INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

CELLULAR ACTIVATION MAPPING (specifying in time and space the electrical activation sequence of cells) is a well-established basic research tool in cardiac (15), neural (2), and to a limited extent gastric (12) physiology. Tremendous strides have been made in the understanding of such common diseases as epilepsy (2) and sudden cardiac death (15) with the use of electrical mapping techniques.

Mapping is accomplished by placement of an array of spatially defined recording electrodes on or in the target organ. The most common approaches to creating electrode arrays are variations of the technique presented by Cohen et al. (6). In their approach, individual wires are carefully cut and placed into rigid (11) or semirigid (17) forms, resulting in electrode plaques. Electrodes as small as 50 µm in diameter, spaced 200 µm apart, have been painstakingly created with the use of this approach (1, 11).

However, as the number of simultaneous recording electrodes has risen to over 1,000, placing individual wires into forms has become a tremendously time-consuming and error-prone procedure. For very fine arrays, with center-to-center spacing of <200 µm, the hand-crafted approach invariably yields irregular electrode spacing. In addition, the connectors (the device responsible for physically and electrically interfacing the electrode array to the recording system) also require each wire to be individually placed and are frequently the cause of broken connections and lost data.

Silicon-based photolithographic manufacturing processes have been proposed (5) to circumvent some of these problems, but these are financially and technically out of reach for most potential users (4). In this paper, we present a novel construction technique for manufacturing electrode plaques on the basis of fine wire ribbon cables. The presented approach enables arrays to be constructed with repeatable electrode spacing, allows for reliable mass termination connectors, and can be used at very high densities (<100 µm center-to-center distance). Electrode arrays can be rapidly constructed without specialized equipment, or, because construction depends on techniques common in the electronics industry, there are many commercial vendors who will supply nearly complete electrode arrays at a relatively low cost.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

One of two materials was used as the foundation for these electrode arrays: fine-pitch ribbon cables or flexible printed circuit (FPC) boards. The fine-pitch, flat ribbon cable (136 series) used here is manufactured by 3M Electronic Products (Austin, TX) for the computer industry. The ribbon cable is made with polyvinylchloride (PVC) insulation separating 30-gauge (254-µm diameter) solid, round copper conductors with a 635-µm center-to-center distance. The cable comes in widths ranging from 20 to 68 parallel conductors and is easily trimmed to nonstandard widths.

FPC boards are constructed of copper foil applied to polyimide. They are common in the electronics industry, leading to a proliferation of providers. Today, most FPC manufacturers can provide 100-µm conductors with a 200-µm center-to-center spacing. Some manufacturers can deliver 50-µm conductors on 100-µm centers. Where larger dimensions are acceptable (above 800 µm), the ribbon cables can be manufactured in any laboratory with the use of a laser printer and one of the transfer/etching kits (sold by most suppliers of electronics). With the use of the smaller dimensions, a photolithographic process is used, which is beyond the capabilities of most laboratories. To obtain fine-pitch FPC ribbon cables, the investigator provides a drawing to a manufacturer stipulating the length of the conductors, the width of the conductors (length of the final electrodes), the number of conductors, the spacing of the conductors (the electrode separation in the x-direction), the thickness of the conductors (width of the electrodes), and the thickness of the insulating layer (electrode separation in y-direction). Many FPC manufacturers will accept these details in the form of rough sketches, like that shown in Fig. 1, step 1. If these specifications are provided in the proper electronic format, the completed ribbon cables can be delivered in a few days. In this work, we manufactured the FPC boards in our laboratory.

View larger version (24K): [in this window] [in a new window]   Fig. 1.   A typical construction sequence takes from 2 to 5 steps. Step 1: if computer ribbon cable (polyvinylchloride; PVC) is used as raw material, then this first step can be skipped. If flexible printed circuit board material (polyimide) is used, the first step is to draw desired geometry of electrodes. Usually a rough sketch, such as that shown here, is sufficient for a manufacturer to create desired polyimide ribbon cables. (In this domain of electronics industry, it is traditional to specify dimensions in inches.) Step 2: PVC ribbon cables are now cut to length. Polyimide ribbon cables will be delivered already cut. Ribbon cables are glued together such that cut ends form active face of array. Step 3: mass termination connectors are added. Most manufacturers of polyimide substrates can also perform step 2 and step 3. In some experiments, copper electrode array completed after step 3 is ready to use after only ~1 h of labor. Step 4: however, in most cases, electrode array is plated with Ag. Step 5: in the most demanding applications, some of Ag at surface of electrodes is converted to AgCl.

