Detection of freezing of gait episodes in patients with parkinson's disease using electroencephalography and motion sensors: A protocol and its feasibility results

Ugur Eliiyi; Turhan Kahraman; Arzu Genç; Pembe Keskinoglu; Ahmet Özkurt; Berril Çolakoglu Dönmez

doi:10.4103/nsn.nsn_104_22

ORIGINAL ARTICLE

Year : 2022 | Volume : 39 | Issue : 4 | Page : 200-205

Detection of freezing of gait episodes in patients with parkinson's disease using electroencephalography and motion sensors: A protocol and its feasibility results

Ugur Eliiyi¹, Turhan Kahraman², Arzu Genç³, Pembe Keskinoglu⁴, Ahmet Özkurt⁵, Berril Çolakoglu Dönmez⁶
¹ Department of Business, Faculty of Economics and Administrative Sciences, Izmir Bakircay University, Izmir, Turkey
² Department of Physiotherapy and Rehabilitation, Faculty of Health Sciences, Izmir Katip Celebi University, Izmir, Turkey
³ Department of Neurologic Physiotherapy-Rehabilitation, Faculty of Physical Therapy and Rehabilitation, Dokuz Eylul University, Izmir, Turkey
⁴ Department of Biostatistics and Medical Informatics, School of Medicine, Dokuz Eylül University, Izmir, Turkey
⁵ Department of Circuits and Systems, Faculty of Engineering, Dokuz Eylül University, Izmir, Turkey
⁶ Department of Neurology, School of Medicine, Dokuz Eylül University, Izmir, Turkey

Date of Submission	03-Jun-2022
Date of Decision	26-Sep-2022
Date of Acceptance	17-Oct-2022
Date of Web Publication	19-Dec-2022

Correspondence Address:
Turhan Kahraman
Department of Physiotherapy and Rehabilitation, Faculty of Health Sciences, Izmir Katip Celebi University, Izmir
Turkey

Source of Support: None, Conflict of Interest: None

DOI: 10.4103/nsn.nsn_104_22

Abstract

Objective: Freezing of gait (FOG) is an important concern for both patients with Parkinson's disease (pwPD) and physicians. In this study, we aimed to introduce a study protocol and our initial data. The data were subsequently used in machine learning models to detect FOG episodes using brain activity signals and motion data in the laboratory setting using complex FOG-evoking activities in a sample of pwPD with and without FOG compared with age-matched healthy controls. Subjects and Methods: An experimental task to evoke a FOG episode was designed. This experimental task was tested on two pwPD with FOG in “on” and “off” periods and one healthy control. Brain activity signals and motion data were collected simultaneously using electroencephalography (EEG) and inertial measurement units (IMUs). Results: The whole procedure took about 2 h, during which around 30 min were spent on walking tasks, involving 35 complete tours in the designed 8-m hallway by pwPD. Both EEG and IMUs sensor data could be collected, accompanied by FOG episode data marked by the neurologist. The video recordings of the patient's walking tasks were checked and reanalyzed by the neurologist sometime after the data experiment for marking the beginnings and ends of the observed FOG episodes more precisely. In the end, 24 stops were marked as FOG, which corresponded to 11% of the sensor data collected during the walking tasks. Conclusion: The designed FOG-evoking task protocol could be performed without any adverse effects, and it created enough FOG episodes for analysis. EEG and motion sensor data could be successfully collected without any significant artifacts.

Keywords: Electroencephalography, freezing, gait, motion, Parkinson's disease

How to cite this article:
Eliiyi U, Kahraman T, Genç A, Keskinoglu P, Özkurt A, Dönmez B&. Detection of freezing of gait episodes in patients with parkinson's disease using electroencephalography and motion sensors: A protocol and its feasibility results. Neurol Sci Neurophysiol 2022;39:200-5

How to cite this URL:
Eliiyi U, Kahraman T, Genç A, Keskinoglu P, Özkurt A, Dönmez B&. Detection of freezing of gait episodes in patients with parkinson's disease using electroencephalography and motion sensors: A protocol and its feasibility results. Neurol Sci Neurophysiol [serial online] 2022 [cited 2023 May 29];39:200-5. Available from: http://www.nsnjournal.org/text.asp?2022/39/4/200/364418

Introduction

Parkinson's disease is the second-most common neurodegenerative disorder.^[1] Bradykinesia, rigidity, and tremor are the most common clinical symptoms seen at the early stages of the disease.^[1],[2] With disease progression, the prevalence of freezing of gait (FOG), postural instability, and falling increases.^[1],[2] The treatment-resistant characteristics of these symptoms have an important negative influence on patients' quality of life.

