Logo for the Journal of Rehab R & D
Journal of Rehabilitation Research and Development
Vol. 38 No. 2, March/April 2001

Assistive technology to improve PC interaction for people with intention tremor

Peter Feys; Anders Romberg; Juhani Ruutiainen, MD; Angela Davies-Smith; Rosemary Jones, PhD; Carlo Alberto Avizzano, PhD; Massimo Bergamasco, PhD; Pierre Ketelaer, MD

National Multiple Sclerosis Center, Melsbroek, Belgium; Katholieke Universiteit Leuven, Faculty of Physical Education and Physiotherapy, Department of Kinesiology; Masku Neurological Rehabilitation Center, Masku, Finland; Multiple Sclerosis Unit, Bristol General Hospital, Bristol, UK; Perceptual Robotics (PERCRO), Scuola Superiore St-Anna, University of Pisa, Italy.

This material is based upon work supported by the "TREMOR DE 3216" project funded by the European Community.
Address all correspondence and requests for reprints to: Peter Feys, Faculty of Physical Education and Physiotherapy, Tervuursevest 101 3001 Heverlee (Leuven), Belgium; email:

Abstract — Many patients with upper limb intention tremor encounter difficulties in mouse-driven interaction with the personal computer (PC). An assistive technology system ("the Tremor Control System"), consisting of a motion-filtering software program that supports multiple interfaces, was developed and validated with 36 persons with Multiple Sclerosis in a multi-center trial. PC-tests, requiring basic functions such as cursor placement and click and drag function, were able to differentiate between patients and control subjects (ANOVA: p<0.05). A significant time improvement on the PC-tests was found when using an optimal alternative interface instead of the standard PC-mouse (paired t-tests: p<0.01 for Point & Click test, p<0.05 for Drag & Drop test and p<0.1 for Double Click test). A significant time improvement was found for the Double Click test (paired t-tests: p<0.05) when the motion-filtering program was implemented. The number of patients able to fully perform the PC-tests increased with the Tremor Control System. Patients with marked intention tremor seemed to profit especially from this assistive technology. These users reported that working with the Tremor Control System was less fatiguing and more comfortable compared to the use of the standard PC-mouse.

Key words: assistive technology, intention tremor, Multiple Sclerosis, personal computer, tremor control.


   Multiple Sclerosis (MS) is the most frequent disabling neurological disease in young adults in North America and Western Europe, with an estimated prevalence between 30 and 120 per 100,000 inhabitants (1,2). Intention tremor in the upper limb is encountered in approximately one-third of the MS population (2-4). Intention tremor is defined as an increase in tremor amplitude toward the termination of a visually guided, goal-directed movement (1,5-7). The tremor has a low frequency (3-6 Hz) and tends to worsen with increasing precision requirements (3,5,8-10). The term cerebellar tremor is often used synonymously with intention tremor and is commonly associated with disruptions of the cerebellum or its afferent or efferent pathways (3,5,11). Intention tremor is a largely underestimated cause of disability, probably because it is usually part of a wider clinical picture where strength often is preserved but movement control is affected (12). The functional impact of tremor may be overlooked easily in standard neurological examination. Nevertheless, even the mildest degree of intention tremor may disrupt a patient's handwriting, PC-interaction or independence in grooming and eating. There is, as yet, no satisfactory treatment for intention tremor (2-4).

   Human-machine interaction is becoming essential for full participation in society because of the increased presence of computers in work and everyday life and their important role in communication and information gathering. Graphical user interfaces, such as Microsoft Windows and graphic packages, are standard in personal computers. The ability to successfully use a mouse or other cursor control device is crucial to enabling rehabilitation to gain functional independence for many persons with movements disorders (12-14). However, a considerable number of patients with upper limb intention tremor experience difficulties operating the standard PC-mouse. Therefore, one key objective of the "TREMOR" project was to develop assistive technology to improve mouse-driven screen functions during PC-interaction or, in the future, to control other computerized devices to assist activities of daily life. The aim of this study was to investigate the validity of the "Tremor Control System", which was developed especially for users with upper limb intention tremor.


Description of the Tremor Control System
   The "Tremor Control System" is an assistive technology system that has been developed by Scuola Superiore St-Anna in collaboration with clinical centers. It consists of a motion-filtering program that supports multiple interfaces.

