Annals of Biomedical Engineering, Vol. 26, pp. 1055–1064, 1998 Printed in the USA. All rights reserved. 0090-6964/98 $10.50 1 .00 Copyright © 1998 Biomedical Engineering Society Recording Properties and Biocompatibility of Chronically Implanted Polymer-based Intrafascicular Electrodes JAMES A. MALMSTROM, TIMOTHY G. MCNAUGHTON, and KENNETH W. HORCH Department of Bioengineering, University of Utah, Salt Lake City, UT (Received 12 August 1997; accepted 11 May 1998) Abstract—We implanted polymer-based longitudinal intrafascicular electrodes ~polyLIFEs! in feline dorsal rootlets acutely and for periods of two to six months to evaluate their electrical properties and biocompatibility. A total of 38 implanted electrodes were analyzed. Some 25 of the 38 electrodes were implanted with an insulative flexible polymer cuff, which was required for recording of afferent activity in situ. Electrode impedances remained stable for the duration of the experiments. The distributions of axons were measured at three levels of the implanted rootlets: the implant level, 1–2 mm proximal to the implant with respect to the cell body, and 1–2 mm distal to the implant with respect to the cell body. Similar measurements were made in five samples of fascicles neighboring an implant and six samples of control tissue from animals in which no implants were placed. The polyLIFEs demonstrated a high degree of biocompatibility, as no adverse effects on axon size were observed in either the implanted fascicle or neighboring neural tissue. However, the insulative cuffs were found to be a source of compression, resulting in necrosis of the neural tissue. © 1998 Biomedical Engineering Society. @S0090-6964~98!01606-3# though chronic studies revealed two problems associated with metal LIFEs comprised of 25 m m diam Pt/Ir wire, the first being drift in the population of cells being recorded, and the second being a decrease in signal-tonoise ratio over time. Both of these were attributed to differential movement of the relatively stiff electrode within the soft neural tissue, resulting in continuing fibrous encapsulation. These observations prompted the development of a significantly more flexible, polymer-based LIFE ~polyLIFE! consisting of a 12 m m diam Kevlar fiber coated with three metal layers: titanium/tungsten, gold, and platinum.14 The increased flexibility and relatively high tensile strength of the polyLIFE yields a better mechanical match to the neural tissue, theoretically improving the long-term recording characteristics previously reported for metal LIFEs. Although LIFEs have demonstrated usefulness in various peripheral nerve applications,4,21 they have not previously been used for study within the central nervous system. The organization of the feline dorsal rootlets of the cauda equina makes this an appropriate anatomical location to extend peripheral nerve techniques into the central nervous system. However, unlike peripheral nerves, the dorsal rootlets lack a perineural layer, making them much more delicate and more susceptible to any mechanical trauma caused by the implanted polyLIFEs. Despite this difficulty, this tissue was chosen for investigation because it undergoes little motion under normal physical activity, minimizing differential movement of the electrode within the neural tissue. The purpose of this study was to evaluate the feasibility of making neural recordings from feline dorsal rootlets with intrafascicular electrodes and to study the long-term biocompatibility of polyLIFEs. Keywords—Neuroprosthetics, Dorsal roots, Spinal cord, Intrafascicular electrodes. INTRODUCTION Functional electrical stimulation ~FES! is a promising method for restoring motor function and sensation to persons with spinal cord injury or stroke. Improvements in FES systems are dependent on the development of better interfaces to record from and to stimulate neural tissue. The longitudinal intrafascicular electrode ~LIFE! is a potential candidate to fill this role. This type of electrode is implanted longitudinally within individual nerve fascicles and has been used to both activate and record from restricted subsets of axons within a fascicle, demonstrating topological selectivity in axonal recruitment for neural stimulation22 and multi-unit extracellular recording.2,12,13 The geometry of the LIFEs makes them reasonably stable during long-term implantations,10 al- MATERIALS AND METHODS Polymer-based longitudinal intrafascicular electrodes ~LIFEs! were constructed from individually metallized, 12 m m diam, Kevlar® @poly~terephthaloylchloride/p- Address correspondence to Ken Horch, Bioengineering, University of Utah, 50 S. Central Campus Dr., RM 2480, Salt Lake City, UT 84112-9202. Electronic mail: k.horch@m.cc.utah.edu 1055 1056 MALMSTROM, MCNAUGHTON, and HORCH phenylenediamine!