Simulating hemispatial neglect with virtual reality
- 8.6k Downloads
Hemispatial neglect is a cognitive disorder defined as a lack of attention for stimuli contra-lateral to the brain lesion. The assessment is traditionally done with basic pencil and paper tests and the rehabilitation programs are generally not well adapted. We propose a virtual reality system featuring an eye-tracking device for a better characterization of the neglect that will lead to new rehabilitation techniques.
This paper presents a comparison of eye-gaze patterns of healthy subjects, patients and healthy simulated patients on a virtual line bisection test. The task was also executed with a reduced visual field condition hoping that fewer stimuli would limit the neglect.
We found that patients and healthy simulated patients had similar eye-gaze patterns. However, while the reduced visual field condition had no effect on the healthy simulated patients, it actually had a negative impact on the patients. We discuss the reasons for these differences and how they relate to the limitations of the neglect simulation.
We argue that with some improvements the technique could be used to determine the potential of new rehabilitation techniques and also help the rehabilitation staff or the patient's relatives to better understand the neglect condition.
KeywordsVirtual World Haptic Device Line Bisection Virtual Reality System Neglect Patient
Hemispatial neglect is a disorder usually observed after a stroke with a lesion in the right parietal lobe. It is defined as a loss of attention for stimuli contra-lateral to the lesion. In severe cases, patients with neglect can also suffer from associated disorders like hemi-paresis and somatoparaphrenia (regarding body parts as though they are someone else's) . The assessment of hemispatial neglect is done with pencil and paper tests consisting of line bisection, target cancellation or copies of drawing tasks [2, 3, 4]. Patients with hemispatial neglect make incorrect bisections, fail to cancel the targets on the left side, etc. A system combining virtual reality and haptic feedback was developed to overcome the lack of proper quantification of neglect that occurs when using these traditional tests. The uniqueness of this system is the use of an eye-gaze tracking device.
Ishiai el al.  studied neglect patients for many years and in particular, has shown that eye-gaze pattern is a valuable information to gain an understanding of the underlying mechanisms of neglect. It can also be used to avoid a false positive result with hemianopia patients or as the basis of a quantitative assessment of neglect.
As we entered the testing phase, we had a lot of difficulties to find hemispatial neglect patients. This lead us to create a virtual equivalent of the neglect condition to artificially increase the number of patients. This paper explains how we simulate the neglect and presents a comparison study between the performance and eye-gaze patterns of patients, healthy subjects and healthy simulated patients on the virtual line bisection test.
Two patients and 44 healthy subjects, shown in table 1, participated in the experiments. The healthy subjects were recruited among the student population and through a company specialized in the short-term employment for retired adults. Inclusion criteria included normal visual acuity (with or without correction), right hand dominance, the absence of any neuromuscular pathology. The healthy subjects were divided into a healthy subjects group and a healthy simulated patients group. Both groups were sub-divided into "young" and "senior" categories.
The two patients, Patient K, a male, and Patient O, a female, were recruited at the rehabilitation center of Saitou Hospital in Ishinomaki (prefecture of Miyagi, Japan). At the time of the experiment, Patient K. was 71 years old. On December 2005, he suddenly experienced difficulties to speak and went to a civic hospital where he was told he had a cerebral infarct. At the beginning of his rehabilitation in February 2006, he could not walk correctly because of the longer movements of his right foot. His unilateral spatial neglect condition was assessed on the basis of daily observations made by the rehabilitation staff. For instance, we were told that Patient K. usually bumps into various objects with his left shoulder or his left foot and often forgets to put on his left shoe. The hemispatial neglect was confirmed with line bisection and target cancellation pencil and paper tests prior to the experiment. Patient O., 73 years old, started her rehabilitation in March 2006. She had a cerebral infarct in January. The rehabilitation staff mentioned that she tends to not pay attention to her left hand. The assessment of the neglect with pencil and paper tests was not conclusive: a few missed or double checked targets and relatively fair results on the line bisection test. The eye-gaze pattern test was carried out in order to obtain additional information.
This system was designed for the rehabilitation and assessment of patients with hemispatial neglect. Several of the tasks were specifically built for the assessment and other, more entertaining tasks for the rehabilitation. For the experiment presented in this paper, we used a virtual replica of the paper and pencil tests that was built to test new rehabilitation techniques and to provide a precise characterization of the neglect . The assessment is based on the combination of performance and the eye-gaze pattern of the subject.
