This action might not be possible to undo. Are you sure you want to continue?
IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 16, NO. 3, JUNE 2008
Automatic Synchronization of Functional Electrical Stimulation and Robotic Assisted Treadmill Training
Mark E. Dohring, Member, IEEE, and Janis J. Daly
Abstract—This work presents a means to automatically synchronize two promising gait training technologies to address gait deﬁcits in stroke survivors: functional electrical stimulation using intramuscular electrodes (FES-IM) and the Lokomat robotic gait orthosis. A system of hardware and software was developed to achieve the automatic synchronization. A series of bench tests were performed to verify the feasibility and reliability of automatic synchronization. The bench tests showed that automatic synchronization of FES-IM to the Lokomat gait cycle was feasible and reliable. Automatic synchronization was more consistent than manually triggered stimulation (10-fold smaller standard deviation of latency), and produced no early or missed stimulations across 634 strides. Automatic synchronization had greater accuracy than manually triggered stimulation, producing stimulation timed to an accuracy of 2.5% of one gait cycle duration (heel strike ). to heel strike
Index Terms—Automation, functional electrical stimulation (FES), gait, neuromuscular stimulation, robots.
can be a provider of passive movement in the swing phase limb, and passive movement is not a desirable motor learning strategy. A second disadvantage could be that the Lokomat produced abnormal muscle activation timings in healthy controls . In contrast, FES-IM can provide normal muscle activation timings. In fact, the combination of Lokomat and FES-IM promises a number of advantages including early practice of close to normal swing phase movement patterns accompanied by electrically induced muscle contractions at the proper timing for swing and stance phases of gait. We investigated manually initiating the FES-IM gait stance phase in synchrony with the Lokomat heel strike. This has proven feasible, but not ideal, since it is subject to human error. Therefore, the purpose of this research was to test the feasibility and reliability of automatically synchronizing FES gait pattern onset with the Lokomat gait robot heel strike. II. METHODS In testing the feasibility, reliability, and consistency of automatic synchronization of FES-IM and the gait robot, the principle questions guiding the testing were: 1) feasibility of automatic synchronization, i.e., that latency must be short enough to ensure that the stance muscles are activated prior to loading more than 50% of body weight onto the stance limb. Otherwise, stance phase knee control practice would not be practiced during limb loading. According to published data for healthy adults , 50% of body weight is accepted after 5% of the gait cycle. Therefore, the acceptable latency must be less than or equal to 5% of the gait; 2) reliability of stimulus activation for each step, i.e., will stimulation occur at each and every step of the gait robot; and 3) consistency of stimulation activation onset, i.e., will the variance of stimulation start times be acceptable (at least as good as manual stimulation). A. FES-IM Technology Stimulation patterns were developed which activated eight, targeted muscles in the proper sequence for the gait pattern . The stimulation gait patterns were handcrafted using an iterative process beginning with templates and customized for individual patient gait deﬁcits. In the current research, a typical patient stimulation pattern was selected for synchronizing with the gait robot. The Lokomat produced a pulse at each right heel strike that was used to synchronize the FES to the robot. This pulse was routed to the Universal External Control Unit (UECU), which has external input/output lines that can be used in the software to trigger stimulation.
ORLDWIDE, 15 million people per year have a stroke. Five million per year are permanently disabled. Up to half of all stroke survivors do not regain functional independence. At six months poststroke, 30% are unable to walk without another person’s physical assistance. Conventional gait training does not restore normal gait in many stroke survivors. There are two promising, new gait-training methods. First, gait robot assist is promising in that it provides passive swing phase robotics assist, as well as weight support and robotic assistance during stance phase , . Second, functional electrical stimulation (FES) with intramuscular (IM) electrodes (FES-IM) is promising in that it is clinically and statistically signiﬁcantly advantageous in gait restoration compared with a comparable comprehensive gait-training program without FES ,  . Each has disadvantages that are compensated by the other. One disadvantage of the FES-IM gait system is that it did not produce an absolutely normal swing phase limb ﬂexion pattern for all subjects . One disadvantage of the Lokomat is that it