# fibers.14 The active zone in this type of electrode is a 1 mm long region along the shaft of the fiber approximately 1 cm back from the tip. This region is platinized by placement within a drop of platinic chloride and passage of 60 Hz, ac current ~10–50 m A! for approximately 30 s. The remainder of the electrode is insulated with a 1–3 m m thick layer of medical-grade silicone elastomer ~Dow-Corning, Silastic MDX4-4210!. Electrode impedance was measured using a 1 kHz constant current sine wave applied to the electrode and returned through a large diameter hypodermic needle inserted into the quadriceps muscle. The return electrode had an impedance below the resolution limit of the measuring system, and so did not contribute to the measured impedances of the LIFEs. Current applied during impedance testing was limited to the nanoamp range to prevent any damage to the surrounding tissue. Three groups of domestic cats were used in the experiments. The first group characterized the acute recording properties of the electrodes when implanted in feline dorsal rootlets. The second group was used to study how the recording properties change during chronic implantation, and the third group was used to determine longterm biocompatibility. Acute Recording The acute experiments involved six animals. Anesthesia was induced with intraperitoneal sodium pentobarbital ~40 mg per kg! and maintained throughout surgery and subsequent experimentation with the same agent delivered intravenously. A laminectomy of vertebrae L4–L7 was performed to expose the spinal cord. The dura was incised and reflected back to expose the cauda equina. Individual rootlets were separated and a single polyLIFE was threaded longitudinally into the fascicle for a distance of approximately 1 cm with the recording zone centered in this region. Up to five electrodes were implanted per cat. An extrafascicular reference electrode was placed in the same region of the spinal roots. Recordings were made differentially between an individual intrafascicular electrode and this reference electrode, and the animal was grounded through a hypodermic needle placed in the quadriceps muscle. We found it relatively easy to isolate rootlets for the required length because vasculature in this location is almost solely contained within the rootlets, and there is very little connective tissue between the rootlets.19 We observed that neural signals were generally too small to be detected above instrumentation noise when an implanted rootlet was bathed in cerebrospinal fluid. To remedy this situation, insulating nerve cuffs were fabricated using 1 cm30.5 cm sections of 5 m m thick polyethylene film which were flexible enough to wrap easily around individual rootlets and could be tacked closed with 10-0 suture. After the electrodes had been implanted, and each rootlet had been ensheathed within a cuff, the electrode wires were led out of the incision and the dura was sutured closed. Receptive field mapping was carried out by exploration of the hindlimb and tail using manual palpation and various static and vibratory probes. Neural signals were monitored on an oscilloscope and with an audio amplifier. Individual units were identified as described previously.3,13,14 For each unit which could be individually activated, the signal-to-noise ratio ~based upon peakto-peak amplitudes! and sensory modality were determined, and the locations of the receptive fields were noted for cutaneous mechanoreceptor units. Chronic Recording A total of 25 electrodes were implanted in six cats. Each cat was anesthetized using Halothane. A laminectomy was performed as described above. In addition, a percutaneous connector system was installed to permit monitoring of neural activity on a chronic basis. The percutaneous connector system was completely assembled prior to implantation. Electrode lead wires were attached with conductive epoxy to a multipin brass connector ~Microtech, Inc., Boothwyn, PA 19061! which had been fitted into a custom-machined Delrin pedestal whose base was shaped to fit atop the cat’s sacrum. The electrodes passed through the pedestal and into a 2 cm long silastic tube, intended to provide some strain relief. All connections were potted in silicone prior to implantation. The pedestal was mounted to the cat’s sacrum with bone screws and sutures looped through holes drilled in the dorsal sacral processes. The silastic tube was positioned and the electrodes were then implanted in the rootlets as described above. Polyethylene cuffs were wrapped about each implanted rootlet and sutured closed. The dura, overlying musculature, fascia, and skin were each sutured closed separately. Initial recordings were made immediately postoperatively while the animal was still under