Electrical Stimulation of Eye Blink in Individuals with Acute Facial Palsy: Progress toward a Bionic Blink

Abstract

Background:

Elicitation of eye closure and other movements via electrical stimulation may provide effective treatment for facial paralysis. The authors performed a human feasibility study to determine whether transcutaneous neural stimulation can elicit a blink in individuals with acute facial palsy and to obtain feedback from participants regarding the tolerability of surface electrical stimulation for daily blink restoration.

Methods:

Forty individuals with acute unilateral facial paralysis, HB grades 4 through 6, were prospectively studied between 6 and 60 days of onset. Unilateral stimulation of zygomatic facial nerve branches to elicit eye blink was achieved with brief bipolar, charge-balanced pulse trains, delivered transcutaneously by adhesive electrode placement; results were recorded on a high-speed video camera. The relationship between stimulation parameters and cutaneous sensation was analyzed using the Wong-Baker Faces Pain Rating Scale.

Results:

Complete eye closure was achieved in 55 percent of participants using stimulation parameters reported as tolerable. In those individuals, initial eye twitch was observed at an average current of 4.6 mA (±1.7; average pulse width of 0.7 ms, 100 to 150 Hz), with complete closure requiring a mean of 7.2 mA (±2.6).

Conclusions:

Transcutaneous facial nerve stimulation may artificially elicit eye blink in a majority of patients with acute facial paralysis. Although individuals varied widely in their reported degrees of discomfort from blink-eliciting stimulation, most of them indicated that such stimulation would be tolerable if it could restore eye closure. These patients would therefore benefit from a biomimetic device to facilitate eye closure until the recovery process is complete.

CLINICAL QUESTION/LEVEL OF EVIDENCE:

Therapeutic, IV.

Eye blink and smile are critically important facial movements, and have traditionally represented primary targets for functional restoration in patients experiencing facial paralysis, with eye closure generally regarded as the top functional priority.¹ Surgical manipulation of the periocular complex does provide benefit²; however, it does not restore high-quality, synchronous, dynamic movement, and it remains invasive.

Because the vast majority of facial paralysis is unilateral, and facial expressions are typically symmetric, movements of the nonparalyzed side may be used to initiate corresponding movements of the paralyzed side. A biomimetic device may be designed to detect movements on the healthy side of the face, and drive activation of contralateral paralyzed muscles, to elicit symmetric facial expressions. The goal of our research is to develop biomimetic applications that complement or replace surgical facial reanimation procedures by means of facial pacing technology.

Several studies have explored the feasibility of devices to artificially stimulate eye blink and smile, spanning from rudimentary conceptual work to indwelling stimulating electrodes.^3–13 Herein, we studied a series of patients with acute facial paralysis by delivering transcutaneous facial nerve stimulation to induce eye closure. We recorded the motor response to electrical stimulation and described the relationship between electrical stimulation parameters and subjective discomfort. The enrolled participants represent the target population for the external blink restoration system we aim to develop.

PATIENTS AND METHODS

Volunteers were recruited prospectively from the adult patient population (age range 25 to 77 years, mean 49.6 years) who visited the Facial Nerve Center at the Massachusetts Eye and Ear Infirmary for acute facial paralysis. Inclusion criteria were: paralysis for a minimum of 6 days and a maximum of 6 weeks, incomplete eye closure, and an intact contralateral facial nerve. Individuals with complete eye closure on the paretic side, with an implanted electronic device (e.g., pacemaker, defibrillator, deep brain stimulator), or who had undergone periocular surgery or chemodenervation were excluded from the study.

A total of 40 patients (18 males and 22 females) were enrolled in the study from June of 2012 to December of 2013. Informed consent was obtained under the Institutional Review Board protocol 12–024H, which was approved by the Massachusetts Eye and Ear Infirmary Human Studies Committee.

