Balance assessment in children with cerebral palsy; methods for measuring postural stability

(1)

Balance assessment in children with cerebral palsy; methods for measuring postural stability

Balansbedömning hos barn med cerebral pares;

metoder för att mäta postural stabilitet MICHAELA SJÖDIN

KTH ROYAL INSTITUTE OF TECHNOLOGY

(2)

(3)

me nd this project and keeping me on the track. I would also like to thank Cecilia Lidbeck at Karolinska Institutet for the opportunity to be a part of her work and for showing interest in this thesis. You have both contributed with a lot of inspiring discussions and valuable feedback, making this project both fun and challenging. Also, thanks to Åsa Bartonek at Karolinska Institutet who contributed with good ideas and interesting perspectives. I would like to thank my seminar group at KTH for providing me with thoughtful feedback and useful criticism; Dennis, Yousef, Adnan, Saa, Zihao and of

(4)

(5)

of the center of pressure (CoP). But some scientists claim that the center of mass (CoM) is what really indicates the sway of the whole body, since the body is a multi- joint system. Many previous studies of human balance have targeted groups with dirent kinds of balance impairments. In a recent study C. Lidbeck investigated factors inuencing standing in children with bilateral spastic cerebral palsy (BSCP).

The conclusion of that study was that the crouched position, that is common with this kind of disability, was not found to be related to strength and not entirely related to the degree of their motor disorders.

In this thesis a number methods were chosen to assess the postural stability of children with BSCP, using both the CoP and the CoM. The hypothesis was that the dierent methods would show dierent aspects of the children's balance impairment.

Also, the inuence of visual stimuli on the crouching position was examined. The long term aim is that the results may contribute to a deeper understanding of the balance disturbances that often accompany this group of children.

16 children with BSCP (GMFCS level I-III) and 20 typically developing (TD) children were included in the study. Data was collected, before the start of this project, using two force plates and an eight-camera 3D motion analysis system with passive markers. The children performed three dierent standing tasks during 30 seconds each; quiet standing, blindfolded and an attention-task. Five methods were chosen (based on previous literature) and implemented in Matlab to examine the postural stability of the two groups during the three tasks.

Result shows that all methods used can clearly distinguish between the balance in the BSCP group and the TD group. When comparing the quiet standing task with the blindfolded task in the BSCP group, there were some signicant results from the statistical evaluation (P<0.05). The result from several of the methods indicated that the children of this group have better postural stability when blindfolded, which is not in agreement with previous literature. In contrast, one method using the total mean velocity indicated that the postural stability decreased. During the attention- task, the methods disagreed with each other, implying a change in balance strategy in the BSCP group that was dierent from the TD group.

Four methods are suggested for future studies, two using the CoP and two using the CoM. These four methods highlighted dierent aspects of the data and in combination they may provide a bigger picture of the postural stability of children with BSCP. Even though there were no signicant dierence in the vertical displacement of the CoM between the BSCP and the TD group, the CoM was slightly elevated during the attention-task in the BSCP group. In the TD group the CoM was lowered during the same task. This indicates that the children with BSCP in this study straighten up a bit when they can focus on something outside of their own body.

(6)

(7)

av the center of pressure (CoP). Men vissa forskare påstår att det är the center of mass (CoM) som visar svängningarna av hela kroppen, eftersom kroppen är ett system som består av många leder. I en nyligen utgiven avhandling av C. Lidbeck undersöks faktorer som påverkar balansen hos barn med bilateral spastisk cerebral pares. Slutsatsen var att den hukande positionen, som är vanlig vid den här typen av funktionsnedsättning, inte var kopplad till styrka och inte heller helt och hållet kopplad till deras motoriska störningar.

I det här examensarbetet kommer ett antal metoder att användas för att bedöma postural stabilitet hos barn med BSCP, med användning av både CoP och CoM. Hy- potesen är att olika metoder kommer att visa olika aspekter av barnens balansned- sättning. Även inuensen av visuell stimulans på den hukande positionen kommer att undersökas. Detta kan i längden leda till en djupare förståelse för balanssvårig- heter som ofta följer med den här typen av funktionsnedsättning.

16 barn med BSCP (GMFCS nivå I-III) och 20 typiskt utvecklade (TD) barn var inkluderade i studien. Data samlades in, innan starten av detta projekt, med hjälp av två kraftplattor och ett 3D rörelseanalys-system med åtta kameror och passiva markörer. Barnen ck utföra tre olika uppgifter stående under 30 sekunder; vanligt stående, stå med ögonbindel och hålla uppmärksamheten på en lm. Fem metoder valdes ut för att undersöka resultatet från de olika uppgifterna (baserat tidigare lit- teratur) och implementerades i Matlab.

Resultatet visar att alla metoder som använts tydligt visar en skillnad mellan BSCP- gruppen och TD-gruppen. När vanligt stående jämfördes med stående med ögonbin- del i BSCP gruppen fanns det signikanta skillnader i resultatet från den statistiska utvärderingen (P<0.05). Resultatet från era av metoderna visade att barnen hade bättre stabilitet med ögonbindel än utan, vilket inte stämde överens med resultatet från tidigare studier. I kontrast visade en metod som använder sig av total me- delhastighet en sämre postural stabilitet. Under uppmärksamhets-uppgiften stämde inte resultatet från metoderna överens vilket indikerar en förändrad balansstrategi i BSCP-gruppen, vilket inte förekom i TD-gruppen.

Fyra metoder föreslås för framtida studier, två som använder CoP och två som an- vänder CoM. De här fyra metoderna understryker olika aspekter av datan och kan i kombination visa en större bild av postural stabilitet hos barn med BSCP. Trots att det inte fanns någon signikant skillnad mellan BSCP-gruppen och TD-gruppen vad det gäller den vertikala föryttningen av CoM, så var CoM något förhöjd under uppmärksamhets-uppgiften i BSCP-gruppen. I TD-gruppen var CoM något lägre

(8)

(9)

CNS Central nervous system CP Cerebral palsy

BSCP Bilateral spastic cerebral palsy TD Typically developing

GMFCS Gross Motor Function Classication System GRF Ground reaction force

CoP Center of pressure CoM Center of mass TMV Total mean velocity CoG Center of gravity

CoMacc Center of mass acceleration CoGacc Center of gravity acceleration SD Standard deviation

AP Anterior/Posterior

(10)

(11)

1 Introduction

Studies of postural stability have been an area of interest for a long time due to its com- plexity and the diculty to distinguish between the dierent systems that contribute to human balance. Even when we are standing perfectly still, the body is constantly adjust- ing itself to remain upright, which is referred to as postural sway. This is not always easy since the body has to balance two thirds of its weight about two thirds from the ground at all times [1].

Many studies have focused on balance impairments, in order to reveal causes or nd out how to improve postural control. Since neuromusculoskeletal diseases result in degeneration in the balance control system, studies on postural stability in children and adults with cerebral palsy (CP) have increased during the last decade [2]. In a recent study C.

Lidbeck investigated factors inuencing standing in children with bilateral spastic cerebral palsy (BSCP). The conclusion was that the crouched position, that is common with this kind of disability, was not found to be related to strength and not entirely related to the degree of their motor disorders. This indicates that sensory and/or perceptual disturbances might inuence motor function, even though the movement and posture disability is mainly presumed to be a consequence of the motor disorder [3].

