THE CARDIOLAB PROJECT

 


Département d’Information Médicale, Université de Rennes I

Objectives

The CARDIOLAB project is dedicated to the development of Computer Assisted Instruction (CAI) and computer aided diagnosis systems in the field of cardiology. CARDIOLAB is a framework that incorporates different kinds computational models of the heart’s electrical activity. The goal is to provide users with an environment capable of predicting and explaining rhythmic disorders. By prediction, we mean computing the heart’s electrical activity when modeled physiological parameter values are given as initial conditions. By explanation, we mean the task of deriving the possible causes of some observed data. Because of these two aspects, CARDIOLAB’s field of application also includes basic, clinical and pharmacological research, and ultimately, intelligent monitoring system design.

Method

CARDIOLAB’s first role is to simulate cardiac electrical activity. A central concern in simulation studies is the adequation of a designed model with respect to its intended goal. Models of cardiac electrical activity may differ in complexity, level of description and representation. Depending on the desired level of detail, analytical, cellular automatas and qualitative models can be used. Their advantages and shortcomings can be summarized in terms space and time complexities, ease of interpretation, and clinical relevance. The framework incorporates all three kinds of models, allowing a spatio-temporal multi-scale approach for studying cardiac electrical activity. Since the clinical context is of prime importance, factors relevant for studying arrhythmias and ischemias are taken into account. They include the durations of cell phases that characterize impulse formation and conduction (e.g., slow-diastolic depolarization), and adaptative properties such as the rate-dependent repolarization durations, or the recovery-state dependency of conduction speed. Transmembrane potentials are also computed for simulating the ECG.

CARDIOLAB’s second role is to explain and diagnose. Explaining an ECG requires qualification and reasoning capabilities. This presupposes that explicit descriptions of simulated processes can be derived from the appropriate representations and the corresponding inference engines. Heart functions and components are made explicit in the framework’s qualitative model. On the other hand, quantitative analysis of rhythmic disorders may require detailed numerical models. For instance, re-entry in ischemic tissue or PVB-induced ventricular fibrillation can only be reproduced by models with a great number of individual elements. In this respect, the system incorporates 2D and 3D cellular automata models composed of thousands of elements.

To summarize, CARDIOLAB is an intelligent framework where both qualification and quantification interact to simulate, and explain arrhythmias in a multi-scale approach. Because of these capabilities, it could be a part of futur developments in the design of knowledge-based CCU monitoring systems.


Plates


Back to LIM.