Whether PVC or FPC ribbon cables should be used depends on the application. PVC ribbon cables are inexpensive and readily available. The FPC cables offer greater flexibility and smaller sizes but are more expensive. In either case, the use of ribbon cables greatly facilitates connection to the mapping system via mass-termination connectors. For PVC ribbon cable, these are physically attached and electrically connected by use of a single crimp. In this work, Thomas & Betts (311 series) connectors were used. For FPC ribbon cable, the connector end of the cable must flair out to meet the specified connector (see Fig. 1). In this work, we used a 1-mm spacing zero-insertion force connector (Hirose Electric, FH21-120S-DSA).

The electrode-to-electrode spacing in the direction parallel to a single ribbon cable is governed by the wire-to-wire spacing of the ribbon cable. The wire-to-wire spacing is tightly controlled by the manufacturer, because the cables are made to mate with standard connectors. In the direction cutting across ribbon cables, the electrode spacing is primarily controlled by the insulation thickness. Again, because these ribbon cables must mate with standardized conductors, the insulation dimensions are also tightly controlled during manufacture. The reproducibility of the insulation thickness and wire-to-wire spacing is responsible for the very regular spacing of the electrodes in these arrays.

However, errors in the location of the electrodes in the final array are possible. If the multiple layers are not carefully assembled, they can be misaligned and/or nonplanar. Careful clamping can eliminate nonplanar errors. Alignment errors are typically not a problem with PVC ribbon cables, because their end-to-end distance is fixed by the insulation thickness. So, stacking PCV cables on end assures alignment. With FPC ribbon cables, the end-to-end distance is controlled by cutting, a relatively rough manufacturing process. To avoid alignment errors with FPC cables, alignment marks are incorporated into the design. Each layer is placed individually, aligning the marks at each stage. Slow-setting cynoacrylate and a dissecting microscope are required to effectively eliminate alignment errors in FPC ribbon cables.

Several prototype arrays were manufactured for this study, using both PVC and FPC ribbon cable. For the PVC ribbon cables, the first step was to cut the required lengths of ribbon cable. Polyimide ribbon cables are delivered cut to the desired length. The PVC cables were epoxied (GC Electronics epoxy type 347) directly to each other, with alignment to the desired electrode plaque shape. FPC ribbon cables were glued together with the use of cynoacrylate. After the glue was allowed to dry, the electrode end was ground flat on a grinding wheel, hand polished (600-grit sandpaper), and rinsed clean in distilled deionized water. Mass termination connectors were attached to the free ends of the cables to complete a copper electrode array. The assembly steps are labeled as steps 2 and 3 in Fig. 1.

One advantage of using FPC ribbon cables is that the FPC manufacturer can perform the assembly steps for the investigator. When assembled at the manufacturer, the electrode array is called a "multilayer flexible printed circuit board." Multilayer FPC is a routine practice in the electronic industry. For example, an investigator could specify a 20-layer board, each layer like that shown in Fig. 1, step 1. The provider would return a complete, 400-electrode recording array, even placing the connectors if desired. One significant advantage of using a commercial vendor is that the errors that can be introduced because of gluing thickness, misalignment, and nonplanar assembly are largely avoided.

In some experiments, a copper array may be usable without alteration. If so, then the electrode plaque is complete and ready to use. Less than 1 h is required for construction, excluding the epoxy cure time. If desired, the completed copper plaque can be purchased from an FPC manufacturer.

However, in most cases, the polarization and half-cell potential variability of copper electrodes introduce excessive noise in the activation mapping recordings. Therefore, plating with silver and then perhaps coating with silver chloride is required.

A number of soluble silver salts can be used to deposit silver. However, the silver cyanide bath is inexpensive, simple to prepare, and stable (3). The major disadvantage of the cyanide bath is the potential toxicity of the cyanide, which is minimized here by use of small quantities at room temperature (3). This study used a silver cyanide bath consisting of 7.2 g AgCN, 9.75 g KCN, and 6.0 g K₂CO₃ dissolved in 150 ml of deionized water. The K₂CO₃ is dissolved first to insure that the solution is never acidified, which might release HCN, the toxic gas. Nevertheless, work should be conducted under an appropriately rated hood.

For silver plating, the cathode is the copper electrodes, and the anode is a strip of 99.9% pure silver foil. The electrodes' connectors can be mated to a strip of shorted compatible connectors to facilitate plating. The area of the anode must be greater than the area of the cathode. We used a silver foil surface area 1.5-2.0 times the copper electrode surface area. When reusing the silver foil, it must be thoroughly sanded because, as the anode, the silver foil will become coated with a white silver salt that will block the solution's access to the pure silver. The electrode array was placed in 20 ml of plating solution, which was stirred continuously. A plating current density of 2.0 mA/1.0 cm² of electrode area (10) was used for 6 min. The same current densities worked for all electrode sizes attempted here.