The most common way to examine postural sway is to use the center of pressure (CoP), since it is easier to obtain than the center of mass (CoM) [4]. But some scientists claim that the CoM is the variable that actually indicates the sway of the whole body since the body is a multi-joint system [1, 5]. It is also shown that the horizontal acceleration of the CoM (CoMacc) is very sensitive to age- or disease-related changes in the postural control system [6, 7].

In this thesis a number methods will be used to assess the postural stability of children with BSCP, using both the CoP and the CoM. The hypothesis is that the dierent methods will show dierent aspects of the children's balance impairment. Also, the inuence of visual stimuli on the crouching position will be examined. In the end, suggestions will be made on how to use the dierent methods in future studies of this kind of impairment.

The long term aim is that the result may contribute to a deeper understanding of the balance disturbances that often accompany this group of children.

(14)

(15)

2 Aim and objectives

The aim of this thesis is to study postural stability in children with BSCP using a number of methods that both include the CoP and the CoM. The hypothesis is that the dierent methods will show dierent aspects of the children's balance impairment. The eects of the visual stimuli on the crouched standing position will also be examined using the vertical CoM displacement. As a result, some suggestions will be made on what methods can be used in future studies of this particular group of children.

The objectives of this thesis are to:

• Find a number of methods to measure postural stability of children with BSCP

• Examine the inuence of visual stimuli on the crouched position

• Suggest which methods can be used in future studies of this group of children

(16)

(17)

3 Method

This study is a follow-up on a dissertation that was performed in 2016 at Karolinska Institutet in Solna. Using some of the data that was collected during the original study, the methods used for the balance assessment were developed. In this section, relevant parts of the original study will be presented, and then the methods and material used in the present study will be described.

3.1 The original study at Karolinska Institutet

At Karolinska institutet in Solna, a PhD-thesis with the aim to investigate factors in-

uencing standing in children with BSCP was conducted by C. Lidbeck [3]. Children of varying standing abilities participated and dierent aspects contributing to postural stability were examined.

3.1.1 Participants

In the original study in total 55 children with BSCP and 46 typically developing (TD) children participated. The study was divided into four dierent parts, where only the participants from part three performed the test required for the present study (including a 3D motion analysis). In part three of the original study 36 children with BSCP and 27 TD children participated. In the TD group, the children only had to be of appropriate age and be overall healthy. A number of inclusion and exclusion criteria were applied when recruiting the children with BSCP for the study.

Inclusion criteria of the children with BSCP:

• A diagnosis of BSCP

• Ability to stand independently with or without the use of hand-held support for at least 30 seconds

• Ability to follow verbal instructions for performing the examinations Exclusion criteria of the children with BSCP:

• Dyskinetic CP and/or presence of dystonia

• Skeletal surgery within the previous year

• Soft tissue surgery within the previous six months

• Botulinum toxin injections within the previous six months 3.1.2 Data collection

All the data used in the original thesis was collected by two experienced physiotherapists (C. Lidbeck and Å. Bartonek) at a Motion Analysis Laboratory at Karolinska Univer- sity Hospital, Stockholm, Sweden [3]. An eight-camera 3D motion analysis system with passive markers (Vicon MX40, Oxford, UK) was used to collect segmental data. Ground reaction forces were simultaneously collected from two force plates (Kistler, Winterthur,

(18)

3.2 The present study 3 METHOD

Switzerland). A full-body biomechanical model and marker set (Plug-In-Gait, Vicon) was used and 21 passive markers were placed on anatomical landmarks. The head, thorax and pelvic segments motions are described in the global coordinate frame and the hip, knee and ankle joint as relative angles between distal and proximal segments. The hands and arms had no markers. The data from the force plates is sampled at 1000 Hz and the data from the 3D motion analysis system is sampled at 100 Hz.

3.1.3 Procedure of the trials

The children included in part three of the original study was recruited between January 2012 and September 2013. All participants went through a physical examination by C.

Lidbeck as well as a neuro-ophthalmological examination at a separate occasion. The 3D motion analysis was performed during 30 seconds while the children performed three dierent tasks shown in gure 1. The tasks included:

(a) Quiet standing - a self-selected standing posture (b) Blindfolded - standing while blindfolded

(c) Attention-task - standing while watching a video recording of a child playing with a dog. The video was shown on a 52x30 cm computer screen that was placed 2 m in front of the child

Figure 1: The three tasks performed during 30 seconds each by one of the children from the original study. Quiet standing, blindfolded and during an attention-task. Illustration is used with permission from the author [3].

3.2 The present study

Most of the criteria for inclusion and exclusion are the same as the original study, where the trials were performed. A few criteria have been added or changed to suit this study and are summarized below. Also, the calculations and materials used are presented in this section.

3.2.1 Participants

The criteria that excluded the largest number of participants from the original study was that more than half of the children with BSCP had to use hand held support to remain standing for 30 seconds. To be able to use the data from the force plates, the entire

6

(19)

weight of the subject needs to be on the platform. The hand-held support was located in front of the force plates during the trials. One child with BSCP began to move the head and arms after only 12 seconds and was therefore not included in this study. A reason that aected the number of TD children participating was technical error in Nexus 5, the program used to collect force plate data. The result was that six children from this group were excluded. Another TD child was excluded for moving around during the blindfold task. The added inclusion and exclusion criteria are listed below.

Inclusion criteria:

• Ability to stand without the use of hand-held support for at least 30 seconds

• The data from the force plates need to be available in the software Exclusion criteria:

• The trial lasted for less than 15 seconds (because of movement of the head or the arms)

The criteria resulted in a smaller number of participants than in the original study. In total 16 children with BSCP and 20 TD children participated. Information about the participants is summarized in Table 1 and 2. A more detailed table with participant data can be found in Appendix B.

Table 1: Participant data of children with BSCP.

Gender 6 female (37.5%) and 10 male (62.5%)

Age 7 - 15 years

Weight 30 - 99 kg Height 1.24 - 1.66 m

GMFCS level 5 level I (31.25%), 9 level II (56.25%) and 2 level III (12.5%) Glasses 5 with glasses (31.25%) and 11 without glasses (68.75%)

Table 2: Participant data of TD children.

Gender 10 female (50%) and 10 male (50%) Age 6 - 16 years

Weight 22 - 72 kg Height 1.21 - 1.81 m

3.3 Data application

From the Vicon Motion Capture Software the CoP coordinates in x- and y-directions were extracted as well as the CoM coordinates in x-, y- and z-directions. In methods containing both the CoP and the CoM signal, the CoP had to be modied to match the 100 Hz CoM signal. This was done by using every 10th sample of the original signal. A 4th order, low pass Butterworth-lter with a cuto frequency of 5 Hz was used on the CoP signal to reduce noise [8]. Four dierent methods to measure body sway were selected based

(20)

3.3 Data application 3 METHOD

on the literature review performed at the start of this thesis. Two methods using the CoP and two methods using the CoG (the x- and y-coordinates of the CoM) were used and will be described in this section. The vertical CoM displacement did not need to be calculated since the z-coordinate of the CoM was included in the data and is equal to the distance needed for comparison. All methods were implemented using the software Matlab (MathWorks, Natick, Massachusetts, United States). The full Matlab-code of each method is included in Appendix E.