anesthesia. At later times, recordings were made from awake or lightly sedated animals at various intervals up to ten weeks following the implantation. Biocompatibility Tissue from four of the animals in the chronic recording group was examined histologically at the end of a survival period of up to six months. An additional 13 electrodes were implanted in three cats for a period of six months. To do so, a laminectomy was performed under Halothane anesthesia as described above. The ends of the electrodes were housed together in a 2 cm long silastic tube for ease in locating them at the end of the study. The distal end of the silastic tube was positioned Intrafascicular Electrodes beneath the dura and sutured into place. The electrodes were then implanted in the rootlets as previously described for the acute studies. The dura, overlying musculature, fascia, and skin were each sutured closed separately. At the end of the survival period, each cat was deeply anesthetized with Nembutal ~40 mg/kg, i.p.!. The impedance of each electrode was measured either through the percutaneous connector, if present, or by making an incision and carefully exposing the ends of the electrodes. The animal was then perfused through the left ventricle with a buffered paraformaldehyde–glutaraldehyde fixative. The entire spinal cord between L4 and L7 was removed with the dura and electrodes still intact, placed in fixative, and refrigerated overnight. The tissue was then cut transversely into pieces approximately 4–5 mm long. Each section was marked with a reference suture looped through the dura to preserve orientation. Control tissue consisting of L6 rootlets was harvested from six animals undergoing acute experiments in which the tissue of interest was exposed to a saline bath for a maximum period of 24 h, but was otherwise undisturbed. The animals were perfused as described above and the rootlets dissected from the spinal cord, placed in fixative, and refrigerated overnight. The sections of implanted and control tissue were then washed in 0.1 M phosphate buffer solution, osmicated, dehydrated in graded alcohols and acetone, and embedded in Durcupan™ Araldite. Using a microtome ~Reichert, Austria! with a diamond knife ~Dupont Instruments!, several 3–4 m m thin transverse sections were obtained at three different levels of the rootlet: ~1! the implant level, ~2! 1–2 mm proximal to the implant with respect to the cell body, and ~3! 1–2 mm distal to the implant with respect to the cell body. Further serial sectioning was done at 75 m m intervals in order to follow the fascicles of interest longitudinally along the rootlet. Similar sections were also obtained at comparable levels of the control tissue. The sections were placed on a glass slide, stained with methylene blue, and mounted with Baxter Su P™ Accu Mount 60J Mounting Medium. At each level, the fascicles were scanned at 403 with light microscopy ~Nikon Diaphot 200, Melville, NY! into the computer using a charge-coupled device camera and image processor ~Argus-20, Hamamatsu Photonic Systems, Bridgewater, NJ!. The scanning was performed on a Power Macintosh computer using the Argus Plug-In ~Hamamatsu Photonic Systems, Bridgewater, NJ! and NIH Image Version 1.59 ~developed at the U.S. National Institutes of Health and available via anonymous ftp from zippy.nimh.nih.gov!. At 403, it was necessary to create a montage of images to form the picture of an entire fascicle. A series of macros developed for NIH Image Version 1.59 ~see the Appendix! was used to 1057 measure the inner diameter and circumference of the myelin sheath of each myelinated axon in the fascicle. A listing of these macros is available from the National Auxiliary Publications Service. Fiber diameters were plotted at each of the three levels for each animal. The diameter distributions were then pooled, using the values for each bin in the distributions from the individual animals, according to source and level. The pooled distributions thus represented ~1! fascicles from implanted cats with no implants ~2! fascicles neighboring an implant ~3! implanted fascicles with cuffs, and ~4! implanted fascicles without cuffs. The number of axons measured ranged from 250 to 4000 fibers per fascicle. Comparisons between pooled diameter distributions were made with the chi-squared goodness-of-fit test. The five nearest Aa fibers to each implanted electrode were identified to obtain their mean distance from the electrode. This distance represents the upper limit of capsular thickness. Encapsulation essentially moves the electrode farther away from the signal sources resulting in degradation of recording characteristics. The fine structure of the capsule was not examined in detail, as it appeared under light microscopy to be similar to the