168

IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 14, NO. 2, JUNE 2006

REFERENCES
[1] J. del R. Milln, F. Renkens, J. Mourio, and W. Gerstner, Non-invasive brain-actuated control of a mobile robot by human EEG, IEEE Trans. Biomed. Eng., vol. 51, no. 6, pp. 10261033, Jun. 2004. [2] R. Grave de Peralta, M. Murray, C. Michel, R. Martuzzi, and S. Gonzalez Andino, Electrical neuroimaging based on biophysical constraints, NeuroImage, vol. 21, pp. 527539, 2004. [3] R. Grave de Peralta, S. Gonzalez Andino, L. Perez, P. Ferrez, and J. del R. Milln, Non-invasive estimation of local eld potentials for neuroprosthesis control, Cognitive Process., vol. 6, pp. 5964, 2005. [4] F. Sharbrough, G. Chatrian, R. Lesser, H. Lders, M. Nuwer, and W. Picton, Am. electroencephalogr. society guidelines for standard electrode position nomenclature, Clin. Neurophysiol., vol. 8, pp. 200202, 1991. [5] T. Kohonen, Self-Organising Maps, 2nd ed. Berlin, Germany: Springer-Verlag, 1997. [6] W. Gehring, M. Coles, D. Meyer, and E. Donchin, The error-related negativity: An event-related brain potential accompanying errors, Psychophysiology, vol. 27, p. 34, 1990. [7] M. Falkenstein, J. Hoormann, S. Christ, and J. Hohnsbein, ERP components on reaction errors and their functional signicance: A tutorial, Biol. Psychol., vol. 51, pp. 87107, 2000. [8] C. Holroyd and M. Coles, The neural basis of human error processing: Reinforcement learning, dopamine, and the error-related negativity, Psychol. Rev., vol. 109, pp. 679709, 2002. [9] B. Blankertz, G. Dornhege, C. Schfer, R. Krepki, J. Kohlmorgen, K.-R. Mller, V. Kunzmann, F. Losch, and G. Curio, Boosting bit rates and error detection for the classication of fast-paced motor commands based on single-trial EEG analysis, IEEE Trans. Neural Syst. Rehabil. Eng., vol. 11, no. 2, pp. 127131, Jun. 2003. [10] L. Parra, C. Spence, A. Gerson, and P. Sajda, Response error correctionA demonstration of improved humanmachine performance using real-time EEG monitoring, IEEE Trans. Neural Syst. Rehabil. Eng., vol. 11, no. 2, pp. 173177, Jun. 2003. [11] G. Schalk, J. Wolpaw, D. McFarland, and G. Pfurtscheller, EEG-based communication: Presence of an error potential, Clin. Neurophysiol., vol. 111, pp. 21382144, 2000. [12] P. Ferrez and J. del R. Milln, You are wrong!Automatic detection of interaction errors from brain waves, in Proc. 19th Int. Joint Conf. Articial Intell., 2005. [13] J. del R. Milln, On the need for on-line learning in brain-computer interfaces, in Proc. Int. Joint Conf. Neural Networks, 2004. [14] N. Schraudolph, Local gain adaptation in stochastic gradient descent, in Proc. 9th Int. Conf. Articial Neural Networks, 1999. [15] A. Butteld and J. del R. Milln, Online classier adaptation in brain computer interfaces IDIAP, IDIAP Res. Rep. 06-16, (2006) [Online]. Available: http://www.idiap.ch/publications.php

Prolonged Cognitive Planning Time, Elevated Cognitive Effort, and Relationship to Coordination and Motor Control Following Stroke
Janis J. Daly, Yin Fang, Elizabeth M. Perepezko, Vlodek Siemionow, and Guang H. Yue

AbstractUnderstanding cortical function can provide accurately targeted interventions after stroke. Initially, stroke survivors had prolonged cognitive planning time and elevated cognitive effort, highly correlated with motor control impairments. Exploratory results suggest that neurorehabilitation, accurately targeted to dyscoordination, weakness, and dysfunctional task component execution, can improve cognitive processes controlling motor function. Index TermsBrain, electroencephalography (EEG), motor control, movement-related cortical potential (MRCP), stroke.