3.3.1 Global CoP

The CoP coordinates were separate from each force plate and had to be summarized using equations (1) and (2) to obtain the global CoP [9].

x = x₁F z₁

F z + x₂F z₂

F z (1)

y = y₁F z₁

F z + y₂F z₂

F z (2)

x, y Global x and y coordinates of the CoP

x_1,2, y1,2 x and y location of the CoP measured by Plate 1 and 2 F z_1,2 Vertical force measured by plate 1 and 2

F z Total vertical force measured by both plates 3.3.2 Total mean velocity

The total mean velocity (TMV) is obtained by measuring the distances between each sample in the collected CoP-signal and then summarizing the distances to obtain the total sway path (TSP). This distance is then divided by the time of the trial. Calculations are shown in equation (3) and (4).

T SP =

n

X

i=1

p((x_i+1) − x_i)²+ ((y_i+1) − y_i)² (3)

T M V = T SP

t_tot (4)

x, y Global x and y coordinates of the CoP t_tot Total time of trial

3.3.3 95% prediction ellipse

A prediction ellipse use the CoP-samples in the x- and the y-directions to t an ellipse, with as small area as possible, around the plot of the sway path. This was performed by using a simple Matlab-code by Schubert and Kirschner [10]. Parts of the code is presented in this section and the full code is shown in Appendix E.4. The value of the chi-square cumulative distribution function with 2-degrees of freedom at 95% probability level is found shown in eq. (5). The sample variance in both directions and the covariance between x and y are found. Then the eigenvalues are calculated from the variance covariance matrix shown in eq. (6). To obtain the area, the square root of the product of both eigen values

8

(21)

are multiplied with π and the value from eq. (5), the Matlab-code used for this is shown in eq. (7).

chisquare = chi2inv(0.95, 2); (5)

[vec, val] = eig(cov(x, y)); (6)

P EA = pi ∗ chisquare ∗ prod(sqrt(svd(val))) (7) 3.3.4 Acceleration of the CoG

To nd the acceleration of the CoG (CoGacc) in the x- and y- directions, the distance between each sample in each direction is calculated. Each distance is then divided by the total time of the trial two times to be able to see the change in acceleration during the entire time of the trial. The standard deviation (SD) of this array is then calculated.

Calculations are shown in eq. (8), (9) and (10).

∆x_n= (x_n+1) − x_n → Acceleration of ∆x_n= ∆xn

t²_tot (8)

∆y_n= (y_n+1) − y_n → Acceleration of ∆y_n= ∆y_n

t²_tot (9)

SD = v u u t 1 n

n

X

i=1

((x_i, y_i) − µ)² (10)

∆x_n, ∆yn Distance between two samples in x- and y-directions of the CoG signal x, y Global x and y coordinates of the CoG

ttot Total time of trial

µ Mean value of samples in x- and y-directions 3.3.5 CoP-CoG displacement

This method uses the dierence between the x- and y-coordinates of the CoP and the CoG. The SD of this array is then calculated. Calculations are shown in eq. (11), (12) and (10).

∆x_n = CoP x_n− CoGx_n (11)

∆y_n= CoP y_n− CoGy_n (12)

∆x_n, ∆yn Distance between two samples in the CoP and the CoG signal in the x- and y-directions

CoP xn, CoP yn x-and y-coordinates in the CoP signal CoGx_n, CoGy_n x-and y-coordinates in the CoG signal

(22)

3.4 Statistical evaluation 3 METHOD

3.4 Statistical evaluation

The variables obtained from the dierent methods were not found to be normally dis- tributed. This was conrmed by using a KolmogorovSmirnov test in Matlab. Therefore, two dierent nonparametric statistical methods were used, the MannWhitney U test and the Wilcoxon signed-rank test. In the Wilcoxon signed-rank test the two samples that are being compared need to be of equal size, which is not necessary in the Mann-Whitney U test. Since one group was larger than the other (nBSCP=16, nT D=20), both methods were used.

The Wilcoxon signed-rank test was used to nd out if there were any signicant dierences between the tasks performed in the trial. The trial of quiet standing was tested against the blind standing task and the attention task respectively. This was done both for the BSCP group and the TD group to nd out how the tasks diered from the quiet standing trial.

The MannWhitney U test was used in two ways to nd out if there were any signicant dierences between the BSCP group and the TD group. First, each method and task was tested against the same method and task in the other group. The MannWhitney U test was also used when comparing if the dierence from the quiet standing task and the other standing tasks were signicant between the BSCP group and the TD group. The dierence was calculated by subtracting the variables of the quiet standing task from the blindfolded task, and in the same way from the attention task. These results were then compared between the BSCP and the TD group to nd out if there were any signicant dierence between the two groups.

To be able to compare the vertical displacement of CoM between the groups, the variables had to be normalized. This was done by dividing the height of the CoM by the total height of the person to get the percentage of body height. This normalized data was used in the statistical evaluation. In all statistical methods, the level of statistical signicance was set at P < 0.05.

10

(23)

4 Results

In this section the results are presented as the mean and SD from the methods used on both the BSCP group and the TD group. For the CoGacc and CoP-CoG methods, the results will be divided into anterior/posterior (AP) and medial/lateral (ML) directions.

Bar plots will help to visualize the results from the table in Appendix C. A box plot is used to show the results from the vertical displacement of the CoM, because the dierences are too small to visualize in a bar plot. Also, the results from the statistical analysis will be summarized in this section. The full result from this analysis are presented in Appendix D.

4.1 Mean and SD of the results

Overall, the mean values of the results are higher in the BSCP group and lower in the TD group. Also, the SD is larger in the BSCP group than in the TD group in all the included methods.

4.1.1 The BSCP-group

If the results from the quiet standing task is compared to the attention task, the mean of the attention task is lower in the following methods in the BSCP group: TMV, CoGacc AP and CoP-CoG AP (Figure 2, 4 and 6). In contrast, the mean is higher in the Area, CoGacc ML and CoP-CoG ML methods for this group (Figure 3, 5 and 7). The vertical displacement of the CoM is higher in the attention task than in the quiet standing task, which both the mean (Appendix C) and the median (Figure 8) of the result is indicating.

4.1.2 The TD-group

In the TD group, the mean of the attention task is higher than (or similar to) the mean of the quiet standing task in all methods. Also, the mean of the blindfold task is higher than the mean of the quiet standing task in all methods. The exception is the CoM v.

disp. where both the mean and the median is lower than (or similar to) the normal task in both the blindfold task and the attention task (Figure 8).

Figure 2: Mean and SD of both groups from

the TMV method for each task performed. Figure 3: Mean and SD of both groups from the Area method for each task performed.

(24)

4.2 Results from the statistical analysis 4 RESULTS

Figure 4: Mean and SD of both groups from the CoGacc method in the AP direction for each task performed.

Figure 5: Mean and SD of both groups from the CoGacc method in the ML direction for each task performed.

Figure 6: Mean and SD of both groups from the CoP-CoG method in the AP direction for each task performed.

Figure 7: Mean and SD of both groups from the CoP-CoG method in the ML direction for each task performed.