Manuscript received July 12, 2007; revised December 12, 2007; accepted December 30, 2007. This work was supported by the Department of Veterans Affairs, Ofﬁce of Rehabilitation Research and Development under Grant B4036I and Grant B5080S. M. E. Dohring is with the Louis Stokes Cleveland Department of Veterans Affairs Medical Center, Cleveland, OH 44106 USA (e-mail: markdohring@ieee. org). J. J. Daly is with the Louis Stokes Cleveland Department of Veterans Affairs Medical Center, Cleveland, OH 44106 USA and with the Department of Neurology, Case Western Reserve University School of Medicine, Cleveland, OH 44106 USA (e-mail: email@example.com). Digital Object Identiﬁer 10.1109/TNSRE.2008.920081
1534-4320/$25.00 © 2008 IEEE
respectively. according to trigger recognition. The maximum trigger recognition latencies were 43. 2) First Pulse Latency: Fig. Table I shows ﬁrst pulse latencies for automatic stimulation. respectively. for 118 of the ranged from 20 to 25 ms. The latency was uniformly distributed over a narrow range of 25. Automated stimulation had a latency that was less variable than manual stimulation (10-fold smaller standard deviation.4% of the steps.7% of the steps. Our analysis of the outcome was ulation pulse. the latency of 68 ms for signal transmission was only 2. Total latency had means of 68. for comtency. Since stance phase extended to 60% of the gait cycle.1 ms and Fig. The minimum latencies for trigger recognition were 17. 2 shows a comparison of automatic and manual stimulation. . For example.8 and 97. t . B. The la. We recorded the time instants of the trigger pulses from the Lokomat. Shows the trigger recognition latency.3 to 97.8 ms. almost all of stance phase and all of swing practice could be performed with combined FES-IM and Lokomat. The short contributes to the feasibility of combined duration of FES-IM and Lokomat gait training. referenced to the Lokomat trigger pulses at the beginning of stance phase of the right leg.1 ms. clinician 1 was later than automatic triggering for 4. i. which was the majority of steps. Latency 1) Trigger Recognition Latency: Fig. when the UECU recognized the trigger signal.3 ms) contribute to the feasibility of automatically synchronized FES-IM and Lokomat gait training. 1) Analysis: Data from two 30-min test trials were recorded and analyzed to determine how quickly the UECU could respond to the Lokomat right heel strike gait event.4 and 30.5% of the entire gait cycle. In addition. Equations (1) and (2) show the method of calculation of two latencies: the trigger recognition latency. The latency met the criterion of being less than 5% of the gait cycle. and for 130 of the strides. the time from the right heel strike trigger pulse from the Lokomat to the recogni. the time from the right heel strike pulse to the delivery of the ﬁrst stimulation pulse. Dtpulse. and ranges of from 40. Fig. RESULTS AND DISCUSSION Automatic synchronization of FES gait patterns and the Lokomat was feasible and more accurate and repeatable than manually delivered electrical stimulation patterns during Lokomat use.0% of the time. Table II).e. i.5 and 17.DOHRING AND DALY: AUTOMATIC SYNCHRONIZATION OF FUNCTIONAL ELECTRICAL STIMULATION 311 B. for the two test runs.4 ms across strides for the two trials. Each clinician initiated stimulation manually during a 30-min test run as he/she would during an actual training session using FES-IM and the Lokomat simultaneously. This was strictly bench testing. the UECU initiated the stimulation pattern for each step without missing any and without any . 1 1 TABLE I FIRST PULSE LATENCY from 41. for two test runs using automatic synchronization of FES to the Lokomat gait cycle. A.8% of the time. latency. t tency ranged from 25 to 30 ms. and when the UECU delivered the ﬁrst stim. and so there was no subject in the Lokomat during testing. respectively. the system was bench tested while connected to the Lokomat during a 30-min test run.2 ms. the lastrides.. therefore. respectively.3 and 43.8 s. (1) (2) 2) Comparison to Manual Initiation of Stimulation: Manual synchronization of FES-IM and the Lokomat was tested by two clinicians. and clinician 2 was early 22. This means that they initiated stimulation prior to the Lokomat gait event of right heel strike. III. 1. The mean trigger recognition latencies were 30. The time for the Lokomat gait cycle was 2.. FES Delivery Reliability In two trials of 634 steps each. There was no accumulation of delay over the 30-min runs. respectively.7 and 69. .5 ms. during walk trial 1 (solid line). The narrow range of this latency and the short duration (30. Note that both clinicians sometimes initiated stimulation early. and the ﬁrst pulse latency. Quality Control Bench Testing To test the latency between the desired stimulation onset and the actual stimulation onset of the newly developed UECU application software. Table II shows trigger recognition latency for automatic and manual stimulation.e. 1 shows a plot of the distribution of the trigger recognition latency on the left. was calculated for each clinician run. and clinician 2 was later than automatic triggering for 64. Clinician 1 was early with stimulation 92. parison to automatic stimulation.8 ms. and the ﬁrst pulse tion of the trigger by the UECU. 1 also shows a plot of the distribution of the ﬁrst pulse latencies for the two test trials.
