Hemineglect assessment and rehabilitation using a
robotic serious game
V. Montedoro, M. Alsamour & M. Edwards
S. Dehem, T. Lejeune & G.Stoquart
Université catholique de Louvain
Institute for Research in Psychological Science
Belgium
[email protected]
Université catholique de Louvain
Institut de Recherche Clinique et Expérimentale
Belgium
S. Heins, D. Galinski & B. Dehez
Université catholique de Louvain
Institute of Mechanics, Materials and Civil Engineering
Belgium
Abstract— Hemineglect is a condition where brain-damaged
patients are impaired at perceiving and responding to the
contralesional part of objects (allocentric hemineglect) and/or the
contralesional part of space (egocentric hemineglect). The
condition can also be accompanied by motor impairments.
Although treatments exist, none are completely effective. The
pipeline objective of our research is to develop a serious game that
evaluates and rehabilitates the different forms of hemineglect
according to the patient’s impairment severity. The first step
towards this objective is the development and validation of a
robotic assessment for egocentric hemineglect. We tested 12 stroke
patients and 40 healthy control participants with the robotic
hemineglect assessment. Patients were also assessed with two
hemineglect standardized tests. Significant correlations between
the robotic and standardized tests demonstrated a good validity for
the assessment of egocentric hemineglect. Moreover, the robotic
assessment showed a good sensitivity and specificity. Based on
these results, we can identify the captures / components necessary
for the serious game rehabilitation to adapt difficulty level in
correspondence to the patient’s egocentric hemineglect
impairment. Future steps in development will focus on the
validation of the allocentric and motor hemineglect assessment.
Keywords— hemineglect; spatial attention; assessment; serious
game; robot; brain-damage
I.
BACKGROUND
Recent research shows that approximately one third of acute
stroke patients present hemineglect (also named visuospatial
neglect) [1]. The condition is heterogeneous, and leads to many
different forms. Egocentric hemineglect refers to a failure to
respond to contralesional stimuli from the patient’s viewpoint
[2], whereas allocentric hemineglect is the failure to attend to
the contralesional side of objects, irrespective of the object
location from the patient’s viewpoint [2]. As well as
visuospatial deficits, hemineglect can also lead to motor
problems, either corresponding to an underutilisation of the
contralesional limb in absence of muscle, strength, or sensation
impairments, or corresponding to defective movements towards
the contralesional hemifield [3], [4].
Hemineglect significantly hinders many daily activities [5] and
current clinical techniques are not completely effective (e.g.
limited transfer to everyday life) [6]. The use of serious games
in neuropsychological assessment and rehabilitation has shown
promising results during the last years. For example, Kim and
colleagues [7] demonstrated that hemineglect patients using
interactive virtual reality serious game tasks of catching
coconuts or balls across different parts of space, led to
improvements in recovery compared to other patients that
followed conventional treatments. Also, research has shown that
use of an ecological virtual reality serious game environment
(e.g. street crossing) led to a greater transfer to daily life
activities [8]. However, to our knowledge no research has so far
used robot interaction, and the focus of existing serious game
rehabilitation has always been to patients with egocentric rather
than other forms of visuospatial neglect.
II.
OBJECTIVE
The objective of our overall project is to develop a new
serious game to evaluate and rehabilitate egocentric, allocentric
and motor forms of hemineglect (Fig. 1). Moreover, we chose to
integrate the serious game on a robot in order to provide a
means to evaluate and rehabilitate motor components of the
hemineglect impairment.
In the first phase of the project reported here, we developed
the egocentric component of the robotic hemineglect
assessment. Our principal objective was to determine the
captures necessary to dynamically adapt the game difficulty in
correspondence to the patient’s egocentric hemineglect
impairment during the rehabilitation. The second aim of the
current phase of the project was to collect normative data on
healthy control participants in order to determine the sensitivity
and specificity of the robotic hemineglect assessment.
Fig. 1. Schematic representation of the overall project.
III.
METHOD
Twelve stroke patients (63,8 ± 10,2 years and 13 ± 12,7
months post-onset) and forty mean age paired healthy controls
(61,4 ± 12,7 years) were tested. The study was approved by the
St-Luc hospital ethics committee and all participants gave their
consent.
The REAplan rehabilitation robot (Axinesis) was used for
the development of the hemineglect assessment. The endeffector robot uses position and force sensors to allow for
precise action measures in the horizontal plane. The robot has a
visual interface for the patient, and a separate monitor for the
therapist. For the present phase, the robotic hemineglect
assessment consisted of a target pointing task that required the
participant to find a musical conductor (target) randomly
presented among musical instruments (distractors). The targets
were complete or incomplete (where the object was missing
left- or right-side allocentric features). For each display, a single
target was randomly presented somewhere in the lateral space
(egocentric, left-to-right). As soon as the complete target was
perceived, the participant was instructed to make a push
response of the end-effector towards a conductor’s baton image.
However, if the target was incomplete, the participant made a
pull response towards an image of a musical stand.