4.2 Results from the statistical analysis

When comparing the BSCP and the TD group, signicant dierences were found in all methods used (P<0.005) except in the CoM v. disp. method. In this method no signicant dierence was found except in the quiet standing task (Pnormal = 0.047, Pblindf old = 0.058, Pattention = 0.12).

The BSCP group showed signicant dierences (P<0.05) from the quiet standing task in two of the methods using a blindfold, the area method and the CoM v. disp. In all other methods and tasks there were no signicant dierences found in this group (P = 0.33-0.88).

In the TD group signicant dierences from the quiet standing task were found in the following methods; CoGacc AP using a blindfold (P<0.05) and the CoM v. disp., both using a blindfold and during the attention task (Pblindf old, Pattention <0.005). In the blindfold

12

(25)

Figure 8: Boxplot of the displacement of the CoM in the vertical direction. The plot shows the median (red line), the 25th and 75th percentile (lower and upper limits of the box), the lowest and highest value (bottom and top lines) and the outliers (red crosses).

task using the TMV method the p-value was close to signicant (P = 0.052). Otherwise, no signicant dierences were found in the other methods and tasks (P = 0.13-0.91).

No signicant dierences were found when comparing the dierences between the quiet standing task and the blindfold task, and the quiet standing task and the attention task (P = 0.10-0.94), between the BSCP group and the TD group.

(26)

(27)

5 Discussion

In this study the postural stability of children with BSCP was assessed using both the CoP and the CoM in the methods. The results show that all the methods used can clearly distinguish between the BSCP group and the TD group. The vertical displacement of the CoM, which is not commonly used in this kind of studies, is an exception. The results show no signicant dierences in the vertical displacement of the CoM between the two groups. In agreement with previous studies the postural stability is better in TD children than in children with BSCP.

5.1 Postural stability during the attention task

The result shows an improvement of the postural stability in the BSCP group during the attention task in several of the methods; TMV, CoGacc AP and CoP-CoG AP. The TMV is a method that has been validated in several previous studies and is a measure that includes the total sway path. The decrease of TMV shows an overall improvement in the postural stability of the children with BSCP during this task. In contrast, the TD children had an increased TMV, SD of the CoGacc and SD of the CoP-CoG (in both directions) during the attention task. This may indicate a change in balance strategy of the children in the BSCP group while performing the attention task, resulting in a better postural stability overall.

At the same time the area, CoGacc in the ML direction and the distance between CoP and CoG in the ML direction increased (indicating a decreased postural stability in this direction). This might be a result of shifting from an ankle strategy, where mainly the AP sway is aected, to a hip strategy aecting the sway in the ML direction.

5.2 Postural stability during the blindfold task

The blindfolded task seems to aect the postural stability greater than the other tasks in both groups, but the eect is not the same. In the BSCP group the postural stability increase during the blindfold task, except in the result from the TMV method, where the stability decrease. The combination of a decreased velocity of the CoP and a smaller sway area may again be a result from a changed balance strategy. The CoM has a lower acceleration (and therefore also a lower velocity) while the CoP has a higher velocity. This result shows that it might be useful to use both the CoP and the CoM since the results are not in agreement. These results may show dierent aspects of the changed balance strategy of this group of children.

The mean values of the CoGacc and the CoP-CoG results decreased in both AP and ML directions during the blindfold task which indicates improved postural stability. This is not in agreement with previous literature, where it is more common for the postural stability to decrease when the vision is excluded. In the TD group, the mean values are all higher during the blindfolded task when comparing with the quiet standing. The higher values indicate a decreased postural stability, which is in agreement with previous literature.

(28)

5.3 The crouched position 5 DISCUSSION

5.3 The crouched position

The mean vertical displacement of the CoM is signicantly lower in the BSCP group during the blindfold task (compared to the quiet standing task), which is in agreement with previous literature. The results also show that the children with BSCP have an elevated mean value of the vertical CoM displacement when they are watching the movie. This indicates that they straighten up a little bit when they focus on something outside of the body instead of just being focused on their posture. Even though there is no signicant dierence between the quiet standing task and the attention task, this result diers from the TD group, where the displacement is signicantly lower in both the blindfolded task and the attention task.

A theory that was discussed during visits at Karolinska institutet, with C. Lidbeck and Å. Bartonek, is that the children with BSCP can experience distrust in their own ability to keep balance. This may result in an increased crouching position if they are confronted with an additional challenge in posture (like for instance a blindfold), as an attempt to feel more secure. The same theory might explain why the vertical CoM displacement is elevated during the attention-task. When the children focus on something, their attention is averted from the insecurity about keeping their balance.

5.4 Suggestions for future studies

The methods used all targeted dierent aspects in the CoP and CoM data and therefore, gave dierent results. A clear example of two methods where the results disagree with each other is during the blindfolded task in the BSCP group. The TMV indicates a decreased postural stability, while the other methods indicate an increased postural stability. This does not have to mean that one of them is wrong, it might just show dierent aspects of the children's strategies to retain their balance. For instance, it is common in children with this kind of disability to distrust their ability to stand upright. When the children have to wear a blindfold it might result in their body to stien which may result in a smaller sway area. To compensate for the loss of vision the velocity of the CoP increases which make it possible to remain standing.

In the TD group, the results from the dierent methods all agreed with each other. This indicates that in studies of postural stability in TD children, it might not be necessary to use several dierent methods, at least in trials of quiet standing. This, of course, depends on what you want to obtain from the results. But in studies of postural stability in children with BSCP it might be necessary to use a few dierent methods and consider the results from all methods as a whole. If, for instance, only the area method had been used, it would only show us part of the picture.

As a suggestion for future studies, I think that both CoP and CoM methods should be used in comparison. The TMV is a method that is widely validated in previous literature and showed dierent results from the other methods during the blinfolded task. The area, CoGacc and CoP-CoG methods all showed similar results, but the statistical results from the area method were more extreme. The results from the two methods using the CoM data are correlated with each other, making it unnecessary to use both at the same trial. I want to suggest the methods; TMV, Area and either the CoGacc or the CoP-CoG method. These three ways to assess postural stability can in combination give a bigger

16

(29)

picture of the postural stability of children with BSCP.

The relevance for the vertical displacement of the CoM depends on the research question.

In this thesis it was of interest to measure the hight of the vertical CoM displacement and I think it might be to other studies of this particular group of children as well. Even though there only were small changes, the results were dierent from the TD group and should be further investigated.

If these methods that have been suggested, in combination with the vertical displacement of the CoM, can be used together in future studies it might increase the understanding of the balance impairments that often acompany this specic group of children. In these results it is clear that the balance strategies of the BSCP group is dierent from the TD group and therefore needs to be assessed by appropriate methods that may dier from methods used in previous studies.

5.5 Sources of error

When comparing the changes between the groups there were no signicant dierences.

This might be a result from the size of the groups, in particular the BSCP group which was even smaller than the TD group. Also, the BSCP group was more heterogeneous with larger SDs in the bar plots, which also suggests that a larger group would be benecial to nding more signicant dierences.

The methods were chosen based on the literature study performed as a part of this thesis.

The number of methods was limited because the time, therefore some interesting methods were excluded. For instance, frequency-based methods might be of interest for future studies of this particular group of children.

