PATIENT INFORMATION—ANYTIME ANYWHERE
Image, Signal, and Distributed Data Processing for Networked eHealth Applications A View from the Guest Editors © ARTVILLE, LLC
ILIAS MAGLOGIANNIS, MANOLIS WALLACE, AND KOSTAS KARPOUZIS
C
omputer-based patient record systems are continuously expanding in order to support more clinical activities and serve healthcare professionals more efficiently. For this reason healthcare organizations and providers are asking their staff to interact more often with computer systems during their everyday work. Hospitals and healthcare centers are nowadays sufficiently rich in their computing infrastructure to handle the internal administrative and clinical processes for both inpatients and outpatients. The data stored in computer-based patient record systems include medical documents relating to the past, present, or future physical condition of a patient, the results of examinations in the form of multimedia (text, medical images, sounds, and videos), and financial and demographic information [1], [2]. In addition, the rapid development in communications through fixed or mobile networks has opened new ways in computer-based health systems by providing the capability of remote and distributed access to patient’s medical data [3], [4]. Remote patient monitoring in terms of telemedicine and the provision of clinical guidelines used for the patient's care from distant locations [5] are supported within networked computer-based patient record systems, while retrieval of medical data [6] and remote teleconsultations between healthcare professionals are also possible [7]. Medical data are captured and transmitted, received or updated, stored or retrieved securely and in real time by users in geographically distributed and organizationally independent organizations or distant locations. In this era of distributed computing the trend in medical informatics is toward achieving two goals: the availability of software applications and medical information anywhere and anytime and the invisibility of computing; computing modules are hidden in multimedia information appliances that are used in everyday life [8]. Both aforementioned goals require distributed data processing modules that will be able to automatically analyze data provided by medical devices and sensors, exchange
Digital Object Identifier 10.1109/EMB.2007.901781
14 IEEE ENGINEERING IN MEDICINE AND BIOLOGY MAGAZINE
knowledge, and make decisions in a given context [9]. Natural user interactions with such applications are based on autonomy, avoiding the need for the user to control every action, and adaptivity, so that they are contextualized and personalized, delivering to the medical personnel the right information and decision at the right moment [10]. A typical application falling into the specific concept, for instance, is pervasive patient telemonitoring (an important part of telemedicine), which involves the sensing of a patient’s physiological and physical parameters and transmitting them to a remote location, typically a medical center, where expert medical knowledge resides. A typical telemonitoring system has the ability to record physiological parameters and provide information to the doctor in real time through a wireless connection, while it requires sensors to measure parameters like arterial blood pressure, heart rate, electrocardiogram, skin temperature and respiration, glucose, or patient position and activity. Filtered signals and medical data are either stored locally on a monitoring wearable device for later transmission or directly transmitted, e.g., over the public phone network, to a medical center [11]–[13]. Thus, the field of networked eHealth has clearly already shown its potential, facilitating exchange of information between clinicians or between institutions, reducing costs, extending the scope and reach of medical facilities, enhancing the quality of service offered on- and off-site, providing new means of medical supervision and preemptive medicine, and so on. Currently, the integration of medical networking and medical information systems is treated as an obvious and irrefragable rule. Although major progress has been made with respect to the integration of distributed processing and communication capabilities offered, the field is still considered to be at a premature stage. In addition to interchangeability and data exchange, computer science can also offer intelligent computing services to the medical sector; thus, clinical decision support is a main focus. In this framework, stateof-the-art image, signal, and data processing is required, so that useful information is extracted and reliable automated recommendations are achieved. 0739-5175/07/$25.00©2007IEEE
SEPTEMBER/OCTOBER 2007
In this era of distributed computing the trend in medical informatics is toward achieving two goals: the availability of software applications and medical information anywhere and anytime and the invisibility of computing. In this Issue
Within this special issue we review some of the most promising prospects for the state-of-the-art and emerging field of image, signal, and distributed data processing for networked eHealth applications. The case of pressure ulcer detection is an appropriate example for this. With the number of individuals that belong to high-risk social groups for the development of pressure ulcers, it is quite difficult for a national healthcare system to have sufficient expert medical personnel distributed throughout a country and dedicated to the detection and treatment of pressure ulcers. Kosmopoulos and Tzevelekou [19] go further than existing manual techniques and propose an automated system for the analysis of two-dimensional (2-D) images and classification of the stage of pressure ulcers present in them. Following a support vector machine approach, they achieve high rates of tissue classification to healthy or one of six different stages of pressure ulcers. Such a system can be utilized as a first filtering step, so that the amount of data that has to be manually processed by clinicians is both reduced and labeled for gravity. In the development of such systems, the availability of training data is often an issue. Clinical