Catálogo de publicaciones - libros

In most case-based reasoning (CBR) systems there has been little research done on validating new knowledge, specifically on how previous knowledge differs from current knowledge by means of conceptual change. This paper proposes a technique that enables the domain expert who is non-expert in artificial intelligence (AI) to interactively supervise the knowledge validation process in a CBR system. The technique is based on formal concept analysis which involves a graphical representation and comparison of the concepts, and a summary description highlighting the conceptual differences. We propose a dissimilarity metric for measuring the degree of variation between the previous and current concepts when a new case is added to the knowledge base. The developed technique has been evaluated by a dermatology consultant, and has shown to be useful for discovering ambiguous cases and keeping the database consistent.

So far, Case-Based Reasoning has not become as successful in medicine as in some other application domains. One, probably the main reason is the adaptation problem. In Case-Based Reasoning the adaptation task still is domain dependent und usually requires specific adaptation rules. Furthermore, in medicine adaptation is often more difficult than in other domains, because usually more and complex features have to be considered. We have developed some programs for endocrine therapy support, especially for hypothyroidism. In this paper, we do not present them in detail, but focus on adaptation. We do not only summarise experiences with adaptation in medicine, but we want to elaborate typical medical adaptation problems and hope to indicate possibilities how to solve them.

The paper proposes a methodology based on a neural network to process the signals related to the hands movements in response to the Transcranial Magnetic Stimulation (TMS) in order to diagnose the pathology and evaluate the treatment of the patients affected by demency diseases. First a time-frequency analysis of such signals is carried out to identify the main signal variables that characterize the demency diseases. Then these variables are processed by a neural network in order to classify the responses into four classes: healthy subjects, people affected by Subcortical Ischemic Vascular Dementia (SIVD) and/or Alzheimer. A comparison between the proposed method and a fuzzy approach previously developed by the authors is presented.

The paper deals with the integrated use of Information Visualization techniques and clustering algorithms to analyze Magnetic Resonance Imaging (MRI) data sets. The paper also describes the criteria we followed in designing and implementing the prototype, according to the above approach. Finally, some preliminary results are given for the considered medical application.

In present clinical practice, information about the heart electrical activity is routinely gathered through ECG’s, which record electrical potential from just nine sites on the body surface. However, thanks to the latest technological advances, body surface potential maps are becoming available, as well as epicardial maps obtained noninvasively from body surface data through mathematical model-based reconstruction methods. Such maps can capture a number of electrical conduction pathologies that can be missed by ECG’s analysis. But, their interpretation requires skills that are possessed by very few experts. The Spatial Aggregation (SA) approach can play a crucial role in the identification of patterns and salient features in the map, and in the long-term goal of delivering an automated map interpretation tool to be used in a clinical context. In this paper, the focus is on epicardial activation isochrone maps. The salient features that characterize the heart electrical activity, and visually correspond to specific geometric patterns, are defined, extracted from the epicardial electrical data, and finally made available in an interpretable form within a SA-based framework.

Cephalometric analysis is a time consuming measurement process by which experienced orthodontist identify on lateral craniofacial X-rays landmarks that are needed for diagnosis and treatment planning and evaluation. High speed and accuracy in detection of craniofacial landmarks are widely demanded. A prototyped system, which is based on CNNs (Cellular Neural Networks) is proposed as an efficient technique for landmarks detection. The first stage of system evaluation assessed the image output of the CNN, to verify that it included and properly highlighted the sought landmark. The second stage evaluated performance of the developed algorithms for 8 landmarks. Compared with the other methods proposed in the literature, the findings are particularly remarkable with respect to the accuracy obtained. Another advantage of a CNN based system is that the method can either be implemented via software, or directly embedded in the hardware, for real-time performance.