Testing began by mapping stimulated blink response from a range of electrode contact points. Response mapping was performed by using a hand-held bipolar stimulating probe (Model 9523–7; Carefusion, Inc., San Diego, Calif.). Pulses were delivered to the lateral periocular skin via two electrodes separated by 2 cm. The cathode was positioned 1 cm lateral to the orbital rim. The anode was moved in an arc around the cathode, spanning 90 degrees (30 degrees above to 60 degrees below the cathode) in 10-degree increments. The electrode position that produced the strongest contraction of the orbicularis oculi muscle was selected for further testing. For subjects in whom mapping failed to identify a site of blink stimulation on the paralyzed side, mapping was performed on the healthy side, and a mirror image of the best stimulation location was used on the paralyzed side, assuming symmetric facial nerve branching patterns. Electrodes used for blink stimulation were disposable adhesive Ag/Cl pads trimmed to an 80-mm² area (reference 019–400400, VIASYS; Carefusion). The cathode was placed 1 cm lateral to the orbital rim (as done in the mapping phase), and the anode was positioned in the location that was determined previously (Fig. 1).

Fig. 1.

Experimental set-up. Two epicutaneous electrodes were placed on the paralyzed side of the face, above the facial nerve branches to the orbicularis oculi muscle.

Pulse train parameters were set using software (Clampex 10.2; Molecular Devices, LLC, Sunnyvale, Calif.), generated by a high-precision digital-to-analog converter (DigiData 1440A; Molecular Devices, LLC), and amplified by a constant-current stimulator (STMISOL; BIOPAC Systems, Inc., Goleta, Calif.). Preliminary electrical stimulation experiments have been performed in anatomically intact human volunteers, indicating that the best stimulation pattern for eliciting natural-appearing movement of the eyelids without discomfort is 100-to 200-Hz pulse trains.¹⁴ Similar train characteristics, with shorter intervals between the first two pulses,¹⁵ were used in the present study.

Stimulation pulse trains for mapping ranged from 100- to 150-Hz frequency at 3- to 5-mA amplitude, with a pulse width of 1 ms.

Pulse trains used for blink stimulation ranged from 0.4- to 1-ms pulse width, 100 to 150 Hz, and 1 to 15 mA. All trains were bipolar, and charge density was limited to 0.5 μC per mm² per pulse. Pulse trains were repeated every second, beginning at a subthreshold amplitude. With each repeated stimulation train, the amplitude increased by 0.5 mA. Pulse train repetitions ended for each stimulation location when either (1) the highest stimulation level (15 mA) was reached, (2) complete eyelid closure was achieved, or (3) the subject reported discomfort and wanted the stimulation series to stop.

The relationship between stimulation parameters and cutaneous sensation was continuously evaluated using the Wong-Baker Faces Pain Rating Scale.¹⁶ Participants controlled a hand-held device that allowed them to indicate level of discomfort as stimulation intensity increased, in real time, on a scale of 0 to 10. The device sent an output voltage to the data acquisition system indicating the degree of discomfort so that subject pain scale responses could be saved in data files along with delivered pulse train characteristics and the video synchronization transistor–transistor logic signal (Fig. 2).

Fig. 2.

The experimental setup is shown, depicting both the blink stimulation pathway and multiple data sources. Blink stimulation originated from the data acquisition system and used a constant-current stimulator with a ratio of 1:10 for input volts to output milliamps of current. The delivered current was monitored by measuring voltage across a 100-ohm resistor located between the stimulator and the participant’s face. Additional data sources included digital video recordings at 30 or 1000 frames per second (synchronized with facial stimulation pulse train recordings via output pulses from the camera) and a variable voltage level from the Faces Pain Scale switch (see Patients and Methods), which indicated the reported pain level throughout testing. Participants looked at a large version of the Wong-Baker Faces Pain Scale that was positioned approximately 2 meters in front of them during blink stimulation testing, which illuminated the particular face corresponding to their reported pain level on a handheld six-position switch they controlled. LED, lightemitting diode; TTL, transistor–transistor logic.

In addition, the kinematics of elicited blinks were studied (Fig. 2). A single, suprathreshold pulse train was delivered through the same electrodes used for blink stimulation. A high-speed video system recorded the eye at 1000 frames per second (240 × 256 pixel CCD camera model 25105100; KayPentax B/W System, Montvale, N.J.). Individual video frames were analyzed using ImageJ software (available at http://rsbweb.nih.gov/ij/download.html) to track palpebral fissure height in relation to iris diameter over time (temporal resolution of 10 ms per image). Blink characteristics, including maximal closure velocity, were measured for paralyzed eyelids with and without electrical stimulation and for the recorded spontaneous blinks of the healthy eye (see Video, Supplemental Digital Content 1, which demonstrates an attempted blink, http://links.lww.com/PRS/B417; and Video, Supplemental Digital Content 2, which demonstrates a stimulated blink, http://links.lww.com/PRS/B418). A paired two samples for means t test was used to compare kinematics of bionic blinks on the paralyzed side with spontaneous blinks on the healthy side:

Finally, spontaneous recovery and any occurrence of synkinesis in participants were evaluated at 12 weeks and also at average 12-month follow-up.

