A System for Real-Time Signal Processing in Cardiovascular Modeling

	ABOUT THE AUTHOR The Autors' Department was created in 1993 by the former Cardiovascular Engineering Group existing since early 80s. Research activity is presently devoted to the study of cardiac output regulation and the interaction between the ventricle and circulatory network. Details on research activity are reported at www.cardio.itbm.rm.cnr.it.

Cardiovascular engineering includes several research topics characterized by the use of engineering techniques. These topics include hemodynamics, signal analysis, active and passive cardiovascular prosthetics components, automated diagnostic procedures, and others. The research activities of Cardiovascular Engineering Department of our Institute are, in general, oriented to the study of interaction between the ventricle and circulatory network. To this aim, researchers devote special attention to hemodynamics evaluation under different physiological and pathological conditions. Initially, studies are usually done with hydraulic and biological (sheep) models. These methodologies are then transferred into clinical environments. In these environments, there is a large demand for real-time monitoring systems. Such systems must be flexible, easily reconfigurable to accommodate new experiments, and must consider the possibility of interfacing to different medical devices, such as patient monitors, biological amplifiers, and so on. Furthermore in order to introduce these new methodologies in daily clinical activities, data presentation must be designed considering standards used by traditional medical devices.

The proposed system resides on a Windows-based PC implemented with a ADC/DAC conversion board from National Instrument (6040E and 6041E families). The PC uses a Pentium II 366 MHz processor with 132 MB RAM and 5 GB HD to allow the system to use computational capability; however, less powerful machines can also be used.

The software is designed using National Instrument's LabVIEW, an environment operating under Windows that creates programs in block-diagram forms. LabVIEW offers a built-in library for data acquisition, processing, analysis, and display. The library is implemented in the graphical language G. LabVIEW programs are called virtual instruments (VIs), because their appearence and operation imitate actual instruments. However functions from the G library are analogous to functions from conventional language programs.

Generally speaking, VIs comprise a block diagram (Figure 1) and user interface or front panel (Figures 2 and 3). The block diagram gives a pictorial solution to a programming problem and contains the source code for VIs shown as icons and connectors. The front panel simulates a panel of physical instruments and contains both controls and indicators in such a way that the operator can input data using keyboard and mouse and then then view results on a computer screen.

Figure 1:   This block diagram of the proposed VI system includes a data acquisition system with an ADC system.

The core of the software driving the system is the acquisition section. The G code for this part of the software is represented in Figure 1. Standard LabVIEW functions set up the system's hardware capabilities before running ADC conversions. The ADC board is designed to acquire data in DMA using a 20,000-sample circular buffer. Data retrieval occurs in a while loop structure (the thick line in Figure 1). During each iteration, ten samples are transferred in a LabVIEW array for computation.

Using this approach objects performing different functions can be added using visual wiring capabilities. Non-expert users can create simple objects based on system library VIs. More complicated analysis can be implemented by an expert programmer and simply wired into the main program. Furthermore, since LabVIEW supports custom codes written in other languages, you can integrate writen graphical language software with source code written in C++, interfaced to the main program using some special components of these tools (for example, the Code Interface Node or CIN).

A further capability of our hardware/software set up is to communicate, or network, with other processes. This allows our system to be applied in a LAN or to be used as an Internet server to share real-time data on the Web.

Innovative Patient Monitoring
Patient monitoring and consequent data interpretation offers the possibility to have both information on patient's status and on the changes produced by therapy. By this point of view, the combination of different signals, even if by simple computations, is useful in clinical practice. The implementation of these methodologies supports the extraction of further information on hemodynamics and energetic parameters from existing signals. This capability can reduce the global invasiveness of the measurement system and permits an easier uptake of research results from bioengineering to the clinical routine. Starting from these considerations, we have developed an application able to implement an innovative signal-processing application in intensive care and during surgery. In this occurrence the system is based on a battery powered notebook PC to avoid problems regarding safety according to IEC 601 safety requirements. Visualization and simple manipulation of data coming from different devices can be easily performed. In this way it is possible to test new and simple multimodal analysis. The on-line study of the interaction of respiratory and cardiovascular systems has been implemented to reduce invasiveness in left-ventricular preload assessment.

Figure 2:  Our instrument's output for in vivo studies. The figure shows, from top to bottom, ECG, arterial pressure, venous pressure, and expired CO₂. The right side of the figure shows required on-line computations.

Benchmark for Mechanical Heart Assist Devices
In the study and in the evaluation of mechanical heart assist devices (such as IABP and VAD) special attention is given to both the device mechanical properties and to their effect on the circulatory system under different hemodynamic conditions.

Modeling these phenomena can be helpful in a wide range of clinical and research applications and frequency domain analysis has been widely applied. In particular, frequency domain representation of arterial impedance as well as pressure transfer function of different vascular districts are used to evaluate mechanical properties in the presence of IABP and VAD. In this context, hydraulic circuits to simulate different conditions of the heart and different arterial loads are used.

This section of the article summarizes how we have developed a set of virtual instruments to perform real-time frequency-domain data analysis to characterize cardiovascular devices and their loads.

Figure 3:  Output of our system for in vitro benchmarks. On left side acquired signals (two pressures and flow); on right side computed input impedance and pressure transfer function (modulus and phase waveforms).

Figure 3 is the front panel of the system computing harmonic response (in terms of input impedance and pressure transfer function) of an hydraulic network simulating systemic circulation. The acquisition of pressure and flow signals is triggered to cover a whole cardiac cycle.

In benchmarking it is moreover often necessary to integrate numerical and hydraulic models and allow the exchange of data between them. The problem of real time data acquisition and analysis is clearly central also in this context. It must be considered that often numerical models are described using different languages like for example C. We found a solution to this problem using the real time capabilities of LabVIEW and integrating it with source codes written in C++.

The proposed system is useful in cardiovascular engineering for innovative prototyping and/or limited production of new devices.