I. INTRODUCTION Although a number of variables may inuence central nervous system (CNS) reorganization in response to rehabilitation, it is not only the number of remaining neurons, but how the neurons function that will determine functional capability [1]. Very little is known about cortical function in terms of cortical planning time and cortical effort level during functional task component performance in stroke survivors. With a better understanding of cortical function, we could more accurately target sophisticated interventions for stroke survivors. During the planning and execution of a motor task, a characteristic electroencephalographic (EEG) response is produced, known as the motor-related cortical potential (MRCP). The early components of the MRCP are associated with preparation for movement execution [2][4]. The MRCP planning component has a linear correlation with muscle force and electromyography (EMG) [3] and sensitively differentiates cortical control signals among different types of movement [5][7]. MRCP has excellent time resolution providing important information regarding cognitive planning time. Using the MRCP measure, the two-fold purpose of this study was: 1) determine the effects of stroke on cognitive planning time and cognitive effort level for shoulder/elbow movements used during functional reaching and 2) quantify the relationships for both cognitive planning time and effort level impairments with coordination and motor control decits. Additionally, in a subsample, we explored whether therapy would improve cognitive planning time and effort level. II. METHODS Eighteen subjects were enrolled and tested (ten stroke survivors with persistent (>12 months) arm coordination decits and eight

Manuscript received July 14, 2005; revised March 20, 2006. This work was supported in part by the Department of Veterans Affairs, Rehabilitation Research and Development Service under Grant B2801R and by the National Institute of Health under Grant HD 36725. J. Daly is with the Department of Neurology, Case Western Reserve University School of Medicine, Cleveland, OH 44195 USA. She is also is with the Department of Research, Cleveland VA Medical Center, Cleveland, OH 44106 USA (e-mail: jjd17@case.edu). Y. Fang, V. Siemionow, and G. H. Yue are with the Departments of BME and Rehabilitation Medicine, Cleveland Clinic Foundation, Cleveland, OH 44195 USA. E. M. Perepezko was with the Department of Research, Cleveland VA Medical Center, Cleveland, OH 44195 USA. Digital Object Identier 10.1109/TNSRE.2006.875554 1534-4320/$20.00 2006 IEEE

IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 14, NO. 2, JUNE 2006

169

age-matched, healthy controls). For a subsample of three stroke survivors, motor learning intervention was provided and treatment response was measured. III. TEST PROCEDURES A. Motor Task A motor target task was displayed on a PC that required an accurate 14-cm linear movement in the horizontal plane, from the center of the workspace to a target placed in a direction directly away from the subject, and was accomplished using shoulder exion and elbow extension (50 trials, with 2 min rest/ten trials). B. EEG Data Acquisition Simultaneous data was obtained for EEG, EMG, movement onset (custom goniometric device), and position in the workspace. A NeuroScan system and software (NeuroScan Labs, El Paso, TX) were used to acquire and process surface EEG signals (Quick-Cap Neuromedical Supplies, El Paso, TX; International 1020 System [8], electrode diameter of 8 mm and cavity depth of 5 mm). Impedance at lower than 10 000
was obtained prior to data collection [3], [5]. The active electrodes were referenced to the common linked left and right mastoid surface electrodes. EEG signals were amplied (275 000), low-pass ltered (0100 Hz), and digitized (500 sample/s). C. Motor-Related Cortical Potential (MRCP) Raw signal recordings were visually inspected to detect artifacts induced by eye blinks, facial muscle contractions, or head movements and removed either manually or with software lters; otherwise, the trials were discarded if noise or artifacts were not correctable. Trials containing noise from EMG signals were discarded [9]. MRCP was derived from movement onset trigger-averaged EEG signals (50 trials per subject). The EEG signal included an 8-s window (4 s before and 4 s after the onset event). The mean MRCP start time and amplitude were calculated for each electrode. We dened cognitive effort level as MRCP amplitude (V ; Fig. 1) and cognitive planning time as the duration of time between MRCP onset and EMG onset (Fig. 1). D. EMG The EMG signal was acquired (Grass Telefactor system, Astro-Med, Inc., Rhode Island) using bipolar electrodes (8-mm recording area) applied on the anterior deltoid and triceps, agonist muscles. EMG onset time was dened as: 2 S.D. increase in amplitude above the resting baseline that was maintained for at least 100 ms. The EMG signal was amplied (21000), ltered 102000 Hz, digitized (2000 samples/s), and synchronized with EEG. E. Trajectory Maintenance The InMotion2 shoulderelbow robot and accompanying graphics (InMotion Technologies, Inc., Cambridge, MA) were used to standardize the 14-cm shoulder/elbow task as well as obtain data for position of the arm, using built-in precision potentiometers (0.9 K
/rad) [10][12]. Movement trajectory maintenance was dened as the mean lateral deviation from the desired path (Table I). F. Coordination The FuglMeyer Coordination Scale subtest for shoulder and elbow movements is a sensitive, reliable, and valid measure that assigns a score for ability to move volitionally either dependent upon or independent of total limb exor or extensor synergistic patterns (max points = 28) [13], [14].

Fig. 1. Top graph shows schematic of MRCP, characteristic EEG signal present prior to and during performance of motor task. Vertical axis shows amplitude of signal and horizontal axis represents time (in seconds). First 3 s shows baseline EEG signal. MRCP onset is labeled at 3 s with upward vertical arrow. Vertical distance from baseline to MRCP peak is labeled as MRCP amplitude or cognitive effort level. Bottom graph shows schematic of EMG signal, and EMG onset is labeled at 6 s with downward vertical arrow and associated vertical line (vertical dotted line on right). MRCP onset is indicated across both top and bottom graphs with vertical dotted line on left at 3 s. Duration of cognitive planning time is horizontal solid line between MRCP onset (3 s) and EMG onset (6 s). TABLE I LATERAL DEVIATION OF MOVEMENT FROM STRAIGHT LINE CONNECTING TWO POINTS

IV. TREATMENT AND TEST PROCEDURES FOR SUBSAMPLE A subsample of three subjects received motor learning for a total of 5 h/day, 5 days/week, for 12 weeks. The motor learning intervention included practice of coordinated movements of the shoulder and elbow, based on principles of motor learning [15], and included rest periods to ensure no difculty in participation. Specically, progression of exercises and task component practice was nely incrementalized, according to a hierarchy of increasing difculty of coordination and dexterity. Pre-/post-treatment measures were: cognitive planning time; level of cognitive effort; coordination of shoulder/elbow movements (described above); strength of shoulder/elbow muscles (summed, with max points = 80) [16]; and everyday functional ability, using the shoulder/elbow functional movement components of the Arm Motor Ability Test (AMAT S/E; in seconds), a reliable and valid measure of 13 complex tasks [17]. V. DATA ANALYSIS We performed a two-tailed t-test to compare stroke versus controls regarding cognitive planning time and cognitive effort level. We analyzed a Pearson correlation model to determine the relationship be-

170

IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 14, NO. 2, JUNE 2006

Fig. 2. (a) Vertical axis is planning time in milliseconds and on horizontal axis, cortical electrode location is listed (C3 for sensorimotor area or Fz for frontal area). In both sensorimotor and frontal regions, there was abnormally prolonged cognitive planning time for stroke (shaded bars) versus healthy controls (white bars; p < :05), and in Fig. (2b) vertical axis is amplitude of cortical signal or cognitive effort level. On horizontal axis, cortical electrode location is listed (C3 for sensorimotor area or Fz for frontal area). In both sensorimotor and frontal regions, there was abnormally elevated cognitive effort level for stroke (shaded bars) versus healthy controls (white bars; p < :05).

Fig. 3. (a) Vertical axis is cognitive planning time in milliseconds (ms) and on horizontal axis, points of pre- and post-treatment data acquisition are identied. For three subjects, after treatment, there was signicant improvement in cognitive planning time. In Fig. (3b), vertical axis is amplitude of EEG signal at C3 electrode (microvolts), dening cognitive effort level. On horizontal axis, points of pre- and post-treatment data acquisition are identied. For three subjects, after treatment, there was signicant improvement in cognitive effort level.

tween cognitive motor planning time and movement trajectory maintenance and a Spearman correlation model for cognitive effort level and shoulder/elbow coordination. For pre-/post-treatment comparisons, we generated descriptive statistics and performed a t-test for measures of cognitive planning time and cognitive effort level. VI. RESULTS In comparison to controls, stroke patients exhibited abnormally prolonged cognitive planning time in the sensorimotor [mean 6 S.D.; 2734 6 1205 ms versus 1466 6 779 ms, controls; p = :03; Fig. 2(a)] and frontal [2596 6 1082 ms versus 1511 6 559 ms, controls; p = :04; Fig. 2(a)] regions. Stroke patients had signicantly abnormally elevated amplitude of MRCP in the sensorimotor [8:0 6 2:1 V versus 4:9 6 1:8 V, controls; p = :009; Fig. 2(b)] and frontal (9:8 6 2:4 V versus 5:7 6 1:1 V, controls; p = :011) regions [Fig. 2(b)]. There was a moderately high and signicant association between abnormally prolonged ipsilesional motor and prefrontal region planning time and movement trajectory maintenance (r = :50, p = :048;

r = :52; p = :050, respectively). There was a very high and significant correlation between shoulder/elbow coordination and prefrontal cognitive effort level (r = 0:74; p = :02). There was a signicant pre-/post-treatment reduction toward normal in cognitive motor planning time in the motor area [2214 6 459 ms, 1490 6 589 ms, respectively; p = :050; Fig. 3(a)]. There was a significant post-treatment reduction toward normal in cognitive effort level in the motor area [8:8 6 1:7 V, 6:2 6 0:8 V; p = :050; Fig. 3(b)]. There were mean pre-/post-treatment improvements toward normal in the following measures: trajectory maintenance (34.5 6 21.8 cm; 27.2 6 15.3 cm); coordination (14 6 1.7; 17.3 6 6.8); strength (28.9 6 3.2; 39.8 6 1.1); and every day shoulder/elbow task components (AMAT S/E; 905.9 6 368.2 s; 684.0 6 407.7 s, respectively). VII. DISCUSSION The study extended the literature by providing characterization of EEG signal characteristics in stroke survivors in comparison to healthy controls, quantifying the presence of abnormally prolonged cognitive planning time and elevated cognitive effort level, specically

IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 14, NO. 2, JUNE 2006

171

for a shoulder/elbow movement used during functional reaching. The results indicated that it may be necessary to take into account existing abnormalities of signal characteristics in stroke survivors during braincomputer interface applications of EEG signals. The study extended the literature by providing evidence for the relationship between cognitive function and motor control in stroke survivors performing shoulder/elbow functional reaching. First, ipsilesional motor and prefrontal areas showed a moderate and signicant association between motor planning time and ability to coordinate execution of a desired straight-line movement trajectory. Those who were least able to maintain a desired trajectory, required more time to prepare for the motor task. Second, prefrontal effort level showed a high and signicant association with shoulder/elbow coordination. Those with the greatest difculty performing a coordinated movement generated higher amplitude cortical signal in the prefrontal region, potentially reecting attempts to employ strategies to compensate for perceived inadequate motor coordination. The study extended the literature by providing pilot data that supports a treatment response according to measures of cognitive planning time and effort level. Mean cognitive planning time improved to a normal range. This improvement may reect the simultaneous gains in strength and coordination that occurred. Though there was mean improvement in cognitive effort level, it did not reach normal. At post-treatment, elevated cognitive effort may have been exerted in order to generate compensatory strategies needed in the presence of residual weakness and dyscoordination. These pilot data suggest that EEG measures of cognitive function may be a useful measure of the recovery of cortical processes that control motor function. VIII. CONCLUSION For a functional reach task, stroke survivors had abnormally prolonged cognitive planning time in ipsilesional sensorimotor and frontal areas, as well as elevated cognitive effort level. Both prolonged cognitive planning time and elevated cognitive effort were associated with dyscoordinated movements. The pilot results for the subsample suggested that the EEG-derived measures of cognitive motor planning time and level of cognitive effort effectively discriminated change in CNS activity in response to treatment. After treatment, not only was there an improvement in cognitive planning time and effort level, but also there were simultaneous mean gains in shoulder/elbow muscle strength and coordination and functional task execution. The results suggest that neurorehabilitation can improve the cognitive processes that control motor function when it is accurately targeted and applied.