encapsulation normally seen in neural tissue implanted with biocompatible materials.17 For all the animal experiments, principles of laboratory animal care ~NIH publication No. 86-23, revised 1985! and protocols approved by the University of Utah Animal Care and Use Committee were followed. RESULTS Nerve Recording Recordings were made from 28 electrodes in the six animals in the acute recording group and 19 recordings were made immediately postoperatively from the six animals in the chronic, cuffed group for a total of 47 shortterm recordings. Each major type of cutaneous mechanoreceptor and both tonic and phasic motor afferents were represented in the recordings. The number of units present in the recordings from a given electrode varied from 2 to 17 with a mean of 7.863.8 ~s.d.!. The signalto-noise ratio ranged from 2 to 8 with a mean of 3.6 61.3 ~s.d.!. Individual spike amplitudes were as high as 120 m V peak to peak, while background noise varied between 10 and 20 m V. Figure 1 shows a recording from a single guard hair unit stimulated with puffs of air from a pipette. Roughly half ~22! of the recordings contained only cutaneous mechanoreceptor units. Eleven recordings contained only units which responded to joint and/or muscle movement. In the remaining nine recordings, activity from both cutaneous and motor afferents was present. 1058 MALMSTROM, MCNAUGHTON, and HORCH FIGURE 1. Single unit recording made with a polymer-based longitudinal intrafascicular electrode „polyLIFE… implanted acutely in a dorsal spinal rootlet showing the response of a guard hair mechanoreceptor to bursts of air from a pipette. Rectangles at the bottom of the plot show when the stimulus was present. The signal-to-noise ratio of this unit was calculated to be 4.3. Viability of the chronic implants for recording was low primarily due to nerve damage from the nerve cuffs, as described below. However, we were able to record unit activity from a few animals for several weeks after implantation. An example of action potentials recorded at the dorsal rootlet level from a Pacinian corpuscle afferent innervating the foot pad at ten weeks postimplantation is presented in Fig. 2. The unit was activated in a phaselocked manner by a vibratory stimulus, and Fig. 2 is a superposition of several cycles of the stimulus and neural response. This provides an idea of the variability in action potential wave forms seen in a single unit with this recording method. The average electrode impedance was 14.463.9 ~s.d.! kV measured on 20 electrodes immediately postimplant, 12.562.5 ~s.d.! kV measured on 4 electrodes at week 10, and 1362.4 ~s.d.! kV measured on 5 electrodes at week 24. Six of the electrodes became nonconductive within two months of implantation. Although a failure mode analysis was not done on these electrodes, subsequent in vitro studies indicated that the most likely cause of failure was loss of adhesion between the Kevlar® substrate and the conductive layer of gold due to the electrodes being bent sharply at some point, such as where they exited from the silicone sleeve. Nerve Fiber Histology FIGURE 2. Dorsal root recording of phase-locked action potentials from a Pacinian corpuscle evoked by a vibratory stimulus applied to the cat’s hindlimb foot pad at ten weeks postimplantation. The bottom trace shows the relative displacement of the stimulator. Fifteen stimulus cycles are superimposed. Measurements were made of the distribution of myelinated axonal diameters from each of the tissue samples. If the fibers were atrophic, an increase in the percentage of smaller fibers would be evident in the distribution of axonal diameters. A total of ten implanted fascicles with cuffs, 13 implanted fascicles without cuffs, five samples of fascicles neighboring an implant, and six samples of control tissue from animals with no implants were scanned and measured. Representative examples of fascicles from control tissue in animals without implants and tissue neighboring an implant are shown in Fig. 3. Figure 4 shows the Intrafascicular Electrodes FIGURE 3. Representative examples of a fascicle from „a… control tissue from an animal in which no implants had been placed and „b… tissue neighboring an implant. The fascicle is comprised of blood vessels „asterisk… and axons „arrows…. No visual differences between these two groups of tissue were seen, an indication that there are no adverse effects in neighboring neural tissue due to the implanted electrode. Calibration bar: 100 m m. distribution of axonal fiber diameters for the control group without implants and the control tissue neighboring an implant at three levels of the rootlet: proximal to the implant level, at the implant level, and distal to the implant level. There is no statistically significant difference