504 OPTICS LETTERS / Vol. 30, No. 5 / March 1, 2005

Optical stimulation of neural tissue in vivo

Jonathon Wells, Chris Kao, Karthik Mariappan, Jeffrey Albea, E. Duco Jansen, Peter Konrad, andAnita Mahadevan-Jansen

Departments of Biomedical Engineering and Neurosurgery, Vanderbilt University, 2201 West End Avenue, Nashville, Tennessee 37235

Received August 18, 2004

For more than a century, the traditional method of stimulating neural activity has been based on electricalmethods, and it remains the gold standard to date. We report a technological breakthrough in neural acti-vation in which low-level, pulsed infrared laser light is used to elicit compound nerve and muscle potentialsin mammalian peripheral nerve in vivo. Optically induced neural action potentials are spatially precise,artifact free, and damage free and are generated by use of energies well below tissue ablation threshold.Thus optical stimulation presents a simple yet novel approach to contact-free in vivo neural activation thathas major implications for clinical neurosurgery, basic neurophysiology, and neuroscience. © 2005 OpticalSociety of America

OCIS codes: 170.3890, 140.3460.

Since the 1800s, whether one is studying single cellsor modifying nerve cell function in human beingsduring clinical procedures, the standard for activatingneurons and understanding how they work has beenbased on electrical methods.1 –3 However, limitationsof electrical stimulation include electrical interferencefrom the environment, the presence of high-frequencyartifacts associated with the stimulation signal thatlimit data analysis and prevent simultaneous stimu-lation and recording of adjacent areas, intrinsicdamage caused by direct contact with the electrodesused for stimulation, a population response to therecruitment of multiple axons, and, in general, poorspatial specificity.2,4,5

It is well known that an action potential can betriggered in neurons by use of many stimuli includingelectrical, magnetic, mechanical, thermal, chemical,and optical means. To circumvent the limitationsof electrical stimulation, several groups of scientistshave pursued mechanical methods of stimulation,including the use of ultrasound and magnetic en-ergy.6,7 Optical stimulation of a neural bundle by ashort-pulsed ultraviolet excimer laser, with energiesat the tissue-damage threshold, was reported.8 Morerecently, the activation of cultured neurons by mul-tiphoton excitation with a femtosecond laser wasreported.9 However, we are aware of no publishedreports of the almost simplistic concept of using pulsedinfrared light to stimulate neural potentials.

Here we demonstrate a fundamentally novel ap-proach to neural activation in which low-level, pulsedinfrared laser energy can be used to consistently andreproducibly stimulate peripheral nerves. This laserexcitation results in compound nerve and musclepotentials in frogs and rats in vivo with no appre-ciable tissue damage by use of radiant exposureswell below damage threshold. The results describedhere validate our finding that optical stimulationcan circumvent the many limitations of electricalstimulation, including lack of spatial specificity andthe presence of electrical artifacts.

Neural stimulation is essential in fundamentalneurophysiological studies as well as in a numberof clinical applications. Electrical stimulation isutilized in applications such as identification of the

0146-9592/05/050504-03$15.00/0 ©

connectivity and funtionality of specific nerve roots toguide resection10 and in cortical surface mapping toguide brain-tumor resection.11,12 Here we describe acontact-free, damage-free, spatially selective, artifact-free method that may have significant advantagescompared with electrical methods for a potpourri ofdiagnostic and therapeutic clinical applications.

Proof-of-concept studies were initially performedin vivo on the sciatic nerve of a leopard frog and weresubsequently validated on Sprague–Dawley rats; pro-tocols approved by the Institutional Animal Care andUse Committee were used. In all results shown here,electrical stimulation (0.3–0.6 V for 5 ms) served asthe gold standard for comparison. The main trunk ofthe sciatic nerve was exposed, and nerves were stimu-lated proximally to the first nerve branch point on thenerve trunk from which an isolated muscle responsewas elicited, to demonstrate spatial selectivity in themuscle response. Subsequent compound nerve actionpotentials (CNAPs) and compound muscle actionpotentials (CMAPs) were consistently observed andrecorded by conventional electrical recording methods(MP100; Biopac Systems, Santa Barbara, California).

Figure 1 shows a direct comparison of the CNAPsand measured CMAPs measured by use of optical stim-ulation and electrical stimulation with pulsed infraredlight at 4 mm. Similar recordings were obtained ata variety of wavelengths from 2 to 10 mm with twodifferent laser sources. The similarity in shape andtiming of the signals from optical and electrical stimulias illustrated in Fig. 1 shows that conduction veloci-ties, represented by the time between the CNAP andthe CMAP, are equal, implying that measured actionpotentials are identical regardless of the mechanismof activation. Of particular importance is the factthat the stimulation artifact that is typically linkedwith electrical methods is not observed in the opticallyelicited nerve and muscle responses, thus facilitatingsimulatneous stimulation and recording from adjacentportions of a neuron. A consistently evoked responsewas recorded from an optically isolated nerve, indicat-ing that the incident light is directly responsible for theCNAP and CMAP observed. These results explicitlyprove that pulsed infrared laser energy incidentupon a sciatic nerve can generate an artifact-free,

2005 Optical Society of America

March 1, 2005 / Vol. 30, No. 5 / OPTICS LETTERS 505

Fig. 1. Compound nerve and muscle action potentialsrecorded from a sciatic nerve in a frog in vivo: (a) CNAPrecorded by use of electrical stimulation �l � 4.0 mm�,(b) CNAP from optical stimulation, (c) hamstringCMAP recorded by use of electrical optical stimulation�l � 4.0 mm�, (d) hamstring CMAP by optical stimulation.

physiologically valid CNAP, which propagates to theinnervated muscle and results in a visible muscletwitch that can be recorded by electromyography.