   The motion-filtering program is a signal processing software program that allows different options for filtering the individual features of tremor during mouse-driven PC interaction. The voluntary movement characteristics are extracted, allowing the patient with intention tremor to interact more accurately with PC applications. The program only manipulates the cursor movements on the screen without physically affecting the movement of the interface, i.e., only the relationship between the mouse movement and cursor movement is altered.

   The configurable parameters of the motion-filtering program and the design of the filter, respectively, are shown in Figures 1 and 2.

A graphic of a computer screen showing the configurable parameters in the
motion-filtering program
Figure 1. Computer screen showing the configurable parameters in the motion-filtering program.

A chart showing filter design
Figure 2. Filter design.

   The features of intention tremor during mouse-driven PC-interaction may differ in individual patients. Therefore, the configurable filtering parameters allow the user (therapist and patient) to determine the best filtering options for individual optimum use.

   The interface support option allows simultaneous input from the standard PC-mouse and alternative interfaces so that two individuals, one disabled and one able-bodied can share PC use. The alternative interfaces used in this study are presented in Figure 3. These included a game joystick, a force-control joystick, a trackball (Rollerball), a cordless mouse (unilateral or bilateral use), and a helmet carrying an infrared movement sensor.

A photograph of interface support options
Figure 3. Interface support option: from left to right: game joystick, force-control joystick, trackball, cordless mice (bilateral and unilateral shaped), standard mouse, helmet.

   The Tremor Control System is an easy access interface for computer systems based on the Windows 95/98 operating system. The system allows normal interaction with common multimedia programs available for personal computers such as Internet Browser, graphic packages, text editors, or games.

Patient Selection Criteria and Assessment Methods
   MS patients with upper limb intention tremor were selected from three European Rehabilitation Centers (National Multiple Sclerosis Centre, Melsbroek, Belgium; Masku Neurological Rehabilitation Center, Finland; and the MS Unit of Bristol General Hospital, United Kingdom). All patients gave their informed consent to participate in the trials. Patients with paresis in the upper limb used in the trials, or who manifested visual or mental impairment or an exacerbation of their MS over the month before testing, were excluded. Paresis in the non-tested upper limb was acceptable.

   The clinical assessment to document level of intention tremor consisted of the Finger-to-Nose test, scored according to Fahn's Tremor Rating Scale (15), and the Nine Hole Peg test (16). Fahn's Tremor Rating Scale is a five-point rating scale where zero is no tremor and four is severe tremor. The Nine Hole Peg test is a functional performance test to measure hand dexterity and is commonly used in MS (17,18). The number of pegs (maximally nine) placed within 50 s is counted.

   PC-tests were constructed to measure objectively the mouse-driven PC-interaction. The tests required the basic functions used in controlling an interface, i.e., cursor positioning, single and double click, and drag functions, which imply accuracy and stability in controlling the cursor. Time needed to complete each test was measured and the presence or absence of compensation techniques was observed. The three tests were as follows:

  1. Point & Click test. The subject had to click on three targets on the screen in a predefined order.
  2. Drag & Drop test. Five objects representing files had to be dragged from one directory box into another. The destination box was smaller than the pick-up box, requiring accuracy in placing the transferred files. The number of inaccurate trails (i.e., file placed outside the destination box) was additionally counted.
  3. Double Click test. The patient was asked to open one file by making a double click. The number of "click attempts" was additionally recorded.

   Patient satisfaction with the alternative interface and motion-filtering program (confer "test procedure") was addressed with four questions: 1) "Does the interface make it easy to control the cursor?"; 2) "Can you easily push on the buttons?"; 3) "Does using the system fatigue you?"; and, 4) "Is the system comfortable to use?". Outcome measure was the Visual Analogue Scale ranging from "not at all" (zero) to "fully" (ten).

Test Procedure and Data Analysis
   The user decided which hand was to be used for operating the interfaces. The performance of each patient during mouse-driven PC-interaction was clinically observed. The potential benefit of the Tremor Control System was explored during "trial and error" sessions. Several interfaces were tested and an optimal one chosen for each subject. Configurations of parameters in the motion-filtering program were tried out and adapted to the individual needs of the patient.

   Finally, the patient was evaluated with the PC-tests in three different stages during a separate evaluation session. All PC-tests were first performed with the "standard mouse" (stage 1), secondly with the "optimal interface support" (stage 2) and finally with the "optimal interface support AND optimal configuration of parameters in the motion-filtering program" (stage 3). Control subjects performed the PC-tests using the standard PC-mouse.