between the two distributions ~p.0.87! at any of the levels. A representative sample of an implanted fascicle with a cuff is shown in Fig. 5. The electrode ~arrow! has been encapsulated by a connective tissue layer. A polyethylene cuff surrounds the fascicle. In two implants, zones of focal demyelination were observed in the cuffed region of the rootlet. Additionally, in one animal no viable myelinated axons were observed at the level of the four electrodes with cuffs. Figure 6 shows the distribution of axonal diameters for the implanted tissue with cuffs and a control group comprised of the combined data from the two control 1059 FIGURE 4. Pooled distributions of axon diameters for the unimplanted control group „light bars… and for control fascicles neighboring an implant „dark bars… at three different levels: „a… proximal to the implant, „b… at the implant, and „c… distal to the implant with respect to dorsal root ganglia. Error bars in each bin indicate the standard error of the mean. Chi-squared goodness of fit comparing the two distributions at each level returned high p values „i.e., showed no statistical difference in the distributions, X 257.8, df517, and p 50.971, proximally; X 2510.7, df517, and p50.872 at the implant; X 257.1, df517, and p50.983, distally…. groups described above. The distribution of fiber diameters of implanted tissue with cuffs is shifted to the left compared to the control tissue. This shift is not statistically significant ~p50.067! proximally, but is highly significant ~p,0.0001! at and distal to the implant. In fact the difference between the two groups was more pronounced than this Fig. 6 indicates, as 10 of the 30 tissue samples from the cuffed fascicles showed no evidence of myelinated fibers. Figure 6 includes only data from the remaining 20 sections in which myelinated axons were present. A sample implanted fascicle without a cuff is shown in Fig. 7. There is no statistically significant difference ~p.0.2 at all levels! between the axonal size distribu- 1060 MALMSTROM, MCNAUGHTON, and HORCH FIGURE 5. Representative example of an implanted fascicle fitted with a cuff. Duration of this implant was three months. The electrode „arrow… has been encapsulated by connective tissue. There is a dramatic decrease in both the number of axons and cross-sectional fascicular area. Note the layer of scar tissue „asterisk… resulting from necrosis surrounding the remaining axons due to compression from the cuff. Calibration bar: 100 m m. tions in the implanted tissue without cuffs and the combined control tissue ~Fig. 8!. Encapsulation An encapsulation layer of connective tissue was present on all implanted electrodes. The mean distance from the implanted electrode to the five nearest viable Aa fibers for ten implanted fascicles with cuffs and 13 implanted fascicles without cuffs was 29.2 6 9.9 ~s.d.! m m and 35.5 6 12.7 ~s.d.! m m, respectively. This distance is less than the reported encapsulation thickness of chronically implanted metal LIFEs ~50 m m!.10 The capsule is comprised of a layer of dense connective tissue immediately surrounding the electrode. This layer then gives way to regions of loose, connective tissue. There is also a small area of reduced axon density in the immediate vicinity of the electrode. It is not clear whether this is an artifact from the implantation process or a response to the presence of the electrode. Some demyelinated axons can also be found among the fibers surrounding the implant. DISCUSSION Recording The recordings demonstrate that it is possible to use polyLIFEs to monitor multiunit activity in the spinal rootlets of the cat. However, to do so, the rootlets had to be wrapped in an insulating cuff. We were able to record from each major type of cutaneous mechanoreceptor unit, as well as from both tonic and phasic motor afferent units. The recording performance of polyLIFEs in cuffed spinal rootlet fascicles closely resembles that of FIGURE 6. Pooled distribution of axon diameters for the combined control group „light bars… and implanted fascicles with cuffs „dark bars…. Format as in Fig. 4. The distributions do not differ statistically at the proximal level „ X 2526.5, df 517, and p50.067…, but do at the level of and distal to the implant „ X 2565.0, df517, p<0.0001, and X 2569.6, df517, p <0.0001, respectively…, indicating a decrease in the size of axonal diameters in the presence of a cuff. One third of the tissue samples from the cuffed population showed no viable myelinated axon profiles, suggesting an even more severe effect than these plots would indicate. metal LIFEs implanted in peripheral nerves both in terms of number of active cells in the recording and in signalto-noise ratio characteristics. In peripheral nerves, fascicles are ensheathed within