The accepted definition of the FOG is “brief, episodic absence or marked reduction of forward progression of the feet despite the intention to walk.”^[3] The pathogenesis of FOG is still unclear, and the treatment options have poor efficacy, which makes it an important concern for both patients and physicians. In the clinical setting, the assessment of FOG mostly relies on clinical observation and self-reported history using simple questions or validated questionnaires.^{[3],[4],[5],[6]} However, due to the unexpected nature of the FOG episodes, its clinical observation is usually difficult. Patient-reported outcome measures are better to reflect the presence and severity of FOG compared with clinical examinations.^{[3],[4],[5],[6]} However, they have some disadvantages such as memory and response bias. In addition, akinesia related to the “off” state is frequently mistaken as the FOG episode, or very brief FOG episodes can be overlooked by patients and their significant others (e.g., formal/informal caregivers, life partners, and children). Several attempts have been used to address these issues in parallel to technological advancements.^[7]

Over the last 15 years, wearable sensors have been used to detect and predict FOG episodes.^[7],[8] Many studies used accelerometers, gyroscopes, force-sensitive resistors, electromyography, electroencephalography (EEG), galvanic skin responses, goniometers, telemeters, or camera-based motion capture for this purpose.^[8] Despite many sensor options, accelerometers, and gyroscopes, and their combinations, named inertial measurement units (IMU) are the most used sensors in the literature because of their relatively low cost and ease of use.^[8] Although many studies have been performed, there are still important challenges and a lack of large FOG datasets, which are important to increase the performance of machine-learning models, especially for deep learning. To overcome this problem, multimodal sensors would be a better solution. Recently, Wang et al.^[9] developed a multimodal model system using brain activity from EEG and motion data from accelerometers in patients with Parkinson's disease (pwPD). Using EEG to detect FOG episodes is not a new method,^{[10],[11],[12],[13],[14]} but it has been recently used in combination with motion sensors.^[9] However, previous studies have important shortcomings such as a lack of pwPD without FOG symptoms and age-matched healthy controls, a lack of assessments in both “off” and “on” stages, and they are mostly limited to some basic FOG-evoking activities such as sit-to-stand, and turning and walking.^{[8],[9],[10],[11],[12],[13],[14]} A study protocol overcoming these shortcomings would provide better datasets to increase the performance of detection and prediction models for FOG events. Here, we introduce a study protocol and our initial data. The data were used in machine learning models to detect FOG episodes using brain activity signals from EEG and motion data from IMUs in the laboratory setting using complex FOG-evoking activities in a sample of people with Parkinson's disease with and without FOG compared with age-matched healthy controls and the performance of the algorithm was investigated based on a physician's judgment of FOG episodes.

Subjects and Methods

Design and ethical considerations

This was an experimental-method-finding study, in which a theoretically designed walking platform was tested as a whole, concerning FOG-evoking in pwPD, the collection of EEG and motion signal data from IMU sensors while walking, and marking of the data by a neurologist corresponding to FOG events. All required experiments were approved by the Noninvasive Clinical Research Ethics Committee of Dokuz Eylül University, İzmir, Turkey because patient practices were also included in this preliminary study (Date: September 07, 2017, Approval number: 2017/21-06). Written informed consent was obtained from all participants of the study.

Experimental task to evoke a freezing of gait episode

pwPD were required to cease Parkinson's disease medication at least 12 h before arriving at the laboratory. They took the last dose at 20:00 the evening before the test and arrived at the assessment room at around 09:00 the following morning. This period was deemed to be an “off” period and was confirmed by a neurologist to be a true “off” period.

Sensors used in the study were placed on the participants. The participants were asked to walk a few times making sure that the sensors were properly placed and that the correct signal could be received. The participants were then seated on a standard 43-cm high armchair where they rested for 3 min.