decision support is most desired in rare medical cases for which a single clinician usually cannot have adequate practical experience. In such cases, analyzing medical history records from a single clinic or medical facility does not typically suffice for the development of a robust and reliable highaccuracy system. Thus, we need to combine data residing at different locations. Additionally, to the ordinary data format and channel bandwidth considerations that govern most data interconnection problems, in the medical case we also need to consider legal and ethical issues related to data ownership and confidentiality. Medical data transfer is also required in order to offer medical services at the time and place they are needed, by making centralized services available electronically through permanent and ad-hoc networks. In both cases, healthcare information needs to be accessible by authorized users only, while its fundamental security properties, namely integrity, availability, and confidentiality, must be retained. The technological challenges for this are presented and discussed by Gritzalis et al. in [18]. Vastly augmenting communication bandwidths and rapidly diminishing costs for digital storage have allowed the realization of the above-mentioned integrated systems that handle or exchange large amounts of medical data. Following this trend, the acquisition, digitization, storage, remote access to, and analysis of large amounts of medical data are becoming typical things in the field of IEEE ENGINEERING IN MEDICINE AND BIOLOGY MAGAZINE
eHealth. Although the results of this progress are certainly beneficial and promising, it is becoming obvious that some important technological factors may have been overlooked in the initial enthusiasm. Specifically, we may have originally underestimated the way in which the sudden availability of cheap and trustworthy storage and communication solutions would affect the amount of medical data retained in digital format. More and more medical data are becoming available in digital formats and to greater details. Although this provides more processable information and thus enhanced computing opportunities, it also augments the requirements in storage and bandwidth, often overweighting the rise in available resources. Clearly, we desire to have techniques that will limit the size of the acquired data but not the information richness, thus enhancing the balance between available and required resources for optimized medical services. Doukas and Maglogiannis in [15] provide an excellent review of such techniques in the field of medical image compression, the medical data that consumes the vast majority of available resources. Region of interest coding is analyzed and compared and classified in 2-D, three-dimensional (3-D), and video sequence categories. In all of the above we have assumed that distributed medical repositories cooperate fully and, thus, that the only considerations for data exchange are related to data format, channel capacity, and security. Unfortunately, this also constitutes an oversimplification. In reality, in most cases individual medical archives are bound by legal constraints not to freely redistribute the medical history records they possess; ownership and cost issues may also interfere. As a result, owners of medical information are often quite reluctant to permit its transmission to other sites so that it can be processed remotely. One of the main fields in which electronic healthcare has been making a major contribution to clinical medicine is clinical decision support. In this framework, medical history data is processed automatically using state-of-theart algorithms in order to develop automated recommendation systems. With today’s augmented computing power resources, using the best available processing algorithms is not an issue. What is an issue, on the other hand, is the availability of medical history data due to the fact that issues related to security, ownership, and confidential personal medical data prohibit the gathering of this data to a main medical database. Given the statistical nature of utilized algorithms, the independent processing of the data and the combination of the individual results cannot provide overall results of equivalent quality to those of centralized processing of all SEPTEMBER/OCTOBER 2007
15
Although major progress has been made with respect to the integration of distributed processing and communication capabilities offered, the field is still considered to be at a premature stage.
the data. Megalooikonomou and Kontos in [17] propose a novel hierarchical processing algorithm in order to tackle this. In their approach, although data are only processed at their original location so that confidentiality issues do not arise, the distributed processing components are in constant communication with each other throughout the procedure, thus achieving a coordination that allows for results equivalent to those acquired when applying centralized processing algorithms. With methodologies such as the above we can generate automated systems able to process medical information and provide decision support recommendations to clinicians; this might not be the only field of application for distributed eHealth. It is becoming more and more obvious that healthcare systems cannot possibly offer full coverage for the geographic and population spread of a country. On the other hand, we now know that asymptomatic diseases are difficult to detect at an early stage without comprehensive medical examinations; without an early detection the chances for treatment are severely narrowed. The only realistic solution is to identify high-risk population groups so that medical advice and monitoring can be targeted. In [14] Kyriacou et al. see the utilization of automated analysis techniques in the field of medical risk analysis. With stroke being the leading cause of death in Western societies, and with the process building up to it often being free of symptoms, the identification of high-risk individuals or groups is desired. Such information can help greatly reduce strokes and related deaths or resulting disabilities. In the work presented in this article, in