RESULTS

Forty individuals with unilateral (16 right, 24 left) facial paralysis were enrolled in the study. They each participated only once, on the day of initial consultation, between 6 and 58 days (mean 20.2 days) from the onset of paralysis. Etiologies included Bell’s palsy in 29 cases, Lyme disease in four cases, Ramsay Hunt syndrome in four cases, temporal bone fracture in two cases, and autoimmune disease in one case (see Table 2).

Table 2.

Demographics, Characteristics of the Paralysis, Motor Response to Electrical Stimulation, and Recovery

Case No.	Sex	Age (yr)	Days of Denervation	HB Score	Side	Etiology	Elicited Blink	Full Blink at 12 Weeks?	Full Blink at 12 Months?	Synkinesis
1	F	59	14	6	L	Bell’s	Full	No	Yes	Yes
2	F	49	4	5	L	Bell’s	Full	N/A	N/A	N/A
3	M	37	17	4	L	Lyme	Full	No	No	No
4	M	39	10	6	R	Ramsay Hunt	Full	Yes	N/A	No
5	M	77	21	4	L	Bell’s	Full	Yes	N/A	No
6	M	25	17	5	L	Bell’s	Full	N/A	N/A	N/A
7	F	28	4	5	R	Bell’s	Full	N/A	N/A	N/A
8	M	39	10	6	R	Bell’s	Full	N/A	N/A	N/A
9	M	71	28	5	L	Bell’s	Full	Yes	Yes	No
10	M	63	7	5	R	Bell’s	Full	Yes	N/A	No
11	M	42	7	4	R	Bell’s	Full	Yes	N/A	No
12	F	34	14	5	L	Bell’s	Full	Yes	N/A	No
13	F	69	5	4	R	Bell’s	Full	N/A	N/A	N/A
14	M	42	35	4	L	Ramsay Hunt	Full	No	N/A	Yes
15	F	37	20	6	L	Bell’s	Full	N/A	N/A	N/A
16	M	55	7	7	L	Bell’s	Full	Yes	Yes	No
17	M	29	6	4	L	Bell’s	Full	Yes	N/A	No
18	F	25	6	5	R	Lyme	Full	Yes	Yes	No
19	M	35	58	4	R	trauma	Full	Yes	N/A	No
20	M	63	5	4	L	Bell’s	Full	Yes	N/A	N/A
21	F	33	12	5	L	Bell’s	Full	No	Yes	No
22	M	51	12	5	L	Bell’s	Full	Yes	N/A	No
23	F	42	17	4	R	Bell’s	Partial	No	Yes	No
24	F	56	47	5	L	Lyme	Partial	Yes	Yes	Yes
25	M	63	38	6	L	Bell’s	Partial	Yes	Yes	Yes
26	F	52	17	6	L	Bell’s	Partial	No	No	No
27	F	52	20	6	L	Bell’s	Partial	No	N/A	N/A
28	F	47	10	5	R	Bell’s	Partial	No	No	Yes
29	M	60	22	5	L	Bell’s	Partial	Yes	Yes	No
30	F	69	5	4	R	Bell’s	Partial	No	No	Yes
31	F	65	9	6	R	Autoimmune	Partial	No	Yes	Yes
32	F	53	18	5	R	Lyme	Partial	N/A	N/A	N/A
33	M	71	29	6	L	Bell’s after trauma	Partial	No	No	Yes
34	F	68	60	5	L	Bell’s	Partial	No	No	Yes
35	M	45	21	5	R	Bell’s	Partial	N/A	N/A	N/A
36	F	37	20	5	R	Ramsay Hunt	Partial	No	N/A	N/A
37	F	41	21	4	L	Bell’s	None	Yes	Yes	Yes
38	F	55	25	6	R	Ramsay Hunt	None	Yes	Yes	No
39	M	66	57	6	L	Bell’s	None	No	No	No
40	F	35	55	5	L	Bell’s	None	No	No	Yes

HB, House-Brackmann; F, female; M, male; L, left; R, right; N/A, not applicable.