“Feasibility of combining multi-channel functional neuromuscular stimulation with weight-supported treadmill training. 307–314. resulting in active muscle contractions contributing to close-to-normal coordinated movement components of gait. Within the bench testing that was conducted. The clinical signiﬁcance is that actual muscle activations can be FES-induced in concert with the robotics-induced passive movement of the lower limbs. The Biomechanics and Motor Control of Human Gait: Normal. Hidler and A. and E. ON. K. A. Jan.5–43. Manual stimulation was 99. and Pathological. 3(b)]. “Effects of locomotion training with assistance of a robot-driven gait orthosis in hemiparetic patients after stroke: A randomized controlled pilot study. Feb. “Prospective. Winter. respectively. Fredrickson. 3(a)] represent the programmed stimulation pattern and the dots represent actual stimulation pulses generated by the UECU during one stride.” Neurorehabil. manual synchronization. Shows a comparison of automatic vs. as deﬁned by the Lokomat trigger pulse. Daly. E.” Clin. McCabe. 2006. 2005. Oct. 105–115. J. 2007. The solid lines [Fig. pp. no. L. Rogers. the integration of automatically triggered FES-IM with the gait robot was feasible. J. Neural Repair. Biomech.. 412–418. J. An 8-channel FES system was automatically synchronized with a gait robot. 2004. Automatic stimulation was 100% reliable. Heller. “A randomized controlled trial of functional neuromuscular stimulation in chronic stroke subjects. Sci. vol. Future work could include testing the integration of FES-IM and Lokomat at different gait speeds. “Intra-limb coordination deﬁcit in stroke survivors and response to treatment. pp. JUNE 2008 Fig. Ruff. FES Delivery Integrity Fig. F. two clinicians missed 2 and 9 stimulation initiations. Matzak. Limitations Gait speed of the Lokomat was constrained at 1. vol. Marsolais. we would use the capability of adjusting the overall pattern timing that was built into our application software. In two trials of 636 steps each.” J. 20. S. 2. 16. TABLE II MANUAL AND AUTOMATIC TRIGGER RECOGNITION LATENCY D. 1991. M. Elderly. Ruff. Wall. Roenigk. 37. IV. vol. K. E.” Stroke. pp. 349–354. Mayr. no. VOL. Waterloo. and L. To our knowledge there is no other currently available method to provide this type of gait practice with these practice advantages.–Aug. Quirbach. pp. 184–193. Koenig. Neurol. Daly and R.5% of the gait cycle. J. “Alterations in muscle activation patterns during robotic-assisted walking. it should pose no problem for synchronizing to other gait speeds. 3 shows the muscle stimulation activations for all eight channels [Fig. Shows desired FES-IM stimulation pattern (solid lines) and actual pulses (points) delivered for one gait cycle plotted in relation to the joint angles at the hip and knee joints of the Lokomat. 0 Fig. Time 0 is right heel strike. The occasional missed initiation of stimulation would not adversely affect training. Gansen. CONCLUSION extraneous initiations. 2.” Stroke.  J. To use our system at different gait speeds. Butler. Mar. Jul. and M. Roenigk. . 3(a)] plotted against the joint angles at the hip and knee of the right leg [Fig. accurate. J. Fredrickson. Feb. blinded. 2. H. 1. E. Canada: Waterloo Biomechanics. REFERENCES  A. Saltuari. vol. Koﬂer. 2006. Holcomb. J. vol. Krewer.” Gait Posture. Dohring. vol. 4. 255.58% reliable for two runs with two different clinicians. no. Daly. randomized crossover study of gait rehabilitation in stroke patients using the lokomat gait orthosis. NO. 21.  J. respectively. Note that the latency for the automatic test runs occupies a very narrow time window (17. 2007. pp.5 km/h for this study.  B. 25.312 IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING. M. Since this capability was already used to adjust pattern timing for our tests. 3. The delivered correspond to the desired pattern to within 2.  D.  J. J. 172–178. 3. and R.69% and 98. K. E. These levels of reliability are sufﬁcient for the clinical use of FES with the Lokomat. The displayed gait cycle is from right heel strike to subsequent right heel strike. C. L. E. C. K. E. no.5 ms) compared to the latencies of the manual test runs ( 330:3 to 287. B. 38. K. Sng. and reliable. M. Husemann. Frohlich..4 ms). Muller. This research developed a new sophistication in technology capability for gait training.  J. pp.
D. Associate Director. OH. OH. FES Center of Excellence. and the Ph. Cleveland. for which she leads a team of interdisciplinary members who study cognitive control and function. Research Career Scientist. Currently. Akron.A. in 2004 as a Research Biomedical Engineer.DOHRING AND DALY: AUTOMATIC SYNCHRONIZATION OF FUNCTIONAL ELECTRICAL STIMULATION 313 Mark E Dohring received the Ph. the M. Cognitive and Motor Learning Research Program. He worked as a post-doc at Case Western Reserve University until taking a research position at the Louis Stokes Cleveland Department of Veterans Affairs Medical Center. OH. Cleveland. degree in physical therapy from Case Western Reserve University (CWRU). Department of Neurology. and develop innovative motor learning methods and technology applications that improve gait and upper limb function for stroke survivors. degree in psychology from University of Akron. OH. and Director. LS Cleveland V. she holds the following appointments: Associate Professor.S. Daly received the B. Medical Center. Janis J. degree from Case Western Reserve University. Cleveland. OH. including FES and robotics to the problem of motor relearning to improve gait and upper limb function in stroke survivors. in 2002.S.D. Oberlin. CWRU School of Medicine. . degree in biology from Oberlin College. His research interests include the application of novel technologies.