the perineurial membrane, a tissue composed of multiple concentric layers of very flattened epithelial cells7 which produce a highimpedance barrier between the endoneurium and the surrounding tissue. Within the spinal canal, the perineurium becomes the arachnoid membrane and enlarges to encompass not individual fascicles, but the entire spinal cord. Thus, individual spinal rootlets are not encased in a high-impedance membrane beyond the level of the spinal ganglia. The lack of a high-impedance natural membrane Intrafascicular Electrodes 1061 FIGURE 7. Representative sample of an implanted fascicle without a cuff. Duration of implant was six months. The electrode „arrow… has been encapsulated by connective tissue. There is an apparent decrease in axonal density in the immediate vicinity of the electrode, but the rest of the fascicle appears unaffected. Calibration bar: 100 m m. around the spinal rootlets is what appears to necessitate the use of an insulative nerve cuff for recording with polyLIFEs. Unfortunately, compression or other damage of the neural tissue from the cuffs resulted in tissue necrosis and loss of nerve fibers when used on a chronic basis. Therefore, successful chronic recording from dorsal rootlets with any form of intrafascicular electrode requires that the problem of providing a biocompatible rootlet cuff be solved. On the other hand, use of polyLIFEs to stimulate spinal roots or to record from peripheral nerve fascicles would not be faced with this problem. If the tendency toward segregation of motor and cutaneous sensory afferent activity that we saw in our recordings can be confirmed, it may assist in extracting somatosensory information reliably. While combining activity from multiple sensory units reduces the number of independent channels of information available, grouping the activity of multiple units which carry coherent information could provide redundancy in estimation of parameters like muscle tension by effectively sampling from a larger afferent population in the muscle.6 The loss of conductivity seen in some of the chronically implanted polyLIFEs appears to be due to delamination of the conductive layer from its substrate. We have found that careful attention to surface preparation and the details of the metal deposition process provides for better and more consistent adhesion. That coupled with care in avoiding stress risers during electrode implantation should alleviate this mode of failure. Biocompatibility Histological examination of the tissue indicated a high degree of biocompatibility for the polyLIFE implants FIGURE 8. Pooled distribution of axon diameters for the combined control group „light bars… and tissue implanted without cuffs „dark bars…. Format as in Fig. 4. There is no statistically significant difference „ X 252.22, df517, and p50.999, proximally; X 2521.0, df517, and p50.226, at the implant level; and X 252.14, df517, and p50.999, distally… at any level. when insulative cuffs are not used. Although the lack of an insulator precludes recording in dorsal rootlets, biocompatibility is an important consideration for use of these electrodes in stimulation applications, where a cuff is not needed, and for recording in peripheral nerves. Biocompatibility of the components used in the polyLIFEs aids in minimizing the formation of an encapsulation layer. Silicone has been shown to be well tolerated when implanted in nervous tissue.11 To the best of our knowledge Kevlar® has never previously been used for implantation in nerve, but it is chemically very stable, and is neither cytotoxic nor mutagenic in cultured bacterial and mammalian cells.1,8,9,14,20 Platinized electrodes have been shown to be biocompatible during longterm implantation in peripheral nerve.10 The biocompatibility of polyethylene has been well established through its use as a surgical implant material 1062 MALMSTROM, MCNAUGHTON, and HORCH for nearly 50 years.16 Although fascicles fitted with cuffs at and distal to the implant showed signs of severe axonal atrophy, the mean distance from the electrodes to the five nearest viable Aa fibers was not significantly different from the implanted fascicles without cuffs, and lower than the reported encapsulation thickness for implanted metal LIFEs.10 The close proximity of viable axons to the electrode indicate that the cause of the damage is not likely due to the presence of the electrode. Rather, the damage appears to be mechanical in nature extending from an external source ~i.e., compression from the cuff!. This is supported by observed regions of focal demyelination and the absence of viable axons in four of the implanted fascicles with cuffs. These reactions are consistent with descriptions of nerve degeneration due to compression.18 Failure to record from