order to acquire this type of knowledge, medical data have been collected from a high number of patients using noninvasive techniques; this data was then analyzed with novel automated processing techniques in order to extract information-rich features. The patients were then monitored for a period reaching up to 84 months. The study of the statistical correlation between the data collected from patients and their latter health status and progress is utilized in the developed medical portal in order to automatically provide estimations of whether specific individuals or groups belong to high-risk categories. With both population and diversity of illnesses augmenting rapidly, it is not possible to provide constant and detailed medical analysis for all people and in all cases, so works such as this that utilize computerized technology to identify high-risk groups provide a much more realistic approach to preemptive medicine and early detection. Once someone has been identified as a member of a high-risk group, we need to be able to provide personalized advice and pervasive monitoring. Ideally, a personal or family clinician would undertake the task. 16 IEEE ENGINEERING IN MEDICINE AND BIOLOGY MAGAZINE
Unfortunately, the amount and diversity of high-risk groups and individuals makes this, too, an unrealistic option. Additionally, few people would be content with this extent of intervention in their day-to-day life. Pervasive, automated, citizen-centered care is probably the most important emerging technological achievement in the scope of distributed eHealth. In this framework, automated digital systems monitor an individual belonging to a high-risk group, providing both personalized lifestyle-related advice and early diagnosis in the individual’s own environment with little or no interference with everyday life. This kind of personalized care was not available in the past outside the clinic and without the support of clinicians and specialized technicians due to the cost, size, and complexity of the required equipment. Recent advances in electronics have made possible the production of miniscule medical observation and processing devices that come in many flavors, with portable and wearable ones being the most promising. Out of these, the wearable type, being the noninvasive one, is the one receiving most attention and providing the best prospects for the future. Gatzoulis and Iakovidis in [16] provide a rich presentation of the state of the art, emerging innovations, and new opportunities in the field. Acknowledgments
The guest editors wish to thank all the authors for their contributions, all reviewers for their efforts and valuable reviews, and especially Editor-in-Chief John Enderle for his advice and encouragement. Although the topics discussed in this special issue could not certainly cover the whole spectrum of distributed eHealth applications, we do hope that the articles included will attract the interest of this publication’s audience and serve as an inspiration to the young researchers working in this field. Ilias Maglogiannis received a Diploma in electrical and computer engineering and a Ph.D. in biomedical engineering and medical informatics from the National Technical University of Athens (NTUA), Greece, in 1996 and 2000, respectively, with a scholarship from the Greek government. From 1996 until 2000 he worked as a researcher in the Biomedical Engineering Laboratory at NTUA, and he has been involved with several European and national projects. In 2001 he joined the faculty of the Department of Information and Communication Systems Engineering at the University of the Aegean. His published scientific work includes two books and five lecture notes (in Greek) on biomedical engineering and artificial intelligence SEPTEMBER/OCTOBER 2007
topics, 26 journal papers, and more than 40 international conference papers. He has served on program and organizing committees of national and international conferences and he is a reviewer for several scientific journals His scientific activities include biomedical engineering, telemedicine and medical informatics, image processing, and multimedia telecommunications. He is a member of the IEEE Engineering in Medicine and Biology Society, IEEE Computer Society, IEEE Communications Society, the International Society for Optical Engineering, the ACM, the Technical Chamber of Greece, the Greek Computer Society, the Hellenic AI Society, and the Hellenic Organization of Biomedical Engineering.
Address for Correspondence: I. Maglogiannis, Dept. of Information & Communication Systems Engineering, University of the Aegean, 83200 Karlovasi, Greece Tel: +30 22730 82239, Fax: +30 22730 82008. E-mail:
[email protected]. References [1] I. Maglogiannis, N. Apostolopoulos, P. Tsoukias, “Designing and implementing an electronic health record for personal digital assistants (PDAs),” Int. J. Qual. Life Res., vol.2, no. 1, pp. 63–67, 2004. [2] A. Dalley, J. Fulcher, D. Bomba, K. Lynch, and P. Feltham, “A technological model to define access to electronic clinical records,” IEEE Trans. Inform. Technol. Biomed., vol.9, no.2, pp. 289–290, 2005. [3] E. Mendonça, E. Chen, P D. Stetson, L. McKnight, J. Lei, and J. Cimino,
Manolis Wallace was born in Athens in 1977. He obtained his Diploma in Electrical and Computer Engineering from the National Technical University of Athens (NTUA) and his Ph.D from the Computer Science Division of NTUA. He is with the University of Indianapolis, Athens Campus (UIA) since 2001 where he serves now as an Assistant Professor. Since 2004 he is also the Chair of the Department of Computer Science of UIA. His main research interests include handling of uncertainty, information systems, data mining, personalization and applications of technology in education. He has published more than 40 papers in the above fields, ten of which in international journals. He is the guest editor of two journal special issues and a reviewer for more than ten journals and various books. His academic volunteering work also includes participation in various conferences as organizing committee chair, session organizer, session chair or program committee member. He is the general chair of SMAP 2006.