The healthy participants were only evaluated with the robotic
hemineglect assessment. Patients were also assessed with two
standardized cancellation tests; the Bells test [9] and the Apple
test [10]. The order of the tests was counterbalanced between
participants. The dependent variables were: percentage of
contralesional space omissions, difference between percentage
of omissions in contralesional compared to ipsilesional space
(i.e., percentage of omission asymmetry), mean reaction time
for correct answers in contralesional space, and difference
between mean reaction time for correct answers in
contralesional compared to ipsilesional space (i.e., mean
reaction time asymmetry). Validity was analyzed with
Spearman correlations. Norms for the robotic assessment were
created using ninety-five percent confidence intervals, and
sensitivity and specificity were calculated by comparing the
number of patients impaired at the robotic assessment with the
number of patients impaired at the two standardized tests.
IV.
RESULTS
Results showed correlation between percentage of omissions
and mean reaction time in contralesional space (ρ = 0.811,
p=0.002), and percentage of omission asymmetry and mean
reaction time asymmetry (ρ = 0.674, p=0.023). Standardized
test comparisons demonstrated correlation between percentage
of contralesional space omissions on the robotic assessment and
percentage of contralesional space omissions on the Bells (ρ =
0.696, p=0.012) and Apple tests (ρ = 0.851, p=0.000).
Similarly, omission asymmetry percentage on the robotic
assessment correlated with percentage of omissions asymmetry
on the Apple test (ρ = 0.620, p=0.031), and showed some
relation with the percentage of omission asymmetry on the Bells
test (ρ = 0.511, p=0.089). Finally, comparison of our
participants to norms showed that the robotic hemineglect
assessment had a sensitivity of 83% and a specificity of 67%.
V. CONCLUSION & PERSPECTIVES
The present phase of our study demonstrated that the robotic
assessment formed a valid tool to evaluate egocentric
hemineglect, matching results from standardized tests.
Moreover, the robotic test had a good sensitivity and specificity
demonstrating value for diagnosis evaluation. These findings
provide the captures necessary to reliably adapt the subsequent
rehabilitation serious game in correspondence to the patient’s
egocentric hemineglect severity. The game will predict when to
trigger visual and motor cues that help the patient achieve a
consistent success rate for re-learning. Future steps in
development will focus on the validation of the allocentric and
motor hemineglect assessment in order to calibrate the serious
game in correspondence with the different hemineglect forms
and severity.
ACKNOWLEDGMENT
This work was supported by the Région Wallonne and
Fondation Saint-Luc. The authors thank Axinesis for the
development of the REAplan, Martin Vanderwegen for his
collaboration and finally all the participants.
REFERENCES
[1]
J. Kamtchum Tatuene, G. Allali, A. Saj., T. Bernati, R. Sztajzel, P.
Pollak and I. Momjian-Mayor, “Incidence, Risk Factors and Anatomy
of Peripersonal Visuospatial Neglect in Acute Stroke,” European
neurology, vol. 75(3-4), pp. 157-163, 2016.
[2] C. Rorden, et al., “Allocentric neglect strongly associated with
egocentric neglect,” Neuropsychologia, vol. 50, pp. 1151-57, 2012.
[3] D.S. Sampanis and J. Riddoch, “Motor neglect and future directions
for research,” Frontiers in Human Neuroscience, vol. 7, pp. 187, 2013.
[4] G. Bottini, R. Sterzi and G. Vallar, “Directional hypokinesia in spatial
hemineglect: a case study,” Journal of Neurology, Neurosurgery &
Psychiatry, vol. 55(7), pp. 562-565, 1992.
[5] P. Pradat-Diehl, F. Poncet, H. Migeot, and C. Taillefer,
“Conséquences dans la vie quotidienne des troubles de la
représentation corporelle,” Revue de neuropsychologie , vol. 2(3), pp.
231-234, 2010.
[6] G. Kerkhoff and T. Schenk, “Rehabilitation of neglect: an update,”
Neuropsychologia, vol. 50(6), pp. 1072-1079, 2012.
[7] Y.M. Kim, M.H. Chun, G.J. Yun, Y.J. Song and H.E. Young, “The
effect of virtual reality training on unilateral spatial neglect in stroke
patients,” Annals of rehabilitation medicine, vol. 35(3), pp. 309-315,
2011.
[8] N. Katz, H. Ring, Y. Naveh, R. Kizony, U. Feintuch and P.L. Weiss,
“Interactive virtual environment training for safe street crossing of
right hemisphere stroke patients with unilateral spatial neglect,”
Disability and rehabilitation, vol. 27(20), pp. 1235-1244, 2005.
[9] L. Gauthier, F. Dehaut and Y. Joanette, “The bells test: a quantitative
and qualitative test for visual neglect,” International Journal of
Clinical Neuropsychology, 1989.
[10] W.L. Bickerton, D. Samson, J. Williamson and G.W. Humphreys,
“Separating forms of neglect using the Apples Test: validation and
functional prediction in chronic and acute stroke,” Neuropsychology,
vol. 25(5), pp. 567, 2011.