In general, CNAP and CMAP represent populationresponses to stimulation that comprise individualall-or-none responses from constituent axons.13,14 Wehave shown that these individual axons can beselectively stimulated to produce isolated, specificmuscle contractions with optical neural activation,which is a major advantage associated exclusivelywith this modality. The precision permits selectiverecruitment of motor axons within a nerve, as can beobserved in a comparison of the relative magnitudesof nerve and muscle potentials elicited from opticaland electrical stimulation (Fig. 1).

To characterize the dependence of optical neuralstimulation on laser wavelength and radiant expo-sures and to find optimal laser parameters that areclinically relevant, we used the Vanderbilt medicalfree-electron laser (FEL) as a continuously tunable�2 10-mm�, pulsed infrared laser source.15 The FELwas used to irradiate four different nerves for threetrials each at six wavelengths, i.e., 2.1, 3.0, 4.0, 4.5,5.0, and 6.1 mm, at or near relative peaks and valleysof tissue absorption.16 For each nerve the stimulationthreshold was determined as the minimum radiantexposure for a visible CMAP to occur with each laserpulse. The variance in stimulation threshold wassmall, resulting in consistent stimulation across mul-tiple samples. Adjacent to the stimulation site, an

ablation threshold was determined as the minimumradiant exposure required for cavitation to occur withten pulses delivered at 2 Hz.

The stimulation threshold exhibits a wavelengthdependence that mirrors the inverse of a soft-tissueabsorption curve (as measured with Fourier-transformspectroscopy). A more useful indicator of optimalwavelengths for achieving damage-free stimulationis the safety ratio, defined as the ratio of thresholdradiant exposure for ablation to that for stimulation,that identifies spectral regions with a large marginbetween excitation and damage radiant exposures.Figure 2 illustrates that the highest safety ratiosare obtained at 2.1 and 4.0 mm, which correspond torelative valleys in tissue absorption and equivalentabsorption coeff icients.

Although the FEL is an excellent source from whichto gather experimental data, it is neither easy to usenor clinically practical. There are few commerciallyavailable lasers that emit light at 4.0 mm; however, theholmium:YAG (Ho:YAG) laser operating at 2.12 mmis a commercial medical-grade laser that is used for avariety of clinical applications. Although the pulsestructure of this laser differs from that of the FEL, thislight can be delivered via optical fibers, facilitatingthe clinical utility of this laser. The Ho:YAG laserwas successfully used for neural stimulation, withan average stimulation threshold radiant exposureof 0.32 J�cm2 and an associated ablation thresholdof 2.0 J�cm2 �n � 10�, yielding a safety ratio of 6.25(Fig. 2, far left). From the results we can infer thatoptimal stimulation will not occur at peaks in tissueabsorption because the stimulation thresholds areroughly equal to the damage thresholds at thesewavelengths. We can also predict that absolutevalleys of absorption (i.e., the visible–near-IR region,300–1500 nm) will not yield optimal wavelengths, asthe increased penetration depth and tissue scatteringwill cause insufficient energy to be delivered to thenerve fibers for an elicited response. Thus the mostappropriate wavelengths for stimulation of peripheralnerves occur at relative valleys in IR soft-tissueabsorption, which produce an optical penetrationdepth of several hundred micrometers (thickness of

Fig. 2. Safety ratio of stimulation by use of the FEL aswell as of the Ho:YAG laser. The solid curve denoteswater absorption as determined by Fourier-transform IRspectroscopy.

506 OPTICS LETTERS / Vol. 30, No. 5 / March 1, 2005

Fig. 3. Histological samples (H&E stain) of rat sciaticnerve tissue (excision immediately after stimulation).The irradiated zones are marked by striped boxes: Left,normal tissue (sham) without laser irradiation; center,stimulation site irradiated by ten pulses at l � 2.12 mmand 0.65 J�cm2; right, control lesion irradiated by 20pulses at l � 2.12 mm and 2.5 J�cm2.

the protective, collagenous sheath surrounding nerve),thus reaching the axons within these nerve bundles.

