Edward J. Ciaccio, Ph.D. Center for Biomedical E ~ g ~ n e e ~ i n ~ /

D e p ~ ~ m e n t of P h a r m ~ ~ o l o ~ y Columbia University

32

ultichannel data in electrocardiology is essential for understanding the

complex interactions between the elec- trical and mechanical components of the cardiovascular system. If the electrical activity of the heart is abnormal, the me- chanical properties of the heart will be af- fected, which can cause inefficient pumping of blood. Sudden death from abnormalities of the electrical activity of the heart is common in cardiac pa- tients, because for many types of heart

Inc higher bandwidth in the digital pre- processing stages. Increased real-time and archival data transfer rate and storage capacity.

9 Increased microProcessor speed for file creation and signal processing. Increased graphics-card Veed for video display of data. ' The to extract "'lent lnforma-

tion from larger data space. Another major consideration is the

disease there are few effective &era- Of proce- pies, Pharmacologic agents, surgery, * dures 'sed to acquire data. Open-heart and ablation methods have been used to invasive and can lead to interrupt cardiac arrhythmias, but the morbidity and pickup

need to reduce the

surgery is

rate is highly dependent on the type of arrhythmia, For example, wolf- Parkinson-White Syndrome (wPw) is

from transducers residing within an in- dwelling catheter is less invasive, but such signals cannot be directly obtained from

highly treatable because often the exact side Of the heart. Depending pathways causing abnormality of the On the type, acquisition data electrical rhythm can be readily identi- sites spaced far apart in the en-

the

from fied, localized, and ablated (i.e., inter- docardium may be rupted with radlo-frequency energy, rather than parallel, owing lo catheter size laser light, or cryothermal means). constraints. Such catheter-based systems Other diseases, for example ventricular for identification and localization of car-

tachycardia caused by re-entry, are less effectively treated because the mecha-

diac electrical need to be robust to time-dependent nonstatlona1-

niSmS by which these arrhythmias are initiated, maintained, and terminated

ties in the data obtained from a given site and the relationships between sites. Pres-

are not known with certainty. There- fore, a major problem in electrocardlol-

ently, prototypical systems Of

arrays that can be fitted into a ogy is the acquisition of the correct data catheter-based device are being tested. and the ability to process these data in These systems can simultaneously semi- such a way so that treatment becomes automatically position electrodes at many

sites along the endocardium, by expan- more efficacious. data Slon of the array, once the catheter is in the

provides more information about the cardiac chamber. Thus, the electrical state of the heart, as both the number of properties Of the endocardium can be recording sites and the digitization rate mapped at points simultaneously. increase. As throughput increases Other catheter-based devices for record-

rate) there is a need for: neously position transducers at the center

tion, and density of front-end trans- there. An inverse solution is required to ducers transform signals acquired from the . Prevention of the additional trans- pickup sites in the center of the cavity, ducers from interfering with the through the blood-filled left ventricular electromechanical properties of the cavity, to the corresponding signals that heart. would be acquired from the endocardial . Prevention of crosstalk in the analog surface. Potentially, epicardial signals can stages when channel density in- also be derived from endocardial signals creases. and/or from signals i n the mid-

A~~~~~~~~~~ of

(more channels and higher digitization ing many endocardial potentials simulta-

, Increased reliability, miniaturiza- Of the chamber and pick

IEEE ENGINEERING IN MEDICINE AND BIOLOGY 0739-51 75/98/$10.00019(18 Janoory/February 1998

myocardium. This would provide infor- mation about the activation of the entire heart from a catheter-based system.

Body-surface recordings are also promising because they are noninvasive and potentially delineate all of the activ- ity that occurs on the epicardial surface of the heart. The problems that arise when using body-surface recordings in- clude accurately positioning the body- surface electrodes and performing the correct inverse calculations from body surface to epicardial surface. These cal- culations depend upon the volume con- ductor, which is in turn dependent upon both the patient and motion of the patient. Another consideration is how to go from epicardial surface signals to endocardial signals. Since many cardiac-electrical

abnormalities originate in the endocar- dium, body-surface acquisition and analysis systems may need to address the problem of relating epicardial to endocar- dial electrical activity.

