiology, Medical School, Izmir, Turkey article info Article history: Received 20 July 2011 Accepted 1 January 2013 Keywords: Ontology Semantic Web Bayesian Network Mammography Uncertainty SQWRL abstract This paper presents an ontology-based annotation system and BI-RADS (Breast Imaging Reporting and Data System) score reasoning with Semantic Web technologies in mammography. The annotation system is based on the Mammography Annotation Ontology (MAO) where the BI-RADS score reasoning works. However, ontologies are based on crisp logic and they cannot handle uncertainty. Consequently, we propose a Bayesian-based approach to model uncertainty in mammography ontology and make reasoning possible using BI-RADS scores with SQWRL (Semantic Query-enhanced Web Rule Language). First, we give general information about our system and present details of mammography annotation ontology, its main concepts and relationships. Then, we express uncertainty in mammography and present approaches to handle uncertainty issues. System is evaluated with a manually annotated dataset DEMS (Dokuz Eylul University Mammography Set) and DDSM (Digital Database for Screening Mammography). We give the result of experimentations in terms of accuracy, sensitivity, precision and uncertainty level measures. & 2013 Elsevier Ltd. All rights reserved. 1. Introduction In 1904, Sir William Osler mused that ‘‘Medicine is a science of uncertainty and an art of probability’’. As time passed, the emer- gence of science in medicine has done much in the last century to reduce the uncertainty surrounding of medicine. Apparently, even a simple search in PubMed, using the terms medical and uncer- tainty, results more than a thousand of recent articles. It evidently shows that it is still an active research area and many researchers still work on to reduce uncertainty. On the other hand, evidence- based medicine aims to provide ways to quantify and communicate uncertainty from a probabilistic way. Nevertheless, uncertainty remains in the nature of medicine as in the very famous quote of Sir William Osler. Mammogram is the gold standard for breast cancer screening and early detection. This is important because, breast cancer is the most treatable cancer when it is detected in early phase. Mammograms can help to detect 85–90% of breast cancers, even before they are felt like a lump [1]. Many researchers have been working on computer-aided diagnosis system (CADx) to detect and identify breast masses automatically in digital mammograms over several decades. All these researches aim to support radi- ologists in the difficult task of discriminating benign and malig- nant breast lesions. Hence, it is not surprising that typically only 15–30% of breast biopsies performed on calcifications will be positive for malignancy [2]. To improve the level of CADx in mammography, there is a need to a system taking the background knowledge of radiologist into account in decision-making process with a more computable way. In this point, ontologies can be a solution to improve the performance of CADx systems in mammography. Ontology is the most common way to represent the knowledge for computers, and defined as a formal, explicit specification of a shared conceptualization and encodes a partial view of the world, with respect to a given domain. It is composed of a set of concepts, their definitions and their relations that can be used to describe and reason about a domain. Ontological modeling of knowledge is vital in many real-world applications and in medicine. In intelligent systems, ontologies are way to transform background knowledge of a domain to machine understandable form. For example, the interpretation of radiological examinations includes years of experience, the knowledge on the respective domain. The medical image interpretation is not solely reached by pattern recognition and it also includes a deep knowledge in medical domain. Therefore, a successful implementation of radi- ological imaging system should be able to model and incorporate such knowledge into a more computable format. In this point, Contents lists available at SciVerse ScienceDirect journal homepage: www.elsevier.com/locate/cbm Computers in Biology and Medicine 0010-4825/$ - see front matter & 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.compbiomed.2013.01.001 n Correspondence to. Dokuz Eylul University, The Graduate School of Natural and Applied Science, Tinaztepe Campus, 35160, Buca, Izmir, TURKEY. Tel.: þ90 232 301 7970. E-mail address: hakan.bulu@deu.edu.tr (H. Bulu). Computers in Biology and Medicine 43 (2013) 301–311