these electrodes over the long term can, therefore, be attributed to a lack of neural activity, rather than failure of the electrodes themselves. Cuffs have long been employed in studies of peripheral nerve and from these studies it is known that careful design and placement is critical to their function.15 Cuffs must fit tightly enough about the nerve to accomplish their insulative function but not so tight as to compress the nerve. Employing self-sizing cuffs, minimizing disruption of local blood supply and choosing an implant site which undergoes little relative motion with surrounding tissue are all important considerations. The polyethylene film we used to construct nerve cuffs in this study was extremely thin and flexible and performed well in the acute experiments, yet appears to have contributed to compression injury of the rootlets in the long term. Spinal rootlets are far more delicate than fascicles in peripheral nerves; their small size and lack of significant epineurial or perineurial supportive tissue make achieving the balance between fit and compression even more difficult than in the periphery. The similarity in appearance of the control tissue and the tissue in proximity to implanted fascicles suggests that the implanted polyLIFEs did not adversely affect neighboring neural tissue. Moreover, the lack of a difference in the axonal diameter distributions between the combined control group and the implanted tissue without cuffs suggests one of two things: either implantation of a polyLIFE within a dorsal root fascicle without a cuff does not lead to significant atrophy of axonal fibers, or this technique is not sensitive enough to detect the pathology that does occur. Strictly speaking, one would need to know axonal density and the distribution of fiber diameters within a single fascicle before and after implantation to answer this question directly.5 However, we have no way of collecting the ‘‘before’’ information without sectioning the rootlet, and the variability in the number and density of axons from one rootlet to the next makes arguments on the basis of population statistics limited in their sensitivity. Nonetheless, the data presented here are consistent with the idea that the increase in fibrous connective tissue around the implant produces a local decrease in axonal density, but no statistically significant change in other characteristics of the nerve fibers in the fascicle. Unlike the clear pathology seen in the cuffed fascicles, there is little evidence of measurable change in the population of dorsal root fibers in an uncuffed fascicle in response to implantation of a polyLIFE. On this basis, we suggest that polyLIFEs are well tolerated by the nervous tissue. In conclusion, this study has shown that polyLIFEs can be used to record from feline dorsal rootlets, provided that an insulative cuff is placed around the fascicle. The polyLIFEs are biocompatible and appear to be viable for chronic implantations in either central or peripheral fasciculated nerve fibers. However, improvements in cuff design and proper strain relief are needed in order to monitor spinal rootlet activity on a chronic basis. ACKNOWLEDGMENTS The authors thank Dr. P. R. Burgess and Dr. V. Hlady for review of early versions of the manuscript, Dr. R. A. Normann for loan of equipment used in the surgery, and Dr. K. Yoshida for technical contributions to construction of the electrodes. This work was supported by a grant from NINDS of NIH. APPENDIX var lower, upper, n, I, pic1, pic2: integer; macro ‘Initialize’ ~Note: Initializes image.! begin Open~‘’!; SetOptions~‘Area’!; IncludeInteriorHoles~true!; SetPrecision~2!; SetScale~4, ‘um’, 1.0!; SetBackgroundColor~0!; SetForegroundColor~255!; SetSaveAs~‘TIFF’!; end; macro ‘Measure ROI ~Region of Interest!’ ~Note: After ‘Initialize’ and before ‘Measure ROI’, the operator selects an area ~ROI! to be measured using the ***** tool.! begin Copy; DrawBoundary; Intrafascicular Electrodes SetParticleSize~20,6000!; SetOptions~‘Area; Perimeter’!; AnalyzeParticles~‘outline’, ‘ignore’, ‘include’!; SetCounter~n!; SetDensitySlice~250,254!; AnalyzeParticles~‘ignore’, ‘include’!; SetDensitySlice~0,0!; AddConstant~-1!; SetThreshold~-1!; AutoThreshold; MakeBinary; SetThreshold~-1!; SaveAs~‘Temp2’!; pic1:5PicNumber; Open~‘’!; pic2:5PicNumber; ImageMath~‘add’, pic1, pic2, 1, 0, ‘Result’!; AddConstant~-30!; SetForegroundColor~252!; SelectAll; Clear; RestoreRoi; Paste; Measure; KillRoi; SaveAs~‘temp’!; Dispose; end; macro ‘Get Total Area’ ~Note: After measuring a number of images, ‘Get Total Area’ saves the results to a file.! begin CopyResults; ResetCounter; NewTextWindow~‘Area’!; Paste; SaveAs~‘Area’!; end; macro ‘Find Edges’ ~Note: ‘Find Edges’ lets the operator choose the image containing axons to