We performed a histological analysis of the ex-cised nerves, both immediately after laser excitationand 3–5 days after stimulation, to assess andquantify the damage accrued within the nerve tis-sue following optical excitation with the Ho:YAGlaser. In the former studies, radiant exposureswell over the ablation threshold were used to gen-erate a damage zone in the rat sciatic nerve in vivofor positive control, with an adjacent stimulationzone with ten laser pulses used for various radiantexposures above stimulation threshold �n � 10�. Insurvival studies, the muscle and skin were suturedfollowing stimulation and the animal was followed fora period of 3–5 days to facilitate assessment of anydelayed neuronal damage and Wallerian degeneration�n � 10�. A sham procedure with no stimulationwas performed in the contralateral leg as a negativecontrol for all procedures. Irradiation sites weremarked with ink and sent for independent and blindhistological analysis.17,18 Indications of thermal dam-age include collagen hyalinization, birefringence loss,disruption and vacuolization of the myelin sheath,and severing of nerves.

Figure 3 shows sample histological images (staincombination of hematoxylin and eosin, commonlyreferred to as H&E stain) of the rat sciatic nervefrom the acute experiments. The extent of histo-logical damage at the site of optical stimulation wasquantified in each acute specimen, which indicatesthat all ten nerves studied showed no sign of thermaltissue damage at the site of stimulation with radiantexposures as much as two times the stimulationthreshold. Histological examination of the nervesfrom the survival study showed no damage to thenerve or the surrounding perineurium in eight of theten specimens, with damage occurring at radiant ex-posures above 2.5 times threshold. These histologi-cal f indings suggest that nerves can be consistentlystimulated by optical means at or near thresholdwithout causing any neural tissue damage.

Whereas the results described here clearly show theeffectiveness of optical stimulation, the mechanismfor this effect is unknown at this time. Possiblemechanisms include photothermal, photomechanical,and photochemical triggers or a combination thereof.The relationship of the stimulation threshold to tissueabsorption indicates that photochemistry is likely not

responsible for this phenomenon, a supposition thatneeds validation. We hypothesize that the basis ofoptical activation is laser-induced temperature tran-sients, which trigger the activation of transmembraneion channels. Studies to elucidate likely mechanismsof optical stimulation are in progress.

A novel approach to neural stimulation by use ofpulsed infrared lasers at radiant exposures well belowtissue-damage threshold is presented. This contact-free, damage-free, artifact-free, spatially specif icstimulation modality has the potential to change thefuture of electrophysiology in the laboratory as well asin medical settings. With the emergence of compactand economical solid-state lasers, the construction of aself-contained, hand-held device for optical stimulationof nerves during clinical procedures and even animplantable device for future neuroprosthetics maysoon become a reality.

We acknowledge the support of the W. M. KeckFoundation Free Electron Laser Center as well ofthe Vanderbilt Medical Free Electron Laser program(grant FA9550-04-1-0045). J. Wells’s e-mail addressis jonathon.d.wells@vanderbilt.edu.

References

1. G. Fritsch and E. Hitzig, Arch. Anat. Physiol. Wiss.Med. 37, 300 (1870).

2. L. A. Geddes and J. D. Bourland, Med. Biol. Eng. Com-put. 23, 131 (1985).

3. J. K. Song, B. Abou-Khalil, and P. E. Konrad, J. Neurol.Neurosurg. Psych. 74, 561 (2003).

4. D. Burke, M. R. Nuwer, J. Daube, C. Fischer,J. Schramm, C. D. Yingling, and S. J. Jones, Elec-troencephal. Clin. Neurophysiol. Suppl. 52, 133 (1999).

5. K. C. McGill, K. L. Cummins, L. J. Dorfman, B. B.Berlizot, K. Leutkemeyer, D. G. Nishimura, andB. Widrow, IEEE Trans. Biomed. Eng. 29, 129 (1982).

6. S. I. Norton, Biomed. Eng. Online 2, 6 (2003).7. T. Wagner, M. Gangitano, R. Romero, H. Theoret,

M. Kobayashi, D. Anschel, J. Ives, N. Cuffin,D. Schomer, and A. Pascual-Leone, Neurosci. Lett.354, 91 (2004).

8. G. Allegre, S. Avrillier, and D. Albe-Fessard, Neurosci.Lett. 180, 261 (1994).

9. H. Hirase, N. Volodymyr, J. Goldberg, and R. Yuste,J. Neurobiol. 51, 237 (2002).

10. H. Ueno, K. Kaneko, T. Taguchi, Y. Fuchigami, H.Fujimoto, and S. Kawai, Int. Orthop. 24, 361 (2001).

11. F. E. Roux and M. Tremoulet, J. Neurosurg. 97, 857(2002).

12. P. A. Starr, Stereotact. Funct. Neurosurg. 79, 118(2002).

13. L. A. Geddes and J. D. Bourland, IEEE. Trans. Biomed.Eng. 32, 458 (1985).

14. E. R. Kandel, J. H. Schwartz, and T. M. Jessell, eds.,Principles of Neural Science, 4th ed. (McGraw-Hill,New York, 2000).

15. G. S. Edwards and R. H. Austin, Rev. Sci. Instrum. 74,3207 (2003).

16. G. M. Hale and M. R. Querry, Appl. Opt. 12, 555 (1973).17. S. Thomsen, in Laser-Induced Interstitial Thermother-

apy, G. Mueller and A. Roggen, eds., SPIE PressMonograph PM25 (SPIE, Bellingham, Wash., 1995),pp. 459–467.

18. S. Thomsen, Photochem. Photobiol. 53, 825 (1991).