   Means and standard deviations (SD) were calculated for all outcome measures. Analysis of variance (ANOVA) was used to investigate whether differences of means were statistically significant. Patients who were unable to perform fully a PC-test in stage 1 were excluded from the data analysis of time-related outcome measures due to lack of an exact numerical value.


Sample and Clinical Assessment
   Thirty-six (17 male, 19 female) MS patients with upper limb intention tremor participated to this multi-center trial. Mean age was 42.9 y (SD=9.8). The results of the clinical assessment are presented in Table 1. One third of the sample, respectively, showed slight, moderate, and marked or severe intention tremor during the Finger-to-Nose test rated with Fahn's Tremor Rating Scale. Eighteen subjects could place 7 to 9 pegs within 50 s during the Nine Hole Peg test, 9 were able to place 4 to 6 pegs and a further 9 subjects placed 0 to 3 pegs.

Table 1.
Clinical assessment of intention tremor.

Fahn's Tremor Rating Scale Nine Hole Peg Test

None 0% 7-9 pegs 50% (n=18)
Slight 34.4% (n=12) 4-6 pegs 25% (n=9)
Moderate 34.4% (n=12) 0-3 pegs 25% (n=9)
Marked 22.9% (n=8)    
Severe 11.4% (n=4)    

PC-Performance Using Standard PC-Mouse
   Results of patient and control groups (16 able-bodied persons, mean age 38.3 y) with the PC-tests, using the standard PC-mouse, are presented in Table 2. Patients who were unable to fully perform the test were excluded from statistical data analysis, as no numerical outcomes were available. These included 3 patients for the Point & Click test, 6 for the Drag & Drop test, and 14 patients for the Double Click test.

Table 2.
Results of patient and control group on PC-assessment using the standard PC-mouse. Patients who were unable to fully perform the test were excluded for data analysis.

  Patients Controls (n=16)

PC-Tests Mean time (SD) Min/max Patients able to fully perform the test Mean time (SD)

Point & Click test 18.1 s (10.8) 5.1-104 33 4.8 s (1.6)
Drag & Drop test 57.8 s (43.4) 17.6-190 30 11.4 s (3.5)
Double Click test 9.8 s (10.5) 1.5-71 22 1.6 s (0.96)

   Mean time to complete the tests for patient and control group, respectively, was 18.1 s and 4.8 s on the Point & Click test, 57.8 s and 11.4 s on the Drag & Drop test, and 9.8 s and 1.6 s on the Double Click test. The difference between the mean performance of patient and control groups was found statistically significant for all PC-tests (ANOVA, t-test; p<0.05).

Choice of Optimal Interface and Motion-filtering Program
   Three patients (8.3 percent) chose the standard mouse as optimal interface, 14 (38.8 percent) the unilateral, and 5 (13.8 percent) the bilateral shaped cordless mouse. Eleven (30.5 percent) chose the trackball, 2 (5.5 percent) the forced-joystick and 1 (2.7 percent) the helmet. Nobody chose the game joystick. The acceptance by subject was addressed with questions concerning the ease of control of the cursor and push on the buttons. Mean outcome was 3.8 and 5.3, respectively, when using the standard mouse, and 6.1 and 7.2, respectively, when using the optimal interface. The difference of means between two conditions was significant for both questions (p<0.01; paired t-test).

   The configuration of parameters differed between individual patients. No significant correlation between the defined configurations and either clinical or PC-assessment was found.

Intervention with the Tremor Control System
   Mean time-performances on the PC-tests during all evaluation stages are shown in Figure 4. Patients who could not fully perform the test using the standard mouse were excluded for statistical data analysis in all stages because of the lack of an exact numerical value with which to refer. Mean time performance (SD) in stages 1, 2, and 3, respectively, was: 18.1 s (10.8), 13.5 s (6.4), and 18.1 s (13.2) on the Point & Click test; 59.4 s (43.4), 44.7 s (24), and 57.9 s (39.1) on the Drag & Drop test; and, 9.8 s (10.4), 6.0 s (4.7), and 5.5 s (3.3) on the Double Click test. A statistically significant improvement in time was found for all PC-tests at stage 2 (paired t-test: p<0.01 for Point & Click, p<0.05 for Drag & Drop and p<0.1 for Double Click test) and for the Double Click test in stage 3 (p<0.05) compared to stage 1.