As we had difficulties in finding many patients to test the system, we developed a simulation of the hemispatial neglect. The first computer model of hemispatial neglect was developed by Mozer  and is referred to as MORSEL. Our approach is simpler because we needed a real-time and interactive model. We use the eye-gaze tracking device to dynamically modify the virtual world to reflect the effect of the neglect. It was shown that the neglect can be object-centric or body-centric depending on the task [8, 9]. In the case of the line bisection test, the neglect is a combination of both because evidence from other studies shows that the bisection of left sided lines has the poorest results and that the bisection of right sided lines is a bit better than that of centered lines [10, 11]. However, given the small workspace and the fact that the patient's head and trunk are in a fixed position when using our system, we have chosen to implement an object-centric neglect. Only the right half of the line was displayed in the initial condition. The left part of the line was displayed depending on the eye-gaze. If the patient looks beyond the current left end of the line then the remaining segment between the eye-gaze and the current left end was displayed. In any case, the right half of the line was always displayed.
After a paper and pencil test evaluation of the hemispatial neglect, the subjects were told how to operate the system. For the normal condition, the subjects were told to "Mark the middle of the line with a pen stroke. The operator will then ask you if you want to try again. If you answer no, the operator will select the next trial.". For the reduced visual field condition, the instruction was "Press the button on the device when the green cross reaches the middle of the line. This will place a mark and display the whole line. The operator will ask you if you want to try again. If you answer no, we will move on to the next trial". The tasks were executed as follows : learning trial in the normal condition, one set in the normal condition, learning trial in the reduced visual field, one set in the reduced visual field, one set in the normal condition. The healthy simulated patients were not told about their particular condition and were given the same instructions.
The state of the virtual objects, the subject's interactions, eye-gaze and the positions of the marked mid-points were recorded. The analysis was done on the eye-gaze patterns and performance. Given the nature of the task, we only considered the lateral component of the eye-gaze. The subject's performance was evaluated by using the distance between the mark and the middle in terms of percents of half the length of the line. In other words, if the line was crossed at one of its extremities, the error would be 100%.
Pattern of eye-gaze
In contrast, there was no difference in the eye-gaze patterns of the healthy simulated patients for the two conditions. We observed a global shift of the distribution towards the left for all the simulated patients in both conditions. For the patient group, we found that the eye-gaze pattern of Patient O. covered a significant portion of the screen in both conditions. Patient K.'s eye-gaze patterns were constrained to the right side of the screen for the two conditions. The lower graph in Fig. 3 shows Patient K.'s eye-gaze during a reduced visual field task. The eye-gaze is clearly limited to the right half portion of the screen. The discontinuities are due to the unsuccessful tracking that happened when the patient's head was leaning toward the right side. However, we did not notice any peek to the left side while monitoring the video output of the eye-tracking device.
We were expecting that both the healthy simulated patients and patients would track the moving area in the reduced visual condition leading to better bisections and improvement for post-training tasks. Unfortunately, it was not the case. The healthy simulated patients had no reason to suspect that the lines were incomplete. Therefore, their strategy was to keep focusing on the middle of this line and wait for the return of the round area to place the mark. By comparison, the healthy subjects followed the round area to determine where the line would end. From there on, their strategy was similar to healthy simulated patients. Despite the similarities between the eye-gaze patterns of Patient K. and the healthy simulated patients, the implications are still unclear. For instance, the patients might have noticed that the line was not complete but could not follow the moving area because it entered their neglected space. Ishiai et al. reported that even if neglect patients are able to scan leftward when specifically asked to do so, their bisections do not improve significantly . For our next experiments, we will specifically ask some of the patients to follow the round area beforehand to determine if the reduced visual field condition results in better bisections.
The difference with the patients' results shows the limit of our simulated hemispatial neglect which is based on the view introduced by Bisiach et al.  that neglect patients correctly bisect the rightward portion of the line that they see. We aim to adjust the algorithm to follow other models such as Marshall and Halligan's model based on scanning  or the view that neglect involves a distortion of the neglected side  and fine tune the model by using the patients' results.
In this paper we have introduced the concept of simulating hemispatial neglect with virtual reality as a tool for the evaluation of new rehabilitation techniques. The present study indicates that this approach can be useful to determine the potential of a particular technique, such as reduction of the visual field, for intervention with hemineglect patients. Further experiments are needed to improve the simulation and to validate its use. Experiencing hemispatial neglect within a virtual reality system appears to benefit both the family and the rehabilitation staff. In order to meet this goal, the technique should be generalized to a wide variety of virtual scenes and extended to other sensory spaces.
The authors wish to acknowledge the advices and help of Kazunori Seki D.M.Sc. and Yasunobu Handa D.M.Sc. of the graduate school of medicine at Tohoku university, Sendai, Japan, Tanaka Akira Ph.D of the college of symbiotic systems science at Fukushima university, Japan. We also would like to thank Kimura Itaru D.M.Sc. member of the national hospital organization, representing the miyagi national hospital. This work was sponsored by a grant from the regional research project of Miyagi.
- 1.Heilman KM, Watson RT, Valenstein E: Neglect and related disorders. Clinical neuropsychology. Oxford University Press; 1985:243-250.Google Scholar
- 7.Mozer MC: A connectionist model of selective attention in visual perception. Proc 10th Annu Conf of the cognitive science society 1998, 195-201.Google Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.