Where polarization is critical, a silver chloride coating is required. For chloriding, the electrodes were thoroughly rinsed in deionized water and then submerged in a second bath consisting of 0.9 g NaCl in 100 ml of deionized water. The cathode is a freshly cleaned and sanded piece of silver foil. A current density of 2.0 mA/cm² was imposed for 30 min. A completed electrode array built from PVC ribbon cable is shown in Fig. 2.

View larger version (111K): [in this window] [in a new window]   Fig. 2.   A completed 400-electrode array. Individual electrodes are visible near center of array, where Ag/AgCl coating has been intentionally removed to increase visibility. Mass termination crimp connectors at other end of each ribbon cable (not shown) directly and reliably connect electrodes to recording system. Individual wiring is not required. This electrode required approximately 2 h to construct.

Once completed, the electrodes were immediately submitted to testing. If the electrodes were perfect, then the potential difference between any two electrodes in a conducting bath would be zero. In reality, the metal of the electrodes and any contaminants react with the bath, creating a potential difference that can vary with time (9). This time-dependent (alternating current; AC) signal is noise to the measurement system. Although a time-independent (direct current; DC) voltage difference can exist as well, it is generally removed by the mapping system (15). The AC root-mean-square (rms) potential was measured (Hi-Z Fluke meter GMM865) between two adjacent, individual electrodes on each prototype. The measurement was repeated after 2 h.

An electrode is nonpolarizable if a rapidly reversible chemical reaction occurs at the electrode-to-electrolyte interface when exposed to a voltage source (18). The testing done here was conducted according to the polarization protocol of Witkowski and Penkoske (18). Their protocol is intended to simulate the recording of the electrical activity of the heart after the application of a defibrillation pulse.

A functional test of the arrays was conducted in guinea pigs. Cellular activation maps were made in the open-chest guinea pig model (7, 14). Large male guinea pigs (800-1,200 g) were anesthetized with 200 mg/kg intraperitoneal cremophor-solubalized propofol initially, followed by 45 mg/h intravenous (penile vein, 24-G catheter) thereafter, or as indicated. Direct laryngeal intubation facilitated continuous positive-pressure ventilation. The body surface electrocardiogram, temperature, and oxygen saturation (pulse oximetry) were continuously monitored. The heart was exposed with a median sternotomy and placed in a plastic pericardial sling with an attached silver-silver chloride reference electrode. For the reference electrode, the surface of a 99.9% pure silver wire was converted to silver chloride using the technique described above.

We used a 400 (20 × 20)-electrode PVC array for the functional test. Cellular activation was recorded from the central 14 × 14 array of electrodes (vs. the reference electrode in the sling). Pacing stimuli were delivered through a stimulus isolation unit (model 880, AM systems) to a cranial row of electrodes. Strong T wave stimuli were delivered from a right lateral row of electrodes from a defibrillator (HVS02; Ventritex, Sunnyvale, CA). The outside ring of electrodes was grounded, serving as a signal shield against noise. Remaining electrodes were grounded. Recordings were made with the use of our mapping electronics, which have been previously described (13). To minimize trauma to the epicardium, the electrode array was held against the heart only during a recording. Electrode contact was confirmed by the low pacing threshold (<1.0 mA for a 2-ms rectangular pulse) and uniform activation of each electrode during plane-wave pacing.

Because the 36 stimulating and 196 recording electrodes are both part of the larger 400-electrode array, we were guaranteed a consistent geometry between electrodes throughout the study. Perhaps more significant is the fact that the geometry was guaranteed to be consistent between animals. Because the electrode construction process is also consistent, it would have been possible to use multiple electrode arrays without concern for a changing electrode geometry.

The negative first derivative of the recorded epicardial potential is often taken as an indication of cellular activation. Here, the first derivative of each recording was made with the use of the three-point central-difference technique followed by a five-point moving-window low-pass filter. The resulting signals were animated with the use of color as the indicator of the derivative's magnitude. Because observation of the data in this manner requires viewing a movie, an alternative technique is required for publication. Here, gray-scale intensity was used to display the time of the maximum derivative, an alternative indicator of cellular activation.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Four PVC ribbon cable arrays and three FPC ribbon cable arrays were constructed. The PVC arrays were sanded and coated a total of 23 times, of which 9 were finished with silver electrodes and 14 with silver/silver chloride electrodes. The FPC ribbon cable was sanded and coated with silver six times. The process was found to be rapid, requiring only 2 h of labor to create a complete electrode array including connectors and silver/silver chloride coating. The 2 h do not include curing or plating times, which require the operator to merely wait. Although it is tempting to use a rapid-setting epoxy to reduce the amount of waiting time, we found that the rapid epoxies would disintegrate during plating. The ribbon cable arrays survived repeated polishing and plating without problems.