A graph showing time performance on the PC-tests for all evaluation stages
Figure 4. Time performance on the PC-tests for all evaluation stages.

   Descriptive data concerning the PC-assessment during all evaluation stages are presented in Table 3. The number of patients able to perform fully the PC-tests was higher in stages 2 and 3 compared to stage 1. The number of patients using compensation techniques to decrease intention tremor during PC-interaction decreased in stages 2 and 3 compared to stage 1. Commonly observed techniques were: stabilization of the conducting arm with the other hand and fixation of the upper extremity against the trunk. The number of inaccurate trials during the Drag & Drop test and the number of click attempts during the Double Click test was lower in stage 2 compared to stage 1. An additional decrease of the number of click attempts was observed in stage 3 for the Double Click test.

Table 3.
Number of patients able to fully perform a PC-test and using a compensation technique, total number of inaccurate trails in Drag & Drop test and total number of attempts for the Double Click test.

  Stage 1 Stage 2 Stage 3

Number of patients able to fully perform a PC-test:      
Point & Click test 33 36 36
Drag & Drop test 30 32 33
Double Click test 22 29 31
Number of patients using a compensation technique 17 10 7
Drag & Drop test: Total number of inaccurate trails 95 30 40
Double Click test: Total number of click attempts 162 100 87

   The satisfaction of patients in using the system was assessed by questions concerning fatigue and comfort and was scored on a visual analogue scale. Mean outcomes for stages 1, 2, and 3, respectively, were 4.5, 2.7, and 3.6 for fatigue, and 4.5, 7.1, and 5.5 for comfort. The mean difference between stages 1 and 2 was significant for both questions (paired t-test, p<0.01), as opposed to a lack of significance of the differences between stages 1 and 3.

Further Data Analysis on Subgroups
   Subgroups were assigned based on the clinical performance on the Nine Hole Peg test. Mean time performance of each subgroup on the PC-tests is shown in Figure 5 for all evaluation stages. Due to the small size of the subgroups, differences between means have not been statistically analyzed. The patient group with the worst clinical performance on the Nine Hole Peg test (0-3 pegs placed within 50 s) showed the greatest benefit from using the motion-filtering program in stage 3, for both the Point & Click and Double Click tests. The subgroup with the best clinical performance did not appear to profit from the implementation of the motion-filtering program although choice of an optimal interface (stage 2) did show some benefit.

A graph showing time performance of subgroups on the PC-tests for all evaluation stages
Figure 5. Time performance of subgroups on the PC-tests for all evaluation stages.

   The degree of satisfaction in using the Tremor Control System is shown in Figure 6. The subgroup with worst clinical performance on the Nine Hole Peg test reported the lowest score concerning fatigue and the highest score concerning comfort, with the motion-filtering program enabled in stage 3. The subgroup with the best clinical scores preferred the implementation of the optimal interface only.

A graph showing the degree of satisfaction
Figure 6. Degree of satisfaction.


   Operating the cursor during PC-interaction is a challenging activity for patients with upper limb intention tremor. The aim of assistive technology systems is to improve functional capabilities in individuals with disabilities (19). The objective of this study was to test whether the Tremor Control System was able to improve mouse-driven PC-interaction in persons with upper limb intention tremor (19).

   The evaluation of the capacity of a person to interact properly with the personal computer, using a graphical user interface, is poorly documented in the literature (20). There is a clinical need for an objective assessment of PC-interaction which is highly functional, in contrast to tracking tasks only (12,13,21). The PC-tests used in this study were able to differentiate between non-tremor subjects and patients with intention tremor (20). The difficulties of patients in using accurately the standard PC-mouse was reflected in their longer performance times, which were even underestimated when the data of patients who were unable to fully perform the PC-test were excluded for statistical analysis. Intention tremor is suggested to be dependent on visual information since it has been found to aggravate during visually guided movements (22-24). The use of visual feedback from the cursor representation on the screen is essential during mouse-driven PC-interaction. Similar to observations of the elderly, making a double click with the standard mouse was found to be a complex task. In our sample, this observation was expected since patients with cerebellar deficits have difficulties in making fast, repetitive movements (20,25).

