Fuzzy Logic in Medicine

The presence of intelligent system applications in the medical environment has been undergoing continual growth [45,47] practically since their earliest days. Such is the case of expert systems, which from their appearance, at the end of the 1960s and the start of the 1970s, has had notable influence in the field of medicine. Some of the best known ones are MYCIN [49], dealing with infectious disease, CASNET [31], in the field of ophthalmology, and INTERNIST [39] focused on the vast field of internal medicine.

Along with the population of high field magnetic resonance imaging (MRI), MR angiography (MRA) imaging with no contrast is rapidly gaining acceptance as a versatile noninvasive alternative to the conventional MRA with contrast and the CT angiography (CTA). To construct the volume visualizations of the cerebral blood vessels from volumetric MRA images of the brain, maximum intensity projection (MIP) technique has been widely used by many physicians [1]. The MIP image is created by selecting the maximum value along on an optical ray corresponding to each pixel of the image. The technique and the mutations have some advantages. For example, it gives densitometric information of raw images without any parameters needing to be tuned, and its implementation is relatively simple [1][2]. However, it also contains critical limitations. They are that it cannot depict the spatial relationship of overlapping vessels, and large bright structures may disturb region of interests (ROIs) along on optical rays from both directions. Some studies investigated the advantages and the disadvantages of three visualization techniques, i.e. MIP, volume rendering (VR), and surface shaded display (SSD) [3][4]. They concluded that SSD is useful to evaluating overlapping vessels, and it provides a better definition of the aneurysm neck and the morphology of saccular aneurysms. However, SSD is not used widely today because there is no application to automatically segment the blood vessel region. To construct the SSD images, a user must manually segment the blood vessel

Breast cancer continues to be one of the most deadly diseases among American women, which is the second leading cause of cancer-related mortality among American women. Currently there are more than 50 million women over the age of 40 at risk of breast cancer and approximately 144,000 new cases of breast cancer are expected each year in the United States. One out of eight women will develop breast cancer at some point during her lifetime in this country [1,2]. Because of the high incidence of breast cancer, any improvement in the process of diagnosing the disease may have a significant impact on saving lives and cutting costs in the health care system. Since the cause of breast cancer remains unknown and the earlier stage tumors can be more easily and less expensively treated, early detection is the key to breast cancer control. Mammography has proven to be the most reliable method and the major diagnosis means for detecting and classifying breast cancer in the early stage. Studies have shown a decrease in both severe breast cancer and mortality in women who undergo regular mammographic screens [3].

The measure went of anaesthetic depth during surgical anaesthesia has always been an inexact science where the experience of the anaesthetist is called upon to provide the control of drug administration. The anaesthetist has to maintain the patient at a suitable level of sedation by carefully controlling several anaesthetic drugs so that the surgical procedure can proceed without causing awareness in the patient. There have been many publications on the subject that have shed much light on the subject and which has as a result improved the control of anaesthetic depth.

The anesthetic management of a surgical patient is a process that relies on the experience of an anesthesiologist, since currently there is no direct means of assessing a patient’s level of consciousness during surgery. The decision for the initial anesthetic level is generally made by using the recommended drug dosages based on various patient characteristics, such as age and weight. The anesthesiologist determines any subsequent alteration in the anesthetic level by observing signs from the patient. These signs, the indirect indicators of the depth of anesthesia (DOA), may include changes in blood pressures or heart rate, lacrimation, facial grimacing, muscular movement, spontaneous breathing, diaphoresis, and other signs that may predicate awareness. However, they are not reliable indicators of changes in a patient’s level of consciousness. Although an anesthesiologist can adjust recommended anesthetic dosages based on individual patient characteristics, these adjustments cannot always account for variability in patient responses to anesthesia or changes in anesthetic requirements during the course of surgery.

The essence of modern medicine is a continuous process of decision-making based on the intelligent evaluation of voluminous yet often inconclusive data gathered from patients. In many clinical setups such as intensive care units and epilepsy care units, monitored patients produce a vast amount of biomedical data from online continuous recordings of ECG, EEG, blood pressure, temperature, etc., as well as from X-ray, CT and MRI imaging. In the current state of affairs, there are objective difficulties in processing and interpreting all this data with the aim of extracting the relevant information.

Every day humans are confronted in numerous occasions with tasks that include the management and the processing of information of various degrees of complexity. Regardless of what the actual information consists of, its degree of complexity, or simplicity, can be associated with the number of recognised parts and the extent of their interrelationship (Klir and Folger 1988). The capability to manage such information considerably depends on the actual understanding of the person(s) involved. The more experienced the person the better the understanding and the information management. Further, although different persons may approach the same problem differently a solution is very often based on a combination of different strategies. This paper has a focus on two strategies:

In current high-dependency clinical environments such as Intensive Coronary Care Unit (ICCU hereinafter), operating rooms and so on, the clinical staff is presented with a large mass of data about the patient’s state. These data can be obtained from the advanced biomedical equipment (especially from electrical and hemodynamical monitors), patient’s history, physical examination findings and test results. This massive flow of information can lead to some well-known problems such as data overload and missing data and misinterpretation [1,13]. In order to avoid these kinds of problems, Intelligent Patient Supervision Systems (IPSS hereinafter) have been developed. IPSSs must be developed to support the interpretation of these data and they should provide information in higher abstraction levels in order to improve the decision making process.

Nowadays, in areas such as medicine, many real-world classification problems rely heavily on large collections of data that are not understandable to human users. Therefore, there is a need for transparent models to represent such databases. In this chapter, we present two methods for learning classification rules which aim at being simplistic and transparent in nature. Both methods use fuzzy sets to describe the universes of discourse since their fuzzy boundaries allow a realistic representation of neighbouring concepts. As a consequence, interpolation effects as well as data compression are obtained in the learned models. Moreover, the fuzzy sets can be labelled with words which allows the inferred rules to be interpreted linguistically. In order to generate these rules, probability distributions need to be extracted from fuzzy sets, which is feasible using the fundamental results of mass assignment theory [2].

Association rules are one of the best studied models for knowledge acquisition in the field of Data Mining. Many papers regarding algorithms, measures and related problems can be found in the literature. A brief summary of the main works (to our knowledge) in this area can be found in the references of this paper.