Sketching is ubiquitous in medicine. Physicians commonly use sketches as part of their note taking in patient records and to help convey diagnoses and treatments to patients. Medical students frequently use sketches to help them think through clinical problems in individual and group problem solving. Applications ranging from automated patient records to medical education software could benefit greatly from the richer and more natural interfaces that would be enabled by the ability to understand sketches. In this paper we take the first steps toward developing a system that can understand anatomical sketches. Understanding an anatomical sketch requires the ability to recognize what anatomical structure has been sketched and from what view (e.g. parietal view of the brain), as well as to identify the anatomical parts and their locations in the sketch (e.g. parts of the brain), even if they have not been explicitly drawn. We present novel algorithms for sketch recognition and for part identification. We evaluate the accuracy of the recognition algorithm on sketches obtained from medical students. We evaluate the part identification algorithm by comparing its results to the judgment of an experienced physician.

This paper presents an effective representation scheme for the statistical shape analysis of the hippocampal structure and its shape classification: Morphometry of the hippocampus. The deformable model based on FEM (Finite Element Method) and ICP (Iterative Closest Point) algorithm allows us to represent parametric surfaces and to normalize multi-resolution shapes. Such deformable surfaces and 3D skeletons extracted from the voxel representations are stored in the Octree data structure. And, it will be used for the hierarchical shape analysis. We have trained SVM (Support Vector Machine) for classifying between the control and patient groups. Results suggest that the presented representation scheme provides various level of shape representation and SVM can be a useful classifier in analyzing the statistical shape of the hippocampus.

Bone scintigraphy or whole-body bone scan is one of the most common diagnostic procedures in nuclear medicine used in the last 25 years. Pathological conditions, technically poor quality images and artifacts necessitate that algorithms use sufficient background knowledge of anatomy and spatial relations of bones in order to work satisfactorily. We present a robust knowledge based methodology for detecting reference points of the main skeletal regions that simultaneously processes anterior and posterior whole-body bone scintigrams. Expert knowledge is represented as a set of parameterized rules which are used to support standard image processing algorithms. Our study includes 467 consecutive, non-selected scintigrams, which is to our knowledge the largest number of images ever used in such studies. Automatic analysis of whole-body bone scans using our knowledge based segmentation algorithm gives more accurate and reliable results than previous studies. Obtained reference points are used for automatic segmentation of the skeleton, which is used for automatic (machine learning) or manual (expert physicians) diagnostics. Preliminary experiments show that an expert system based on machine learning closely mimics the results of expert physicians.

Though multi-agent systems have been explored in a wide variety of medical settings, their role at the primary care level has been relatively little investigated. In this paper, we present a system that is currently being piloted for future rollout in Scotland that employs an industrial strength multi-agent platform to tackle both technical and sociological challenges within primary care. In particular, the work is motivated by several specific issues: (i) the need to widen mechanisms for access to primary care; (ii) the need to harness technical solutions to reduce load not only for general practitioners, but also for practice nurses and administrators; (iii) the need to design and deploy technical solutions in such a way that they fit in to existing professional activity, rather than demanding changes in current practice. With direct representation of individuals in health care relationships implemented in a multi-agent system (with one multi-functional agents representing each patient, doctor, nurse, pharmacist, etc.) it becomes straightforward first to model and then to integrate with existing practice. It is for this reason that the system described here successfully widens access for patients (by opening up novel communication channels of email and SMS texting) and reduces load on the practice (by streamlining communications and semi-automating appointment arrangement). It does this by ensuring that the solution is not imposed on, but rather, integrated with what currently goes on in primary care. Furthermore, with agents responsible for maintaining audit trails for the patients they represent, it becomes possible to see elements of the electronic patient record (EPR) emerging under agent control. This EPR can be extended through structured interaction with the practice system (here, we examine the GPASS system, the market leader in Scotland), to allow rich agent-agent and agent-human interactions. By using multi-agent design and implementation techniques, we have been able to build a solution that integrates both with individuals and extant software to successfully tackle real problems in primary care.