be measured and locates the boundary between the myelin sheath and the axoplasm. This process works best for images with good contrast.! begin Open~‘’!; n:5rCount; AddConstant~0!; Filter~‘find edges’!; AutoThreshold; SetForegroundColor~0!; end; macro ‘Add Background’ ~Note: After ‘Find Edges’ and before ‘Add Background’ the operator manually sets the threshold level.! begin MakeBinary; pic1:5PicNumber; Open~‘’!; AddConstant~-30!; pic2:5PicNumber; ImageMath~‘add’, pic1, pic2, 1, 0, ‘Result’!; end; macro Measure Axons’ ~Note: After ‘Add Background’ and before ‘Measure Axons’ the operator edits the image to help the program recognize valid axons.! begin MakeBinary; Skeletonize; SetThreshold~-1!; 1063 end; macro ‘Measure Missed Axons’ ~Note: The original image is superimposed on the image of recognized axons. The operator manually outlines in black the axons that were missed.! begin SetThreshold~250!; MakeBinary; Skeletonize; AnalyzeParticles~‘ignore’, ‘include’, ‘outline’!; SetThreshold~-1!; CopyResults; NewTextWindow~‘Result’!; Paste; SaveAs~‘Results’!; Dispose; DisposeAll; end; REFERENCES 1 DuPont, Kevlar fiber products: Material Safety Data Sheet. Wilmington, DE: DuPont, 1989. 2 Goodall, E. V., and K. W. Horch. Separation of action potentials in multi-unit intrafascicular recordings. IEEE Trans. Biomed. Eng. 39:289–295, 1992. 3 Goodall, E. V., K.W. Horch, T. G. McNaughton, and C. M. Lybbert. Analysis of single-unit firing patterns in multi-unit intrafascicular recordings. Med. Biol. Eng. Comput. 31:257– 267, 1993. 4 Goodall, E. V., T. M. Lefurge, and K. W. Horch. Information contained in sensory nerve recordings made with intrafascicular electrodes. IEEE Trans. Biomed. Eng. 38:846–850, 1991. 5 Horch, K. W., and S. J. W. Lisney. On the number and nature of regenerating myelinated axons after lesions of cu- 1064 MALMSTROM, MCNAUGHTON, and HORCH taneous nerves in the cat. J. Physiol. (London) 313:275–286, 1981. 6 Horcholle-Bossavit, G., L. Jami, J. Petit, R. Vejsada, and D. Zytnicki. Ensemble discharge from Golgi tendon organs of cat peroneous tertius muscle. J. Neurophysiol. 64:813–821, 1990. 7 Krstic, R. V. General Histology of the Mammal. New York: Springer, 1985. 8 Lee, K. P., D. P. Kelly, and G. L. Kennedy, Jr. Pulmonary response to inhaled aramid synthetic fibers in rats. Toxicol. Appl. Pharmacol. 71:242–253, 1983. 9 Lee, K. P., D. P. Kelly, F. O. O’Neal, J. C. Stadler, and G. L. Kennedy, Jr. Lung responses to ultrafine Kevlar aramid synthetic fibrils following 2-year exposure in rats. Fundam. Appl. Toxicol. 11:1–20, 1988. 10 Lefurge, T., E. Goodall, K. Horch, L. Stensaas, and A. Schoenberg. Chronically implanted intrafascicular recording electrodes. Ann. Biomed. Eng. 19:197–207, 1991. 11 Loeb, G. E., M. J. Bak, and J. Duysens. Long-term unit recording from somatosensory neurons in the spinal ganglia of the freely walking cat. Science 197:1192–1194, 1977. 12 Malagodi, M., K. W. Horch, and A. A. Schoenberg. An intrafascicular electrode for recording of action potentials in peripheral nerves. Ann. Biomed. Eng. 17:397–410, 1989. 13 McNaughton, T. G., and K. W. Horch. Action potential classification with dual channel intrafascicular electrodes. IEEE Trans. Biomed. Eng. 41:609–616, 1994. 14 McNaughton, T. G., and K. W. Horch. Metallized polymer fibers as leadwires and intrafascicular microelectrodes. J. Neurosci. Methods 70:103–110, 1996. 15 Naples, G. G., J. T. Mortimer, and T. G. H. Yuen. Overview of peripheral nerve electrode design and implantation. In: Neural Prostheses, Fundamental Studies, edited by W. F. Agnew and D. B. McCreery. Englewood Cliffs, NJ: Prentice–Hall, 1990, pp. 107–145. 16 Rubin, L. R. Polyethylene as a bone and cartilage substitute: a 32-year retrospective. In: Biomaterials in Reconstructive Surgery, edited by L. R. Rubin. St. Louis: Mosby, 1983, pp. 474–493. 17 Schmidt, S., K. Horch, and R. Normann. Biocompatibility of silicon-based electrode arrays implanted in feline cortical tissue. J. Biomed. Mater. Res. 27:1393–1399, 1993. 18 Sunderland, S. Nerves and Nerve Injuries. New York: Churchill Livingstone, 1978. 19 Wall, E. J., M. S. Cohen, J. B. Massie, B. Rydevik, and S. R. Garfin. Cauda equina anatomy I: Intrathecal nerve root organization. Spine ~Philadelphia! 15:1244–1247, 1990. 20 Wening, J. V., H. Marquardt, A. Katzer, K. H. Jungbluth, and H. Marquardt. Cytotoxicity and mutagenicity of Kelvar®: an in vitro evaluation. Biomaterials 16:337–340, 1995. 21 Yoshida, K., and K. Horch. Closed-loop control of ankle position using muscle afferent feedback with functional neuromuscular stimulation. IEEE Trans. Biomed. Eng. 43:167– 176, 1996. 22 Yoshida, K., and K. Horch. Selective stimulation of peripheral nerve fibers using dual intrafascicular electrodes. IEEE Trans. Biomed. Eng. 40:492–494, 1993.