Preliminary mapping across the lateral periorbital area localized a cutaneous stimulation site in 36 cases. In four cases, the position for testing electrodes was determined based on mapping of the contralateral healthy facial nerve, as described above. Transcutaneous electrical stimulation of the motor branches to the paralyzed orbicularis oculi elicited complete eye closure in 22 (55 percent) of the 40 participants. In 14 cases (35 percent), only partial closure was achieved; two of these participants aborted the trial at an amplitude of 8 mA because of discomfort. In four cases (10 percent), no motor response was observed; these were the four participants with no observed movement during mapping.

Twenty-two individuals received varied pulse width stimulation from 0.4 to 1 ms. Average current thresholds for all four pulse widths fell within the range of 3.57 to 5.71 mA for eye twitch and 6.2 to 7.8 mA for complete eyelid closure. The mean current threshold value was 4.6 mA for all twitch responses and 7.2 mA for all eye blinks in this subset. The inverse relationship between pulse width and current threshold for elicited eye twitches and blinks in the 22 participants who achieved complete eye closure is represented in Figure 3.

Fig. 3.

Average current thresholds (milliamperes) for eye twitch and blink at different pulse durations (milliseconds) in the 22 participants who reached complete eye closure.

Participants did not start reporting pain or discomfort before reaching the twitch response current threshold. Six of the 22 individuals who received four pulse widths did not have any reportable pain throughout stimulation. In the pain scale analysis table (Table 1), the values of these six individuals were included in the percentage column but excluded from the average threshold calculation so as not to skew the data. Of the remaining 16, average current thresholds to elicit blinks fell within the reported average pain score range of 2 to 4. The relationship between pulse width and reported pain is detailed in Table 1.

Table 1.

Pain Scale Report^*

Pulse Duration (ms)	Participants Reporting Pain (%)	Average Current Level (mA)
Pain level 2
0.4	55	6.88
0.6	70	6.56
0.8	70	5.73
1	62	5.19
Pain level 4
0.4	36	7.4
0.6	61	7.37
0.8	61	7.31
1	50	6.67
Pain level 6
0.4	18	8.1
0.6	39	9.1
0.8	39	8.5
1	42	8.1
Pain level 8
0.4	14	8.3
0.6	22	9.12
0.8	22	8.54
1	15	8.175
Pain level 10
0.4	0	N/A
0.6	0	N/A
0.8	4	7.6
1	4	12.2

^*Percentages of individuals reporting pain and average amplitude values corresponding with different pulse widths. Pain level 0 (no pain) is not reported in this table.

High-speed video recordings followed blink stimulation in cases in which a complete eyelid closure was observed (see Video, Supplemental Digital Content 1, which demonstrates an attempted blink, http://links.lww.com/PRS/B417; and Video, Supplemental Digital Content 2, which demonstrates a stimulated blink, http://links.lww.com/PRS/B418). The time frame between the trigger signal and the onset of the movement was 4.5 to 14.5 ms (mean, 9.9 ms). The duration of the down phase of the blink was 55 to 100 ms (mean, 95 ms), and the lid descent velocity was 0.06 to 0.12 meters per second (mean, 0.08 meters per second). On the paralyzed side, the down phase duration of spontaneous upper eyelid down movement, if present, was 103 to 240 ms (mean, 141 ms), and lid descent velocity was 0.003 to 0.02 meters per second (mean, 0.01 meters per second). With regard to the healthy eye, the duration of the down phase of the blink was 87 to 193 ms (mean, 120 ms), and lid descent velocity was 0.04 to 0.08 meters per second (mean, 0.06 meters per second). The upper eyelid speed difference between the paralyzed side bionic blink and the healthy side spontaneous blink was not statistically significant.