“Approach to mobile information and communication for health care,” Int. J. Med. Inform., vol. 73, no. 7-8, pp. 631–638, 2004. [4] E. Hall, D.K. Vawdrey, C.D. Knutson, and J.K. Archibals, “Enabling remote access to personal electronic medical records,” IEEE Eng. Med. Biol. Mag., vol. 22, no. 3, pp. 133–139 2003. [5] C. Finch, “Mobile computing in healthcare,” Health Manage.Technol., vol. 20, no. 3, pp. 63–64, 1999. [6] G. Kambourakis, I. Maglogiannis, and A. Rouskas, “PKI-based secure mobile access to electronic health services and data,” Technol. Healthcare, vol. 13, no. 6, pp. 511–526, 2005. [7] I. Maglogiannis, K. Delakouridis, and L. Kazatzopoulos, “Enabling collaborative medical diagnosis over the Internet via peer to peer distribution of electronic health records,” J. Med. Syst., vol. 30, no. 2, pp. 107–116 2006. [8] G. Abowd, “Software engineering issues for ubiquitous computing,” in Proc. Int. Conf. Software Engineering, Los Angeles, 1999, pp. 5–84. [9] G.E. Barnes and S. Warren, “A wearable, Bluetooth-enabled system for home health care,” in Proc. 2nd Joint EMBS/BMES Conf., Houston, TX, Oct. 2002, pp. 1879–1880. [10] E. Jovanov, “Stress monitoring using a distributed wireless intelligent sensor system,” IEEE Eng. Med. Biol. Mag., vol. 22, no. 3, pp. 49–55, 2003. [11] V. Rialle, J. Lamy, N. Noury, and L. Bajolle, “Telemonitoring of patients at home: A software agent approach,” Comput. Methods Programs Biomed.,
Kostas Karpouzis is an associate researcher at the Institute of Communication and Computer Systems (ICCS) and holds an adjunct lecturer position at the University of Piraeus, teaching data warehousing and data mining. He graduated from the School of Electrical and Computer Engineering of the National Technical University of Athens in 1998 and received his Ph.D. degree in 2001 from the same University. His current research interests lie in the areas of human computer interaction, image and video processing, image interchange infrastructures using the MPEG-4 and MPEG-21 standards, sign language synthesis, and virtual reality. Dr. Karpouzis has published more than 70 papers in international journals and proceedings of international conferences. He is a member of the technical committee of the International Conference on Image Processing (ICIP), the IFIP Conference on Artificial Intelligence Applications and Innovations (AIAI) and a reviewer in many international journals. Since 1995 he has participated in more than 12 R&D projects at Greek and European level. He is also a national representative in IFIP Working Groups 12.5 “Artificial Intelligence Applications” and 3.2 “Informatics and ICT in Higher Education.” IEEE ENGINEERING IN MEDICINE AND BIOLOGY MAGAZINE
vol. 72, no. 3, pp. 257–268, 2003. [12] K. Hung and Y. Zhang, “Implementation of a WAP-based telemedicine system for patient monitoring,” IEEE Trans. Inform. Technol. Biomed., vol. 7, no. 2, pp. 101–107, 2003. [13] J. Andreasson et al., “Remote system for patient monitoring using Bluetooth/spl trade” Sensors 2002, Proc. IEEE, vol. 1, pp. 304–307, June 2002. [14] E. Kyriacou, C.S. Pattichis, M. Karaolis, C. Loizou, C. Christodoulou, M.S. Pattichis, S. Kakkos, and A. Nicolaides, “An Integrated system for assessing stroke risk,” IEEE Eng. Med. Biol. Mag., vol. 26, no. 5, pp. 43–50, 2007. [15] C. Doukas and I. Maglogiannis, “Region of interest coding techniques for medical image compression,” IEEE Eng. Med. Biol. Mag., vol. 26, no. 5, pp. 29–35, 2007. [16] L. Gatzoulis and I. Iakovidis, “Wearable and portable eHealth systems,” IEEE Eng. Med. Biol. Mag., vol. 26, no. 5, pp. 51–56, 2007. [17] V. Megalooikonomou and D. Kontos, “Medical data fusion for telemedicine: A model for distributed analysis of medical image data across clinical information repositories,” IEEE Eng. Med. Biol. Mag., vol. 26, no. 5, pp. 36–42, 2007. [18] S. Gritzalis, P. Belsis, and S.K. Katsikas, “Interconnecting autonomous medical domains,” IEEE Eng. Med. Biol. Mag., vol. 26, no. 5, pp. 23–28, 2007. [19] D. Kosmopoulos and F. Tzevelekou, “Automated pressure ulcer lesion diagnosis for telemedicine systems,” IEEE Eng. Med. Biol. Mag., vol. 26, no. 5, pp. 18–22, 2007.
SEPTEMBER/OCTOBER 2007
17