The arms and hands did not have markers, which can slightly aect the CoM. But since this was a standing trial and the arms were passive in all three tasks, this should not have a large impact on the results.

(30)

(31)

6 Conclusion

Results show that all methods used can distinguish between the BSCP group and the TD group, except the vertical displacement of the CoM. When comparing within the groups the tasks aect the postural stability dierently between the two groups. As a suggestion, the TMV, the Area and either the CoGacc or the CoP-CoG methods can be used in future studies of children with BSCP. These three methods highlighted dierent aspects of the data and in combination they may provide a bigger picture of the postural stability of children with BSCP. Even though there were no signicant dierence in the vertical displacement of the CoM between the BSCP and the TD group, the CoM was slightly elevated during the attention-task in the BSCP group. In the TD group the CoM was lowered during the same task. This indicates that the children with BSCP in this study straighten up a bit when they can focus on something outside of their own body.

Therefore, the vertical CoM displacement is also to be considered as a method for balance assessment in future studies.

(32)

(33)

Appendices

A Background

To be able to maintain balance, a combination of several complex systems are required.

In the following sections a brief overview of balance and posture will be presented as well as a brief explanation of the diagnosis of CP and the consequences in posture due to this motor disorder. At last a number of common methods to measure postural stability will be explained and some recent studies will be mentioned.

A.1 Balance and posture

Postural stability is dened as the ability to maintain the CoM within an area with specic limits referred to as stability limits. This area is principally dened by the length of an individual's feet and the width between them but the stability limits can change depending on the individual's biomechanics, the task performed and various aspects of the environment [11]. The CoM can be described as a hypothetical point where the subjects total mass is visualized to be concentrated and will be described further in section A.5. If this point moves outside of the stability limits the subject will lose balance and have to adjust the body some way to regain it [1]. In a clinical context postural stability assessments are often used when examining sports-related concussions or age- related balance disorders [12, 13].

A.1.1 Quiet standing

The body's position when standing still is often referred to as quiet stance or quiet standing. Even when we are standing perfectly still, the body has to make adjustments to keep the balance using motor and sensory strategies. In a vertical position the gravity is working against us, therefore the body needs to have strategies to remain upright.

Motor strategies include alignment, muscle tone and postural tone and helps to control the body's position in space. The alignment is minimizing the eect of the gravity while the muscles keep us from collapsing. There are also sensory strategies, including visual, vestibular and somatosensory inputs that organize the sensory information. These two strategies do not work one at a time, but rather collaborate and are constantly giving feedback to one another through the central nervous system (CNS) [11].

A.1.2 Motor strategies

The ideal alignment allows the body to maintain equilibrium and to use the least amount of energy possible to remain upright. The muscle tone refers to the force that keeps the muscles from lengthening and can be tested by extending and exing a relaxed patients limbs to record the resistance of the muscles. Postural tone is communicating with the somatosensory, visual and vestibular system and uses active muscles to counteract the gravity. When balance is lost, the body primarily uses movement in the ankles and the hips to restore the CoM to a point of stability [11].

(34)

A.1 Balance and posture A BACKGROUND

A.1.3 Sensory strategies

To be able to control orientation and navigation in space the sensory system is crucial [14]. Inputs from the visual, vestibular and somatosensory systems can provide the CNS with the information needed to give feedback to the motor control system.

The visual system gives us information about the position of the head and surrounding objects [11]. But we can still keep the balance when the eyes are closed thanks to com- pensation from other sensory systems, like the vestibular system [15]. Also, visual signals may be misinterpreted by the brain. For example when you perceive you are moving while standing on a train platform, because a train is moving beside you. In this case, the visual system tricks the CNS to send out the wrong feedback to make you experience movement. Also, when the eyes are closed or blindfolded the body sway increases (based on CoP amplitude) [1].

The vestibular system gives information about position and movement of the head with respect to inertial and gravity forces. There are two types of receptors, the semicircular canals and the otoliths. Angular acceleration of the head is sensed by the semicircular canals and are good at detecting fast head movements. The otoliths can detect linear position and acceleration in horizontal and vertical directions. They are sensitive to slow movements such as postural sway. The vestibular system is a major part of the sensory strategies but without input from the other systems it would be impossible to e.g. distinguish between a head nod and a forward bow [11].

The somatosensory system modulates the body's position in space with reference to supporting surfaces using dierent receptors. The proprioceptors are experts in sensing changes in length and tension in the muscles, communicate the relationship between the segments and the body position in space. The dierent proprioceptors are muscle, joint and pressure receptors. Cutaneous receptors give information from external stimuli such as touch [11, 3].

A.1.4 Balance impairments

There are many pathologies that create challenges in the everyday life regarding the ability to maintain balance. A few of them are chronic ankle sprains, chronic degenerative low back pain, scoliosis, paroxysmal positional vertigo, head injury, stroke, cerebellar disease, Parkinson's disease, vestibular decits, peripheral neuropathies, amputation, and CP. Most neuromusculoskeletal diseases result in degeneration in the balance control system, and due to the CNSs ability to adapt, this may not be apparent until the patient is deprived of the compensating system. One of the largest group that is experiencing degeneration in balance control is the elderly, primarily due to neuronal decay. An increasingly ageing population with a longer life expectancy challenges researchers and clinicians to understand more about how the balance system works and has resulted in many research projects that target the balance system [1, 11]. The number of publications on postural decit in children with CP has also increased during the last decade and there has been evidence that conrms a lack of postural control in this group [2].

ii

(35)

A.2 Cerebral palsy

In developed countries CP is the most common childhood motor impairment. The impairments are often recognized before the age of 18 months as diculties in learning to sit, stand and walk. The diculties with static and dynamic motor activities are often presumed to be connected to motor disorders, but the disorders are often accompanied by disturbances in sensation, perception and cognition [16]. For example, about 48% of children with CP also have disturbances in the visual system, compared to 4-5% of TD children [17].

A.2.1 Denition

There have been several attempts to dene the term CP in the past. The diagnosis can vary in severity, patterns of motor involvement, associated impairments such as communication-level, intellectual ability, and epilepsy. The following denition has been proposed by the International Executive Committee for the Denition of CP:

"CP describes a group of permanent disorders of the development of movement and posture, causing activity limitation, that are attributed to non-progressive disturbances that occurred in the developing fetal or infant brain. The motor disorders of CP are often accompanied by disturbances of sensation, perception, cognition, communication and be- haviour, by epilepsy, and by secondary musculoskeletal problems" [16].

Attempts have been made to improve physical function of muscles in individuals with CP, e.g. with stretching of sti muscles, but there is still no clinical signicant evidence in favor of this kind of therapy. On the other hand there is evidence of the disadvantages, such as pain. Colver et al. state that it should be recognized that quality of life and participation is what individuals with CP seek and not improved physical function for its own sake. [18]

A.2.2 Prevalence and incidence

CP is the most common physical disability in children with a prevalence, that has remained steady in the last 40 years, of about 2-3 cases per 1000 live births [18, 19, 20, 21]. The incidence is much higher in premature babies and 75-80% of the cases are due to prenatal injury [20].