Table 1 summarizes the measured voltage differences between adjacent pairs of single electrodes in a 0.9% NaCl solution. Electrode pairs are from randomly selected PVC ribbon cables, with each pair being from separate plating batches. As expected, a silver chloride coating significantly reduces (0.76-0.42 mV; P < 0.001, Z test for mean difference) the fluctuation in the potential difference between two electrodes. Table 2 summarizes the data obtained during polarization testing. Figure 3 shows an example of the raw data from this test. A nearly immediate recovery of the potential difference between adjacent pairs of silver/silver chloride electrodes is observed, whereas the silver electrodes do not recover for more than 4 ms. Larger shocks required 20 ms or more to recover for silver electrodes.

View larger version (26K): [in this window] [in a new window]   Fig. 3.   Results of 1 polarization test using Ag/AgCl electrodes. Recovery of potential between 2 electrodes is nearly immediate after a defibrillation (Defib)-strength shock. However, when this same test is run on Ag electrodes, potential does not return to within boundaries before 4-20 ms after defibrillation-strength stimulus.

Figure 4A shows a recording from a single electrode in the open-chest guinea pig. Figure 4B shows the derivative of this recording. At the start of this recording, the pacing stimuli induce plane-wave pacing. Figure 4C shows the uniform nature of the times of the maximum negative derivative in one such cycle. A defibrillation stimulus early in the T wave had no effect. In a later recording, when the defibrillation stimulus was delivered, the uniform plane-wave pacing changed to what appears to be a circulating wave front during the first 48 ms after the stimulus (Fig. 4D).

View larger version (89K): [in this window] [in a new window]   Fig. 4.   In this experiment, heart is paced for several beats. Then a defibrillation-strength shock is delivered during a critical phase of heart's recovery cycle. A: recording from a single electrode in array. B: derivative of this recording. Because derivative is often signal that is analyzed to determine cellular activity, fidelity of this signal is paramount. C and D: time of peak negative derivative (when cells beneath electrode can be considered to be activating) for every electrode on array from a later recording. In response to pacing stimulus, heart shows uniform plane-wave pacing (C). However, after defibrillation pulse, activation waves begin to rotate, a condition that quickly leads to ventricular fibrillation (D).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

There have been many methods proposed to manufacture silver/silver chloride electrodes for the recording of the electrical activity of cells [see Geddes and Baker for a review (9)]. However, we report the highest-density large-electrode array that can be rapidly produced. This is also the only technique that allows the investigator to purchase nearly complete arrays, requiring only silver plating. The small spacing possible with FPC arrays (50-µm electrodes on 100-µm centers at this writing) is ideal for a study such as that proposed by Spach and Heidlage (16), who hypothesize that micro-reentry (cardiac arrhythmias on the scale of a few hundred micrometers) could be a major cause of morbidity and mortality. Neural (2) and gastric (12) physiology studies in small animals will benefit from the small-electrode spacing possible with this technique. The ability to create a consistent and repeatable electrode spacing is a critical advantage of this array for studying stimulation patterns where the angle between the electrodes is influential.

Where anatomic correlation is critical, marker beads can be glued or sutured to the myocardium to insure that the electrode is placed at the same location for each measurement, assuming multiple placements are made. As suggested by Cohen et al. (6), a mounting jig can also be used.

The AC rms value of the potential difference between adjacent electrodes, a signal that would appear as noise for most mapping studies, was comparable with the 200 µV measured by Geddes and Baker (8). Furthermore, the polarizability of our silver/silver chloride electrodes was nearly identical to that described by Witkowski and Penkoske (18). Thus the performance of the proposed array is adequate, even for studies involving defibrillation-strength stimuli, one of the most challenging mapping applications.

In conclusion, the arrays described here are rapidly created and offer reliable connections. The FPC construction approach is particularly promising, because it lends itself to several extensions. The most direct extension of this approach is to attach electronics directly to the FPC. This is a common practice in the electronics industry. In this way, amplifiers and filters could be placed very close to the sensing electrodes, improving fidelity.