An experimental task design was used to evoke a FOG episode.^[15] The tasks were performed in an 8-m hallway where the last 1 m had a doorway 0.8-m wide. With the command “Start,” the participants rose from the chair. They walked to the doorway and passed through it. After passing through the doorway, they continued to walk and make a U-turn, then passed through the doorway again. After walking approximately 2 m, the command “pivot 360° around yourself to the right” was given. Then the participants continued to walk. After walking approximately 2 m, the command “pivot 360° around yourself to the left” was given. Then the participants continued walking. After the participants reached the starting point, they were asked about their fatigue. If they did not feel fatigued, the same experimental protocol was re-applied. If the participants indicated that they were tired or the neurologist noticed that they were tired, the participants sat on the chair. After an adequate rest period, they started the experimental protocol again. The same protocol was repeated three times in both the pwPD and the healthy participants. However, the protocol was repeated five times in pwPD who were classified as freezers if any FOG events did not occur in the first three trials. During the tasks, for safety reasons, a physiotherapist escorted the participants. A neurologist observed the participants and determined FOG episodes by pressing a button (the button system is described in the following section). All experimental tasks were recorded using two video cameras placed at suitable points to observe the participants at all times for marking purposes. [Figure 1] presents an example of the applications of the measurement systems and experimental tasks.

Figure 1: Diagrammatic overview of the experiment walking track setup used in the study

Click here to view

Hardware and components

Wireless electroencephalography system

During the experimental protocol performed under the control of an experienced neurologist, the EEG data generated on the participant sensors were collected in real-time using this system. EEG helps to measure the part of the electrical signals that are formed by the chemical activity between nerve cells in an active brain that leak to the surface of the head. The potential voltage difference between the measuring tips (probes) in the EEG device, i.e., between the reference probes and the channel probes, is measured. In this study, certified EMOTIV EPOC + 14-Channel Wireless EEG headsets (EMOTIV, San Francisco, CA, USA) were used to collect patient data for ease of use.

Wearable inertial movement unit

The designed system also allowed for the acquisition of signals due to the movements of the limbs during the walking of the participants. Six IMU sensors were placed on the right and left wrists, right and left ankles, and the right and left sides of the waists of the participants. In this study, an MTw Full Development Kit System (XSens, Los Angeles, CA, USA) was used because it has been certified for use on humans.

Freezing of gait episode marking system

A button system was implemented by the researchers to mark the FOG episodes. During the FOG-evoking experimental tasks, the neurologist used this button to generate a marking pulse for the FOG episode. The neurologist kept the button pressed during the period when the patient froze, thus the FOG episodes in the system were marked. These data, which were recorded with time information, were then synchronized with the EEG and IMU data, allowing all sensor data of pwPD to be analyzed according to the FOG episode markers. The entire system is presented in [Figure 2].

Figure 2: Overview of the components used in the freezing of gait episodes marking system

Click here to view

Results and Discussion

The introduced protocol was used successfully in two pwPD. The patients easily tolerated the measurement system and the FOG-evoked tasks. No adverse events were observed. The entire procedure took about 2 h, during which around 30 min were spent on walking tasks, involving 35 complete tours in the designed 8-m hallway by pwPD. Both EEG and IMU sensor data could be collected, accompanied by FOG episode data marked by the neurologist. The video recordings of the patient's walking tasks were checked and reanalyzed by the neurologist sometime after the data experiment to mark the beginnings and ends of the observed FOG episodes more precisely. In the end, 24 stops were marked as FOGs, which corresponded to 11% of the sensor data collected during the walking tasks. Nearly, all of these events were also witnessed as involuntary stops by the other team members.

Healthy participant experiment

By conducting the same sensor measurements on a healthy participant, it was confirmed that both EEG and IMU signal data were successfully recorded by the designed system in the walking hallway, regarding the effective transmission of both device signals and the commands given by the neurologist in the designed physical environment's range.