A.2.3 Classication systems

CP has traditionally been described based on the resulting dysfunction and the location of said dysfunction. The three main motor types are spastic, ataxic or dyskinetic where spastic is the most common type [16]. Spastic CP is characterized by spasticity (muscle contractions) that results in tightness on one or both sides of the body [21]. The location, or topography, used to be divided into hemiplegia (where only one side of the body is aected), diplegia (where both sides but usually only the lower extremities are involved) and quadriplegia (where both arms and legs are involved). A more recent way to divide the location of the impairment is recommended by Rosenbaum et al. [16]; unilateral, if the impairment only aects one side, and bilateral, if both sides are involved. According to the Surveillance of CP in Europe (SCPE) BSCP is the most common combination of

(36)

A.2 Cerebral palsy A BACKGROUND

location and motor type [21].

In 2001 the World Health Organization (WHO) published a new model to be used as a classicaton tool. It can be used in a way that integrates several perspectives of this di- verse group of childhood disabilities [22]. The International Classication of Functioning, Disability and Health (ICF) takes several constructs into account including body struc- ture and function, activity limitation, participation, environmental factors and personal factors [18].

Another method was developed as a response to the need of a standardized system to measure the severity of the movement disability in children with CP [23]. The Gross Motor Function Classication System (GMFCS) has since been widely employed internationally and has been found to be a reliable and valid system that classies children with CP by their age-specic motor ability [24, 25]. The system describes the functional ability in ve dierent levels. The children on level I can perform all activity of their age matched peers only with some diculty in speed, balance and coordination while the children on level V have problem controlling their head and trunk and have diculties with involuntary movements [23].

A.2.4 Consequences in posture

Children with CP tend to have more postural sway while quietly standing compared to TD children. This is the consequence of several factors such as a crouched posture, decreased ability to recover balance, delayed response in ankle muscles and inappropriate muscle response sequencing [26]. Children with CP usually learn physical milestones later in life compared with TD children. Because of this they develop compensatory mechanisms to cope with their primary decits, such as leaning backwards or forwards or even medial or lateral shifts. These compensations might in turn lead to secondary impairments, such as muscle imbalance and poor alignment across joints, further muscle weakness and a contracture in the joints [2].

Figure 9: Postural patterns in the children with bilateral CP. Forward-leaning, backward-leaning end balanced posture. Illustration is reprinted with permission from the editorial. [2]

In a recent article, Domagalska-Szopa and Szopa investigate the characteristics of body alignment and postural orientation during quiet standing in children with bilateral CP.

iv

(37)

The results of their research showed three postural compensatory strategies in the examined group (Figure 9); a forward-leaning of the trunk and pelvis (44%), a backward-leaning of the trunk and pelvis (28%) and a balanced posture (28%). The participating children were classied into GMFCS level I (n=34) and II (n=24). [2].

A.3 Balance in biomechanics

The biomechanics of human movement can be described as an interdisciplinary that describes, analyzes and assesses human movement. The physical and biological principles that apply are the same but can be applied to a wide variety of movements [27]. Even small movements, such as the sway of the body during quiet standing, can be analyzed with these principles if you have a suitable method. Ideally, the methods used should be specic for the group that is being analyzed and will be discussed further later in this section. Postural steadiness is often evaluated with methods based on displacement of the CoP, but there are some scientists that believe that both the CoP and the CoM need to be considered in these kinds of evaluations [4].

In biomechanics, balance means that all the torques and forces acting on the body are equal to zero. These forces can be either external or internal. The external forces are for instance the gravity acting on the entire body and the ground reaction force (GRF) acting upwards on surfaces in contact with the ground. The internal forces are for instance breathing, heartbeats or activation of the muscles for dierent tasks to control the body's movements and maintenance [5]. In general, the balance control is adjusted by the ankles in the AP direction and by the hips in the ML direction [1].

A.3.1 Center of pressure

The CoP can be dened as "the location of the vertical location of the vertical reaction vector on the surface of a force platform on which the subject stands" and represents a weighted average of all surface in contact with the ground. The CoP can never travel outside of the stability limits, if it does the balance is lost and we need to take a step or change posture to regain stability [9].

The GRF and the CoP can be obtained with the use of force plates, which is used as the gold standard for this kind of data collection [28]. Usually a force plate consists of a square board with some kind of sensors that can measure the three force components (Fx, Fy and Fz) and the three torques (Mx, My, and Mz) acting on the force plate. [5]

When there is a single force plate it measures the total CoP of both feet. When there are multiple force plates (one for each foot), the total CoP has to be summarized.

A.3.2 Center of mass

The CoM can be dened as the point in the body where the total mass is evenly dis- tributed, therefore it is also the balancing point in the body where the sum of the torques equal zero. It is a theoretical point that change its location during time as a result from all the dierent body segment's movements [1, 30]. Because you have to consider relative motions of each part of the body proportionally to their masses, the CoM can be a bit complex to estimate [4]. Some scientists claim that the CoM is the only variable that characterizes body sway, but despite this it is still not used as frequently as the CoP in

(38)

A.3 Balance in biomechanics A BACKGROUND

Figure 10: Locations of the CoP, CoM and CoG (P) while walking. The illustration is reprinted with permission from the editorial. [29]

balance assessments. This might be due to complex and time-consuming calculations and the need for expensive equipment [5].

The vertical projection of CoM onto the ground is called the center of gravity (CoG) and is often a variable of interest when measuring body sway. The CoG is simply the x- and y-coordinates of the CoM. In Figure 10 the CoP, CoM and the CoG are illustrated during gait to make the locations of each point clear. In general, there is no interest in the CoM variation in the vertical direction, since the sway is shorter then in the horizontal directions [5]. Since the CoM is totally independent of the velocity or total acceleration of the body, so is the CoG. Measuring the AP and the ML dierences between the CoP and the CoG is suggested to be a measure of actual body sway. The shorter the sway frequencies of the body, the shorter is the distance between these two variables. If you look at the amplitude of the CoP- and the CoG-displacements, the CoP will always be larger and have a higher frequency (Figure 11) [1, 5, 31].

Figure 11: The CoP and the CoG (horizontal components of the CoM), illustrating the dierence between the two variables. The CoP signal is always slightly outside of the CoG signal, reacting to changes in the body's CoM.

vi

(39)

A.3.3 Finding the CoM

To get a correct 3D estimation of displacement of CoM, the only technique is optical tracking of as many body segments as possible. This can be done by using reective markers placed on specic places on a person's body and then record with a 3D motion analysis system. Most of segments are fairly rigid and easy to capture, but e.g. the trunk needs several segments because of movement from respiratory and cardiac functions. [27, 32]

According to D.A Winter an example of a complete measurement is an 21-marker, 14- segment model [27]. There have been attempts to simplify these models, e.g. in a study of a model with only 13-markers and 9-segments. The results were close to the complete model when used on healthy populations in studies of quiet stance, but it was not as accurate [32].

To calculate the CoM the most common method is the segmental method. First the CoM of each individual segment is calculated, then the total CoM can be calculated as the sum of all segments. For this method, knowledge of each segment's weight and mass is required and are usually acquired with anthropometric data. The anthropometric data shows what proportions each segment has as a percentage of body weight or height and can estimate where the CoM is placed in the segment. Some of this data is acquired directly from cadavers, some from mathematic geometric modeling and some from mass scanning. This is usually applied to the data collected in 3D-motion analyses, together with information about the subject's weight and height, to nd the CoM. [30]

A.3.4 CoP-methods to measure postural stability

By far the most common way to measure postural control is with methods using the displacement of the CoP [5]. According to a study performed by Crétual in 2015 about two thirds of the studies on postural stability use this kind of methods [4]. The main reason for this is that the CoP can be measured directly and is easily obtained without the use of an expensive 3D motion tracking system [33]. The methods used to measure CoP displacement are usually divided into either time-domain or frequency-domain measures [34].

The distance traveled by the CoP can be used in dierent ways to examine postural stability. The path length for instance (Figure 12) is based on the total distance travelled by the CoP during the total measurement. The smaller path length, the better the postural stability. It is considered to be a valid measurement of postural stability in several populations studied [35, 28]. Amplitude of displacement is another way to look at the distance where the maximum and minimum CoP displacement in the AP and ML directions are measured, usually around the mean CoP (Figure 13). The smaller distance from the mean value, the better the postural stability. This is regarded as a reliable parameter and used for instance when analyzing postural instability in patient with neuromotor disorders, such as CP [35].

Velocity or acceleration of the CoP is another way to use the distance measurements. The TMV is calculated by dividing the total sway path by the time of the trial. It shows the eciency of the postural control system to maintain stability, the higher the velocity, the worse postural stability. Doyle et al. recommend a trial of at least 60 s in general, but velocity is a reliable measure even at shorter trials, like 30 s [36]. TMV is known as the method with the highest reliability among trials of postural stability in quiet standing [5].

(40)

Figure 12: Illustration of a typical sway path. The path length is the total distance of the sway path.

Figure 13: Illustration of a typical amplitude curve in AP and ML directions.

The SD and RMS measures the standard deviation and average absolute displacement around the mean value. An increased value implies a decreased postural stability. It has been used by numerous researchers and can easily be combined with other methods such as distance or velocity-based calculations. [35, 31].

The area of the CoP displacement can be calculated in several dierent ways. A common way is to calculate an ellipse that contains 90-95% of the CoP data in the AP and ML directions, often referred to as a 95% prediction ellipse (Figure 14). A smaller surface means less sway and a better postural stability [35, 5].

By using the Fourier analysis of a signal, it is possible to obtain information about the frequencies that compose the signal. A higher frequency indicates faster and smaller postural adjustments. Mean and median frequency is suggested to show ankle stiness,

viii

(41)

Figure 14: Illustration of a 95% prediction ellipse used to measure area of the sway path.

the higher the frequency the more stiness around the ankles [5, 35].

A.3.5 CoM-methods to measure postural stability

In recent studies, it is shown that the CoM plays a central part in postural control when joints above the ankle are involved, as it is suggested that quiet standing is a multi-joint motor task [37]. It is also shown that the horizontal CoMacc is very sensitive to age- or disease-related changes in the postural control system [6, 7]. Unfortunately, the CoM is less common in studies of postural stability than the CoP. According to Crétual only 2.4% of studies use CoM displacement and 7.5% use a combination of CoP and CoM in contrast to 63.9% that only use CoP displacement. Crétual suggests that bioengineering development is necessary to make CoM calculations easier to obtain, and might be a future solution to this problem [4].

Already in 1995 Winter underlined the dierence between the CoP and the CoM and that several researchers often wrongfully refers to the CoP displacement directly as postural sway. He brought up that several studies predicted that the dierence between CoP and CoM is correlated to the CoMacc. The dierence between these two variables (CoP- CoM) in the horizontal direction can be considered the error signal that the balance control system is sensing (Figure 11). Both the magnitude and frequency of this error signal is of importance [1].

A number of scientists refer to Winter in their studies and use the relationship between CoM and CoP when assessing postural sway. Smetanin et al. for instance compared how posture maintenance was aected by dierent visual conditions. They used both the trajectories of the CoG and the dierences between the CoP and the CoG. The CoG was considered a controlled variable and the dierence between the CoP and CoG was supposed to reect the resultant stiness in the ankle joints [38].

Another example of method is to use the CoMacc. Oba et al. examined the development

(42)

of postural control in dierent age groups using this variable to investigate children's postural control during quiet standing. Both the CoMacc and the CoP displacements in the AP directions were found to be able to distinguish between children and young adults as well as dierent visual conditions. The CoM acceleration was also found to have a higher sway frequency in children then in young adults, opening up for future studies using this variable [37].

The SD of the CoMacc was used by Yu et al. in 2008, as well as the dierence between the CoP and the CoM, to examine dierences in postural stability in post-stroke patients and healthy groups with elderly and young people. It was concluded that CoP-CoM dierence and CoMacc was highly correlated in both AP and ML directions and that the CoMacc could distinguish between post-stroke patients, the elderly and the young groups. Yu et al. propose that the CoMacc can be used as an alternative to CoP-CoM dierence in future studies [7].

In 2007 Masani et al. used the SD of the CoP-CoG dierence and the SD of the CoMacc to assess that both variables increase with age. They proposed that, since aging aects the control mechanisms of balance, the CoP-CoG dierence must be aected. And therefore, the CoMacc must be aected as well. They found that the hypothesis was true, the SD of both variables were larger in the elderly than in the young. They also found that the CoP-CoG dierence was proportional to the CoMacc in both groups [6].

x

(43)

B Participant data

Participant data from the children with BSCP included in this study:

Subject Gender Age Weight Height GMFCS Glasses

P1 M 11.4 38.5 141.5 2 No

P2 F 14.3 54.0 164.0 2 No

P3 F 15.9 46.0 158.0 3 No

P4 F 11.2 34.1 136.0 2 Yes

P5 M 15.9 99.7 166.0 2 Yes

P6 M 8.6 32.6 132.0 1 No

P7 M 7.7 30.7 124.0 3 Yes

P8 M 9.0 35.2 130.0 2 Yes

P9 F 12.8 46.3 157.0 1 No

P10 M 10.2 31.6 133.5 1 No

P11 M 14.3 42.0 161.0 2 Yes

P12 M 8.3 32.4 131.0 2 No

P13 M 12.4 45.0 157.0 2 No

P14 F 10.4 52.7 145.0 1 No

P15 F 16.0 49.3 164.0 1 No

P16 M 12.2 37.2 157.0 2 No

Participant data from the TD children included in this study:

Subject Gender Age Weight Height

C1 M 16.9 61.4 175.5

C2 M 16.9 62.7 181.0

C3 F 14.3 40.8 156.0

C4 F 6.9 28.5 121.0

C5 M 14.3 62.0 179.0

C6 M 14.4 72.5 170.0

C7 F 6.9 24.3 123.5

C8 F 8.8 41.9 145.5

C9 F 13.3 55.6 159.0

C10 F 8.8 36.7 137.5

C11 F 7.5 23.7 123.5

C12 F 9.1 24.9 128.5

C13 F 6.7 22.3 121.0

C14 F 9.9 29.9 143.5

C15 M 7.8 23.4 124.0

C16 M 7.4 25.3 122.0

C17 M 9.5 31.2 135.5

C18 M 12.1 40.4 153.0

C19 M 12.3 46.2 159.5

C20 M 11.7 40.5 162.5

(44)