Finally, 32 participants were evaluated at 12-week follow-up; complete eye closure was observed in 75 percent (12 of 16) of participants who previously had complete eye blink response to electrical stimulation, in 25 percent (three of 12) of those with incomplete eye blink, and in 50 percent (two of four) of those with no motor response. Twenty participants were also followed up long term (average, 12 months); full eye closure was observed in 83 percent (five of six) of participants with complete eye blink response to electrical stimulation, 50 percent (five of 10) of those with incomplete eye blink, and 50 percent (two of four) of those with no motor response. Synkinesis was observed in 13 percent (two of 15) of participants who previously had complete eye blink response to electrical stimulation, in 70 percent (seven of 10) of those with incomplete eye blink, and in 50 percent (two of four) of those with no motor response. These results are detailed in Table 2.

DISCUSSION

Facial paralysis causes tremendous deficits in facial function and social interactions. Because of the profound effect of this disorder on patients’ quality of life, a great deal of effort has been focused on rehabilitation of the paralyzed face. The loss of eye blink is one important aspect of facial paralysis, and has thus been the first target of our research.

A thorough understanding of the subtleties of different clinical presentations is paramount for initiation of appropriate therapy. For the purposes of treatment planning, patients with facial paralysis may be classified by category:

1. Recent acute onset with prognosis for complete recovery within 3 months, i.e., Bell’s palsy;

2. Recent acute onset with prognosis for complete recovery over 4 to 12 months, i.e., Lyme disease or longitudinal temporal bone fracture;

3. Acute onset with an injured but intact or grafted facial nerve with prognosis for prolonged, incomplete recovery, e.g., acoustic neuroma;

4. Chronic paralysis with severely injured or extirpated facial nerve and no recovery over the course of 12 months, e.g., parotid gland malignancy; or

5. Chronic paralysis of congenital or childhood origin, e.g., Möbius syndrome or traumatic forceps delivery.

Individuals within the first three categories would benefit from an external biomimetic device to facilitate eye closure during waking hours, until the recovery process is complete. The last two groups of patients would require more conventional treatment regimens, such as lid loading or periocular reanimation surgery with free or regional muscle transfer, and do not represent the target of our present research. However, facial pacing could potentially be applied to neuromuscular reconstructions to provide aid (training feedback, functional augmentation, coordination regulation). Ultimately, engineered muscle tissue and implantable electrodes with movement detection systems for the nonparalyzed side may supplant microneurovascular surgery as the accepted standard for facial reanimation in these patients.

Our research aims to develop a biomimetic application that might partially replace facial reanimation procedures by means of real-time facial pacing technology. Because spontaneous mimetic movements are mainly symmetric, in cases of unilateral paralysis, a biomimetic device may be able to record and process a signal when the healthy hemiface spontaneously moves. The signal recorded from the healthy side may then be processed in real time to create a train of electrical pulses that would initiate a specific movement on the paralyzed side. Prosthetically assisting eye blink, and eventually other facial movements, using noninvasive means, would provide several advantages over current therapy.

With regard to the surgical rehabilitation of acute paralytic lagophthalmos, eyelid springs and weights may lead to ptosis and are sometimes visible as masses beneath the skin surface. In many cases, bulky gold weights have been replaced by lower-profile platinum implants, but they are still often perceptible as upper eyelid contour irregularities. In addition to the cosmetic drawback, indwelling foreign bodies increase the risk of postoperative infections, particularly in diabetic patients¹⁷ or patients receiving chemotherapy, may ultimately extrude through the thin skin of the eyelid, or they may cause astigmatism from pressure on the cornea. Moreover, if orbicularis oculi muscle function ultimately returns, the weight will need to be removed or the tarsorrhaphy will need to be reversed. In cases of acute reversible paralytic lagophthalmos, the application of an external biomimetic device would provide immediate restoration of eye closure and could be used until either the patient recovers sufficient function to no longer require assistance for eye closure, or the decision is made to proceed with further surgery. Also, the development of an eye blink-triggering mechanism would likely decrease the risk of corneal exposure and its attendant morbidity. Finally, there would be substantial cosmetic and functional advantages to real-time pacing for symmetric and synchronous movements.