Patient participant experiments

Patient 1 with Parkinson's disease

Patient characteristics

A 75-year-old male who frequently presented FOG symptoms and had been under follow-up for 21 years with a diagnosis of Parkinson's disease in the Department of Neurology, Dokuz Eylül University Hospital, İzmir, Turkey, where this research was conducted. Because problems specific to the “off” and “on” periods were being investigated, to create off-term standardization with the test starting at 10:00 in the morning, he was told to take his Parkinson's disease medication at 22:00 the previous night. It was learned that the medicine was taken on time in accordance with the physician's suggestion. The patient had dyskinesia during the “on” period but did not have it during the “off” period.

Test findings

”Off” period walking characteristics

Movement was slow, his feet did not lift from the ground in a normal walking pattern, and the first FOG episode was experienced after 5 min (on the second attempt) of the experiment. The protocol was performed under close (1.5 m) scrutiny, where one of the physiotherapist researchers could intervene to ensure the safety of the patient. It was observed that the resumption of walking was not achieved with the intervention of touching the back of the patient. The first patient experienced nine FOG episodes within 14 min, which lasted 66 s in total, in the “off” period without medication; he recovered from all these occurrences and completed the walking track according to the commands.

”On” period walking characteristics

No further tests were performed after he had taken his medication due to prevent any complications or unnecessary exhaustion because the patient experienced very frequent FOG episodes. The patient completed the walking track (full cycle from standing up from the chair to sitting down again afterward) five times, which took 6 min in total (walking + FOG episodes). His EEG and motion sensor data were recorded and FOG episode markings were successfully transmitted via wireless transfer. There was no falling or any other clinical problems observed in this patient.

Patient 2 with Parkinson's disease

Patient characteristics

A 58-year-old female presented FOG symptoms less frequently compared with patient 1, who had had a diagnosed with Parkinson's disease for 14 years and was under follow-up in the same department. To create the off-term standardization for the test starting at 11:00 in the morning, she was prescribed to take her Parkinson's disease medication at 23:00 the previous night. It was learned that the medicine was taken on time in accordance with the physician's suggestion. The patient had dyskinesia during the “on” period but did not have it during the “off” period.

Test findings

The “off” period walking characteristics had the feature that the feet did not lift from the ground in a normal walking pattern and at normal walking speed. No FOG episodes were observed in the “off” period walking cycles. Nonetheless, the protocol was conducted under close (1.5 m) scrutiny, where one of the physiotherapist researchers could intervene to ensure the safety of the patient. She completed all her attempts on the walking track according to the commands.

”On” period walking characteristics: 50 min after taking her medication, she experienced her first FOG episode (within 20 s of her first attempt). Patient 2 experienced 15 FOG episodes within 25 min, with a total duration of 142 s in the “on” period after taking her medication. She recovered from all these occurrences and completed the walking track according to the commands. In both periods (off/on), the patient completed the walking track 30 times, which took 23 min in total; her EEG and motion sensor data were successfully collected and all FOG episode markings were transmitted in full synchronization. There was no falling nor any other clinical problems observed in this patient.

Some descriptive statistics for the collected signal data and FOG detection status in repeated signal measurements of the two patients and one healthy control are presented in [Table 1]. As can be seen, nearly 29 min of data collected from the two pwPD amounted to 98,264 rows of signal data, 14.4% of which are marked as FOG episodes. Using the same data as summarized in [Table 1], average signal frequencies for a selection of EEG channels (corresponding to F3, FC5, AF3, AF4, FC6, and F4 channels of the EMOTIV EPOC + 14-Channel Wireless EEG headset) are given in [Figure 3] for both patients and the healthy control, as a subset of all the available signal channels. These will be used in forthcoming studies using the protocol described herein.

Table 1: Data size of the signal measurements in the protocol experiment

Click here to view

Figure 3: Average signal frequencies (in Hertz) of different EEG channels (F3, FC5, AF3, AF4, FC6, and F4) measurements

Click here to view

For the sake of preliminary observation, the differences in the means between the unmarked and marked data for both patients were tested with a 0.05 significance level. For the first subject, in four of the six channels, as presented in [Figure 3], signal frequency mean values of the FOG-detected parts of the data were significantly different than unmarked ones (data corresponding to normal walking patterns). For the second patient subject, this difference was observed in all six channels as shown in [Figure 3], even with a significance level of 0.01.

The designed FOG-evoking task protocol was performed without any adverse effects, and it created sufficient FOG episodes for analysis. The EEG and motion sensor data were successfully collected without any significant artifacts. The preliminary results indicated that the targeted protocol was feasible and safe and could be used in a targeted full-scale study.