   The selection of an individual, optimal interface improved speed of execution on all PC-tests and decreased the number of inaccurate actions during both click and drag tasks, implying that operation of the cursor was performed more efficiently. The acceptance by the subjects of an alternative interface for the standard mouse was high; its use was less fatiguing and more comfortable. Controlling the cursor and pushing the buttons of the interface was easier, probably because of features such as appropriate position of the buttons, type of conducting mechanism, and size and shape of the chosen interface.

   The implementation of the motion-filtering program improved time performance on most PC tests, but statistical significance was found for the Double Click test only. Further data analyses on subgroups were performed to investigate whether one patient group gained more benefit than another from the motion-filtering program. Ranking the subjects by clinical performance on the Nine Hole Peg test showed that the improvement of time performance seemed to cluster at the lower end of the clinical performance scale. Since more severe tremor is more disabling, any improvement achieved probably implies better functional ability (12). Patients with the worst clinical performance showed the best time performance on most PC-tests when the motion-filtering program was implemented, in contrary to patient groups with better initial clinical performance. Simultaneously, the most severely affected patients reported the highest degree of satisfaction when using the motion-filtering program while patients with slight or moderate tremor experienced no or little benefit. In fact, many found that implementing the software slowed their performance without substantially improving accuracy of cursor control.

   Time measurement might not fully reflect the quality of PC-interaction; therefore, other parameters were taken into account. The number of persons who were able to perform fully the PC-tests with the Tremor Control System increased compared to the use of the standard PC-mouse. It is important to note this, since these patients were not included for statistical data analysis although a time improvement was observed. The quality of PC-interaction improved as operation of the cursor was performed more efficiently and subjects were more relaxed, with less need for additional stabilization of the upper limb.

   The results of this study showed that the movement filtering parameters could be configured to enable improved PC-interaction especially for those with marked intention tremor. The clinical aspects of intention tremor (overshoot, and tremor velocity and amplitude and frequency) were highly variable between patients, and individuals were found to benefit from various combinations of control set-ups. However, no single configuration of parameters was beneficial for all patients. This fact highlights the importance of the flexibility available in the motion-filtering program, allowing a range of configurations to be selected according to individual tremor features, which may change over time. An important advance in the motion-filtering program is the simplification of this configuration phase, which enhances the usability of the system (12,19). Following the initial setting up of parameters, based on clinical assessment, the user is able to alter the parameters using the on-screen software options. Future developments focus on further simplification of the system by employing an automatic configuration procedure, using a mathematical index calculated from the input signal (26).

   An important feature of the motion-filtering system is the potential to share the personal computer with other users, via the standard mouse input. Playing of games and other family, social, or work-related PC-activities can be shared, enhancing the ability of the patient to integrate, and having important psychological benefits (27).

   This study showed the validity of the Tremor Control System for persons with marked intention tremor in the upper limb. Preliminary results of ongoing trials with children with cerebellar ataxia in the upper limb indicate a similar beneficial effect. It is hoped that the Tremor Control System will be available shortly, at modest cost.


   This study demonstrated how persons even with severe level of disabling intention tremor are capable of improving their PC-performance with appropriate technical solutions. The Tremor Control System extended the number of patients with upper limb intention tremor able to interact with the PC and enhanced the quality of operating the cursor. PC interaction was performed faster and more efficiently. Especially patients with marked intention tremor experienced the system as less fatiguing and more comfortable. Health care professionals are challenged to integrate these findings in clinical practice to enhance the functional independence of their patients in man-machine interaction.


   The development of the assistive technology system was supported by the "TREMOR DE 3216" project funded by the European Community. We thank all patients who were willing to participate in the trials. The collaboration of P. Van Asch, chief of the Physiotherapy Department, Melsbroek, during the test trials is highly acknowledged.