C MEAN AND SD OF THE RESULTS

C Mean and SD of the results

Mean and SD of the results from each method used to measure postural stability:

BSCP TD

Method Task Mean ± SD Mean ± SD

TMV (cm/s) Quiet 2.52±0.86 1.39±0.57

Blindfold 2.73±1.24 1.55±0.42 Attention 2.39±1.00 1.50±0.70 Area (cm²) Quiet 18.67±10.49 4.22±4.16 Blindfold 13.26±7.89 4.40±4.55 Attention 18.95±16.40 4.90±6.43 CoG Acc AP (cm/s²) Quiet 2.19±1.33 0.86±0.55 Blindfold 2.13±1.06 1.01±0.43 Attention 1.87±0.98 0.98±0.62 CoG Acc ML (cm/s²) Quiet 2.35±1.52 0.59±0.50 Blindfold 2.01±1.29 0.63±0.36 Attention 2.35±1.14 0.80±0.87 CoP-CoG AP (cm) Quiet 0.43±0.16 0.22±0.09 Blindfold 0.41±0.13 0.22±0.06 Attention 0.40±0.19 0.22±0.09 CoP-CoG ML (cm) Quiet 0.40±0.18 0.18±0.08 Blindfold 0.37±0.16 0.19±0.06 Attention 0.42±0.21 0.19±0.11 CoM v. disp. (% of BH) Quiet 54.62±1.49 55.50±0.80

Blindfold 54.51±1.59 55.38±0.81 Attention 54.70±1.39 55.42±0.79

xii

(45)

D Results from statistical analysis

P-values and signicance of the results from the quiet standing task compared to the blindfolded/attention task using Wilcoxon signed-rank test:

BSCP TD

Method Task P-value Sign. P-value Sign.

TMV (cm/s) Quiet/Blindfold 0.53 0.052

Quiet/Attention 0.72 0.20

Area (cm²) Quiet/Blindfold 0.034 x 0.30

CoG Acc AP (cm/s²) Quiet/Blindfold 0.88 0.033 x Quiet/Attention 0.44 0.26

CoG Acc ML (cm/s²) Quiet/Blindfold 0.47 0.13 Quiet/Attention 0.84 0.16

CoP-CoG AP (cm) Quiet/Blindfold 0.61 0.48

CoP-CoG ML (cm) Quiet/Blindfold 0.80 0.23

CoM v. disp. (% of BH) Quiet/Blindfold 0.020 x 1.32e-3 x Quiet/Attention 0.76 3.59e-3 x

P-values and signicance of the results from the BSCP group compared to the TD group using Mann-Whitney U test:

Method Task P-value Sign.

TMV (cm/s) Quiet 6.21e-4 x

Blindfold 5.52e-4 x Attention 1.71e-3 x

Area (cm²) Quiet 1.85e-5 x

Blindfold 3.26e-5 x Attention 1.84e-4 x CoG Acc AP (cm/s²) Quiet 1.23e-3 x Blindfold 2.36e-4 x Attention 6.98e-4 x CoG Acc ML (cm/s²) Quiet 4.93e-5 x Blindfold 9.62e-5 x Attention 5.64e-5 x

CoP-CoG AP (cm) Quiet 6.46e-5 x

Blindfold 3.26e-5 x Attention 1.62e-4 x

CoP-CoG ML (cm) Quiet 2.46e-5 x

Blindfold 4.93e-5 x Attention 3.42e-4 x CoM v. disp. (% of BH) Quiet 0.047 x

Blindfold 0.058 Attention 0.12

(46)

D RESULTS FROM STATISTICAL ANALYSIS

P-values and signicance of the results from comparing dierences between the quiet standing task and the blindfold/attention task in the BSCP group and the TD group:

Method Task P-value Sign.

TMV (cm/s) Di. Quiet/Blindfold 0.86 Di. Quiet/Attention 0.30 Area (cm²) Di. Quiet/Blindfold 0.11 Di. Quiet/Attention 0.38 CoG Acc AP (cm/s²) Di. Quiet/Blindfold 0.71 Di. Quiet/Attention 0.53 CoG Acc ML (cm/s²) Di. Quiet/Blindfold 0.38 Di. Quiet/Attention 0.64 CoP-CoG AP (cm) Di. Quiet/Blindfold 0.45 Di. Quiet/Attention 0.40 CoP-CoG ML (cm) Di. Quiet/Blindfold 0.94 Di. Quiet/Attention 0.79 CoM v. disp. (% of BH) Di. Quiet/Blindfold 0.81Di. Quiet/Attention 0.10

xiv

(47)

2017-12-14 10:35 C:\Users...\CoP_ReadFile.m 1 of 4

function [CoP_ML,CoP_AP,time,S_CoP_ML,S_CoP_AP,S_time] = CoP_ReadFile (x,length,start)

%x = the current file

%length = the length of the current file

%start = the starting point of the current file

%Import data divided by "," and skip two rows input_CoP = importdata(x,',', 2);

DATA_CoP = input_CoP.data;

%CoP in centimeters

CoP_1_raw = [DATA_CoP(:,2)./10 DATA_CoP(:,3)./10]; %Foot 1 - CoP in cm CoP_2_raw = [DATA_CoP(:,14)./10 DATA_CoP(:,15)./10]; %Foot 2 - CoP in cm

%GRF columns

GRF1 = DATA_CoP(:,10); %Foot 1 GRF2 = DATA_CoP(:,22); %Foot 2

%prelocate CoP, GRF and time columns CoP_1 = zeros(length*10,2);

CoP_2 = zeros(length*10,2);

GRF_1 = zeros(length*10,1);

GRF_2 = zeros(length*10,1);

time = zeros(length*10,1);

k = 1;

%Put CoP-data and GRF-data into arrays + create 30 s time-vector for i=(start*10):1:(length*10)+(start*10-1)

CoP_1(k,1) = CoP_1_raw(i,1);

CoP_1(k,2) = CoP_1_raw(i,2);

CoP_2(k,1) = CoP_2_raw(i,1);

CoP_2(k,2) = CoP_2_raw(i,2);

E.1 Read CoP-data

(48)

2017-12-14 10:35 C:\Users...\CoP_ReadFile.m 2 of 4

GRF_1(k,1) = GRF1(i,1);

GRF_2(k,1) = GRF2(i,1);

time(k,1) = k;

k = k+1;

end

%prelocate CoP, GRF and time arrays

S_CoP_1 = zeros(length,2);

S_CoP_2 = zeros(length,2);

S_GRF_1 = zeros(length,1);

S_GRF_2 = zeros(length,1);

S_time = zeros(length,1);

k = 1;

%Put CoP-data and GRF-data into arrays + create 30 s time-vector

%Shorter version to match 100 Hz CoM-data

for i=(start*10):10:(length*10)+((start-1)*10) %To match with 100 Hz CoM samples

S_CoP_1(k,1) = CoP_1_raw(i,1);

S_GRF_1(k,1) = GRF1(i,1);

S_GRF_2(k,1) = GRF2(i,1);

S_time(k,1) = k;

k = k+1;

end

%prelocate total CoP

CoP_tot = zeros(length*10,2);

S_CoP_tot = zeros(length,2);

Balance assessment in children with cerebral palsy; methods for measuring postural stability