[7] G. H. Yue, J. Z. Liu, V. Siemionow, V. K. Ranganathan, T. C. Ng, and V. Sahgal, Brain activation during human nger extension and exion movements, Brain Res., vol. 856, no. 12, pp. 291300, Feb. 2000. [8] H. H. Jasper, Report of the committee on methods of clinical examination in electroencephalography, Electroencephalogr. Clin. Neurophysiol., vol. 10, pp. 370375, 1958. [9] Y. Fang, F. Xiong, V. Siemionow, and G. H. Yue, Reduction of muscle activity-induced artifacts from scalp EEG signals, Soc. Neurosci. Abstr., vol. 27, p. 938.10, 2001. [10] W. DeWeerdt and S. M. Crossley, Restoration of balance in stroke patients, Clin. Rehab., vol. 3, pp. 139147, 1989. [11] P. W. Duncan, L. B. Goldstein, D. Matchar, G. W. Divine, and J. Feussner, Measurement of motor recovery after stroke, Stroke, vol. 23, pp. 10841089, Aug. 1992. [12] N. Hogan, H. I. Krebs, A. Sharon, and J. Charnnarong, Inventors: Massachusetts Inst. Technology, Interactive robotic therapist, U.S. patent 5 466 213, Nov. 14, 1995. [13] A. R. Fugl-Meyer and L. Jaasko et al., Post stroke hemiplegic patient, Scand. J. Rehab. Med., vol. 7, no. 1, pp. 1331, 1975. [14] P. W. Duncan, M. Propst, and S. G. Nelsen, Reliability of the FuglMeyer assessment of sensorimotor recovery following cerebrovascular accident, Phys. Ther., vol. 63, no. 10, pp. 16061610, Oct. 1983. [15] C. J. Winstein, Motor learning considerations in stroke rehabilitation, in Stroke Rehabilitation: The Recovery of Motor Control, P. W. Duncan and M. B. Badke, Eds. Chicago, IL: Year Book Medical, 1987, pp. 109134. [16] F. P. Kendall and E. K. McCreary, Muscle: Testing and Function, 3rd ed. Baltimore, MD: Williams and Wilkins, 1983. [17] J. J. Daly, I. Krebs, E. Perepezko, J. Rogers, M. Dohring, K. Goyal, N. Hogan, and R. Ruff, Response to upper limb robotics and functional neuromuscular stimulation following stroke, J. Rehabil. Res. Dev., vol. 42, no. 6, pp. 723736, Dec. 2005.

REFERENCES
[1] B. B. Johansson and P. V. Belichenko, Neuronal plasticity and dendritic spines: Effect of environmental enrichment on intact and postischemic rat brain, J. Cereb. Blood Flow Metab., vol. 22, pp. 8996, Jan. 2002. [2] M. Hallett, Movement-related cortical potentials, Electro-myogr. Clin. Neurophysiol., vol. 34, pp. 513, Jan.Feb. 1994. [3] V. Siemionow, G. H. Yue, V. K. Ranganathan, J. Z. Liu, and V. Sahgal, Relationship between motor activity-related cortical potential and voluntary muscle activation, Exp. Brain Res., vol. 133, pp. 303311, Aug. 2000. [4] L. Deecke et al., Voluntary nger movement in man: Cerebral potentials and theory, Biol. Cybern., vol. 23, pp. 99119, Jul. 1976. [5] Y. Fang, V. Siemionow, V. Sahgal, F. Xiong, and G. H. Yue, Greater movement-related cortical potential during human eccentric versus concentric muscle contractions, J. Neurophysiol., vol. 86, pp. 17641772, Oct. 2001. [6] , Distinct brain activation patterns for human maximal voluntary eccentric and concentric muscle actions, Brain Res., vol. 1023, no. 2, pp. 200212, Oct. 2004.