Finally, it would be very desirable to understand cellular mechanisms and phe- nomena from surface or intramyocardial recordings. If the signal morphology can be related to cellular ionic currents, it would allow the effects of pharmacologic agents, ischemic conditions, temperature, and other factors to be studied in the whole heart, rather than cell by cell. This potentially would greatly expand the clinical use of cardiac antiarrhythmic drugs, because their effects on the ar- rhythmogenic heart could be more rapidly and completely identified.

The articles in this issue of IEEE Engi- neering in Medicine and Biology Maga- zine focus on current systems for acquisition and analysis of multichannel data in e lectrocardiology. Several articles are devoted to hardware considerations, others are devoted entirely to software analysis, and a few address complete sys- tems. The purpose of the issue is to direct the attention of biomedical engineers and clinicians toward the need to develop more accurate, efficient, and inexpensive methods for working with multichannel data in experimental and clinical electro- cardiology. It is hoped that the issue will stimulate much fruitful thought, suggest the need to devote more attention to these issues in the literature and at clinical and bioengineering meetings, and perhaps gen- erate a siinilar focus for other signal types.

Student’s Corner (Continued from page 31)

this information may be used to determine levels of chemicals such as glucose.

When a photon of a given energy col- lides with an atom having an equal excita- tion energy, the atom will “absorb” the light, making a quantum jump to a higher energy level. In dense gases, liquids, and solids, absorption occurs over a range or bands of frequencies. As one can imagine, different compounds will exhibit unique absorption spectra in the region of funda- mental absorption frequencies. The amount of radiation absorbed follows Beer’s law, which states that the light in- tensity at a given wavelength in a purely absorbing sample decays exponentially as a function of absorber concentration, pathlength, and absorptivity. The mid- infrared region contains the fundamental absorption bands, which allow identifica- tion of chemical species, but, because of the low energy of the photons, only very short pathlengths can be used. This makes the mid-IR unsuitable for quantitative measurements in most biomedical appli-

cations. Higher-order absorption bands are present in the near-infrared region, where longer pathlengths are achievable. These bands are broader and overlapping, so it is difficult to use these for compound identification. Measurements are compli- cated in the presence of scatterers, so a more complex theory than Beer’s law must be applied.

Absorbed energy is generally dissi- pated by way of intermolecular collisions. However, it is possible that an excited spe- cies emits a photon. Resonance radiation is emitted at the same frequency of the ab- sorbed photon. Luminescence occurs when the excited species drops down to an interim energy state and emits a photon of lower energy than the incident photon. When this occurs rapidly, on the order of hundreds of nanoseconds, it is called fluo- rescence, Tissues exhibit intrinsic fluores- cence (autofluorescence) and fluorescent markers can be used to bind to specific chemicals of interest.

Most light scattered from a sample ir- radiated by a quasimonochromatic source is of the same frequency as the incident ra- diation. However, there are very weak

scattered components of higher and lower frequencies, with the frequency being characteristic of the material. The inten- sity of Raman scattered light is linearly proportilonal to the concentration of the analyte. Raman spectroscopy is used simi- larly to absorption spectroscopy to iden- tify compounds as well as make quantitative determinations. Because it is based on scatter, the information is com- plementary to that obtained from absorp- tion spectroscopy.

Clearly, the applications of optics to solving biomedical problems cover a broad spectrum. There is a tremendous potential for innovative research in devel- oping nlzw methodologies or improving existing techniques. This article only gives a \ ery basic and generic overview of some of the approaches being investi- gated. R.eferences for the topics consid- ered here can be obtained from the authors.

Smita Sampath can he reached at sOs5814@acs.tamu.edu and Mike McShane can be reached at mcshane @acs.tamu.edu.

January/February 1998 IEEE ENGINEERING I N MEDICINE AND BIOLOGY 33