In a preliminary study on a group of six individuals affected by facial paralysis, interferential stimulation at 30 Hz between 6 months and 15 years from symptom onset evoked a motor response below the maximal discomfort threshold in all patients tested.¹⁸ Herein, we have studied a series of individuals with acute, reversible facial paralysis, representing the target population for the system we envision. In this study, we explored both feasibility and tolerability of eliciting eye blink by means of transcutaneous stimulation of the facial nerve. Complete eye closure can be elicited in 55 percent of patients affected by acute facial paralysis. Preliminary mapping could be useful in the clinical setting to determine which patients would benefit from facial pacing. This study tested the excitability of each participant’s facial nerve on a single day during the recovery period; a longitudinal study testing individuals multiple times, at different stages of recovery, would provide information about gradual evolution of the recovering nerve response to electrical stimulation. The envisioned pacer might indeed require fine tuning with time (i.e., lowering delivered current levels) corresponding to any gradual increases in excitability of the neuromuscular structures throughout recovery. With regard to the pain report, participants underwent only one test, which does not predict whether they could tolerate daily pacing that would be required to treat facial palsy until recovery. Tolerability of continuous transcutaneous electrical stimulation will be a determinant of patient compliance in wearing the envisioned device.

With regard to the kinematics of bionic eye blinks compared with spontaneous blinking, data show that the average duration of the down phase was 95 ms for artificially induced eye blinks and 120 ms for the contralateral healthy eye. Changes in frequency across the train length have been shown to play a role in the kinematics of bionic eye blinks,^14,15 although the difference between natural and induced eye blinks would likely not be noticeable if closure of both eyes initiated simultaneously. If the recording apparatus introduced a delay while processing the movement signal from the unaffected side, shorter duration of the bionic eye blink stimulus might balance the overall timing and improve the symmetry of blink restoration. The average 10-ms delay measured between stimulation and the onset of movement is important information for the design of closed-loop facial pacers because the overall delay of the circuit should not exceed 33 ms.^19,20

Interestingly, the motor response to electrical stimulation seems to be predictive of the clinical outcome and the occurrence of synkinesis. For example, 13 percent (two of 15 testable) of subjects with full blink response to electrical stimulation developed synkinesis, whereas 64 percent of subjects with only a partial or no stimulable blink did. Unfortunately, individuals lost to follow-up were not randomly distributed because most of the patients who experienced complete recovery did not come for a second follow-up visit. Specifically, 53 percent of subjects who had a full blink at 12 weeks did not return for a 12-month visit, whereas only 20 percent of subjects without a full blink at 12 weeks missed their 12-month follow-up visit. Likewise, 73 percent of the subjects who had a full stimulable blink were lost to the 12-month follow-up point as were 29 percent of those with a partial stimulable blink and 0 percent of subjects with no stimulable blink. Further investigation of the relationship between blink stimulability and functional recovery is warranted.

This study explored the application of surface electrode stimulation as a potential treatment to artificially elicit eye blink in cases of acute facial paralysis. Electrodes were taped above the orbicularis oculi muscle and likely stimulated intramuscular nerve branches. The low current threshold with short duration pulses that we observed is typical for myelinated axons. In principle, these stimulation parameters should not work if the peripheral motor axons are not viable because the threshold for direct excitation of muscle fibers is generally ten-fold higher than axon stimulation. It remains uncertain whether these same stimulation locations and pulse train patterns would be effective in generating eye blinks in chronically paralyzed individuals. Additional experiments are therefore needed to determine whether it is possible to elicit useful movements from completely denervated facial muscles using surface or intramuscular electrodes.

CONCLUSIONS

Development of an eye blink pacing system is on the horizon. Our previous work demonstrated that blinks can be detected noninvasively with optical-electronic hardware mounted on eyeglasses¹³ or surface electromyographic recording.^21,22 The present prospective study suggests that approximately 90 percent of individuals presenting with acute unilateral facial paralysis will have at least some stimulable blink, and that more than half will be able to achieve full eye closure using non-invasive electrical stimulation. Given that both the blink detection and stimulation hardware in our envisioned pacing system would be worn externally, we are hopeful that the apparatus can be clinically available in the near future in order to provide blink reanimation during recovery from acute facial palsy.

Supplementary Material

Attempted Blink with Paralysis

Video 1. Supplemental Digital Content 1 demonstrates an attempted blink, http://links.lww.com/PRS/B417.

Electrically Stimulated Blink

Video 2. Supplemental Digital Content 2 demonstrates a stimulated blink, http://links.lww.com/PRS/B418.