Financial support and sponsorship

This work was funded by the Department of Scientific Research Projects, Dokuz Eylül University (İzmir, Türkiye), under grant number 2018.KB.SAG.005.

Conflicts of interest

There are no conflicts of interest.

References

1.	Poewe W, Seppi K, Tanner CM, Halliday GM, Brundin P, Volkmann J, et al. Parkinson disease. Nat Rev Dis Primers 2017;3:17013.
2.	Hely MA, Reid WG, Adena MA, Halliday GM, Morris JG. The Sydney multicenter study of Parkinson's disease: The inevitability of dementia at 20 years. Mov Disord 2008;23:837-44.
3.	Nutt JG, Bloem BR, Giladi N, Hallett M, Horak FB, Nieuwboer A. Freezing of gait: Moving forward on a mysterious clinical phenomenon. Lancet Neurol 2011;10:734-44.
4.	Goetz CG, Tilley BC, Shaftman SR, Stebbins GT, Fahn S, Martinez-Martin P, et al. Movement Disorder Society-sponsored revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS): Scale presentation and clinimetric testing results. Mov Disord 2008;23:2129-70.
5.	Giladi N, Tal J, Azulay T, Rascol O, Brooks DJ, Melamed E, et al. Validation of the freezing of gait questionnaire in patients with Parkinson's disease. Mov Disord 2009;24:655-61.
6.	Nieuwboer A, Rochester L, Herman T, Vandenberghe W, Emil GE, Thomaes T, et al. Reliability of the new freezing of gait questionnaire: Agreement between patients with Parkinson's disease and their carers. Gait Posture 2009;30:459-63.
7.	Irrera F, Cabestany J, Suppa A. Editorial: New advanced wireless technologies for objective monitoring of motor symptoms in Parkinson's disease. Front Neurol 2018;9:216.
8.	Pardoel S, Kofman J, Nantel J, Lemaire ED. Wearable-sensor-based detection and prediction of freezing of gait in Parkinson's disease: A review. Sensors (Basel) 2019;19:5141.
9.	Wang Y, Beuving F, Nonnekes J, Cohen MX, Long X, Aarts RM, et al. Freezing of gait detection in Parkinson's disease via multimodal analysis of EEG and accelerometer signals. Annu Int Conf IEEE Eng Med Biol Soc 2020;2020:847-50.
10.	Handojoseno AM, Gilat M, Ly QT, Chamtie H, Shine JM, Nguyen TN, et al. An EEG study of turning freeze in Parkinson's disease patients: The alteration of brain dynamic on the motor and visual cortex. Annu Int Conf IEEE Eng Med Biol Soc 2015;2015:6618-21.
11.	Ly QT, Ardi Handojoseno AM, Gilat M, Nguyen N, Chai R, Tran Y, et al. Identifying montages that best detect the electroencephalogram power spectrum alteration during freezing of gait in Parkinson's disease patients. Annu Int Conf IEEE Eng Med Biol Soc 2016;2016:6094-7.
12.	Ly QT, Ardi Handojoseno AM, Gilat M, Chai R, Ehgoetz Martens KA, Georgiades M, et al. Detection of turning freeze in Parkinson's disease based on S-transform decomposition of EEG signals. Annu Int Conf IEEE Eng Med Biol Soc 2017;2017:3044-7.
13.	Handojoseno AM, Shine JM, Nguyen TN, Tran Y, Lewis SJ, Nguyen HT. Analysis and prediction of the freezing of gait using EEG brain dynamics. IEEE Trans Neural Syst Rehabil Eng 2015;23:887-96.
14.	Palmerini L, Rocchi L, Mazilu S, Gazit E, Hausdorff JM, Chiari L. Identification of characteristic motor patterns preceding freezing of gait in Parkinson's disease using wearable sensors. Front Neurol 2017;8:394.
15.	Tripoliti EE, Tzallas AT, Tsipouras MG, Rigas G, Bougia P, Leontiou M, et al. Automatic detection of freezing of gait events in patients with Parkinson's disease. Comput Methods Programs Biomed 2013;110:12-26.