  1. Ebers GC, Sadovnick AD. Epidemiology. In: Paty DW, Ebers GC, editors. Multiple Sclerosis. Philadelphia: F.A. Davis Company; 1998. p. 5-28.
  2. Bakheit A. The epidemiology of ataxia in multiple sclerosis. In: Ketelaer P, Ruutiainen J, editors. Ataxia. Genoa: Associazione Italiana Sclerosi Multipla (A.I.S.M.) Publishing; 1995. p. 17-22.
  3. Alusi SH, Glickman S, Aziz TZ, Bain PG. Tremor in Multiple Sclerosis. J Neurol Neurosurg Psychiatry 1999;66(2):131-4.
  4. Jones L, Lewis Y, Harrison J. The effectiveness of occupational therapy and physiotherapy in multiple sclerosis patients with ataxia of the upper limb and trunk. Clin Rehabil 1996;10:277-82.
  5. Deuschl G, Bain P, Brin M. Consensus statement of the Movement Disorder Society on Tremor. Mov Disord 1998;13 Suppl 3:2-23.
  6. Bain P. Clinical measurement of tremor. Mov Disord 1998;13 Suppl 3:77-80.
  7. Topka H, Mescheriakov S, Boose A, Kuntz R, Hertrich I, Seydel L, et al. A cerebellar-like terminal and postural tremor induced in normal man by transcranial magnetic stimulation. Brain 1999;122:1551-62.
  8. Vidailhet M, Jedynak CP, Pollak P, Agid Y. Pathology of symptomatic tremors. Mov Disord 1998;13 Suppl 3:49-54.
  9. Elbe RJ. Central mechanisms of tremor. J Clin Neurophysiol 1996;13(2):133-44.
  10. Hallett M. Overview of human tremor physiology. Mov Disord 1998;13 Suppl 3:43-8.
  11. Bain P. A combined clinical and neurophysiological approach to the study of patients with tremor. J Neurol Neurosurg Psychiatry 1993;69:839-44.
  12. Riley PO, Rosen MJ. Evaluating manual control devices for those with tremor disability. J Rehabil Res Dev 1987;24(2):99-110.
  13. Riviere CN, Thakor NV. Effects of age and disability on tracking tasks with a computer mouse: accuracy and linearity. J Rehabil Res Dev 1996;33(1):6-15.
  14. Jacobs R, Hendrickx E, Van Mele I; Edwards K; Verheust M; Spaepen A; Van Steenberghe D. Control of a trackball by the chin for communication applications, with and without neck movements. Archs Oral Biol 1997;42(3):213-8.
  15. Hooper J, Taylor R, Pentland B, Whittle I. Rater reliability of Fahn's tremor rating scale in patients with multiple sclerosis. Arch Phys Med Rehabil 1998;79:1076-9.
  16. Goodkin DE, Hertsgaard D, Seminary J. Upper extremity function in multiple sclerosis: improving assessment sensitivity with Box-and-Block and Nine-Hole Peg tests. Arch Phys Med Rehabil 1988;69:850-4.
  17. Goodkin D, Priore R, Wende K. Comparing the ability of various composive outcomes to discriminate treatment effects in MS clinical trials. The multiple sclerosis Collaborative Research Group (MSCRG). Multiple Sclerosis 1998;4(6):480-6.
  18. Alusi SH, Worthington J, Glickman S, Findley LJ, Bain PG. Evaluation of three different ways of assessing tremor in multiple sclerosis. J Neurol Neurosurg Psychiatry 2000;68(6):758-60.
  19. Scherer M. Outcomes of assistive technology use on quality of life. Disabil Rehabil 1996;18(9):439-48.
  20. Smith MW, Sharit J, Czaja SJ. Aging, motor control, and the performance of computer mouse tasks. Hum Factors 1999;41(3):389-96.
  21. Gonzalez JG, Heredia EA, Rahman T, Barner KE, Arce GR. Optimal digital filtering for tremor suppression. IEEE Trans Biomed Eng 2000;47(5):664-73.
  22. Sanes JS, LeWitt PA, Mauritz KH. Visual and mechanical control of postural and kinetic tremor in cerebellar system disorders. J Neurol Neurosurg Psychiatry 1988;51:934-43.
  23. Mitoma H. Intention tremor exaggerated by visually guided movement. Eur Neurol 1996;36:177-8.
  24. Liu X, Miall C. Role of visual feedback on action tremor in multiple sclerosis during visually-guided wrist tracking tasks. Mov Disord 1998;13 Suppl 3:133-4.
  25. Diener HC, Dichgans J. Pathophysiology of cerebellar ataxia. Mov Disord 1992;7:95-109.
  26. Avizzano CA, Barbagli F, Feys P, Bergamasco M. Adaptive filters for the suppression of tremor. Proceedings of the IEEE International Conference: System, Man and Cybernetics; 2000 Oct 7-11, Nashville, TN; 2000. p. 1805-10.
  27. Thoumie P, Charlier JR, Alecki M, D'Erceville D, Heurtin A, Mathe JF, et al. Clinical and functional evaluation of a gaze controlled system for the severely handicapped. Spinal Cord 1998;36:104-9.

Go to TOP.