MEDICO – Intelligent searches of medical databases
Imaging procedures, such as ultrasound and computed tomography (CT), are a critical part of medical diagnosis and therapy, as they make it possible to detect illnesses at an early stage and initiate the right treatment. Up to now, however, no single instrument has been able to intelligently structure all relevant information – written materials, images, laboratory data – and make it accessible. Researchers involved in the MEDICO application are working to close this gap.
How can a computer learn to interpret images and recognize similarities, as well as to relate them to text-based content? These are the questions with which the THESEUS application MEDICO is concerned. Using semantic technologies, the participating researchers are developing applications for simple, cross-linked searches of medical databases. The aim is to facilitate the work of physicians and other healthcare workers by intelligently compiling all of the relevant information on patients that is derived from image- and text-based findings.
New possibilities for diagnosis and treatment
MEDICO is developing procedures that not only recognize anatomical structures, such as bones, blood vessels and organs, but also automatically classify data and combine reference images and treatment reports from various different databases, making it possible to identify pathological changes. The benefits of MEDICO can be seen, for example, in the diagnosis and treatment of lymphomas: A lymphoma patient is being treated at a local hospital. The attending physician wants to know whether chemotherapy is working. The algorithms developed by MEDICO researchers automatically compare current CT images with those taken prior to treatment, yielding the relevant information: Chemotherapy appears to be working, since the lymph nodes have not increased in size. However, the program detects damage to the spleen, raising new questions. The attending physician has no comparable cases on site that might provide useful information. To clarify the patient’s situation, he takes advantage of another feature of MEDICO that compares his patient’s data with information on numerous lymphoma cases treated at a university medical center. This much larger set of data provides the attending oncologist with specific recommendations for further tests and treatment. Using the THESEUS application ORDO, the physician is also able to conduct a quick search of current scientific literature that is relevant to the illness of this particular lymphoma patient. The most current research results help to determine the optimal approach to treatment.
Support for physicians
The intelligent interpretation and retrieval of a wide variety of images and texts offer a multitude of opportunities for the medical sector. Attending physicians are provided with an efficient and precise tool to help them reach decisions. Patients also benefit, since their physicians are able to base their diagnostic and therapeutic decisions on a wide variety of experience. Moreover, MEDICO will save hospitals time and money by avoiding tests that are unnecessary and often difficult for patients.
The researchers involved in MEDICO are initially concentrating on semantic searches and structuring the information contained in medical image databases. However, the potential applications of MEDICO go much further. As soon as ethical and data protection issues have been resolved, epidemiological researchers and the pharmaceutical industry alike will benefit from quick, efficient access to relevant studies, which can further improve the quality of diagnosis and treatment.
Structuring data appropriately
The purpose of the technologies developed through MEDICO is to enable computers to generate information on their own, using the contents of the images displayed; to combine that information with other medical data; and to structure the results to provide recommendations for the attending physician. This requires the development of numerous components, for example pattern recognition methods and clinical decision-making tools. The central task, however, is to model an appropriate ontology – a system for organizing and cross-linking copious amounts of information, based on medical terminology. Here MEDICO can build on ontologies that have already been developed by medical experts, such as “RadLex,” which characterizes medical images, or the very detailed “Foundational Model of Anatomy,” which – as the name indicates – describes human anatomy.
As methods of diagnosis and treatment improve, physicians will have access to more and more information about patients and their illnesses. MEDICO quickly and easily interprets, links and compares medical information in image and text form, making an important contribution to medical progress.
Visit MEDICO as well on youtube and twitter:
