Abstract Introduction Within the field of neuromotor rehabilitation, the aim of this thesis was to evaluate the efficacy of a vibro-tactile biofeedback device based on electromyography, in improving motor learing in children and adolescents with dystonia. In 2013, an international committee formed by different personalities with years of experience in the field of dystonia met to clarify terminological and classification aspects related to this pathology, which was defined as follows (A. Albanese et al., 2013): ‘’Dystonia is a movement disorder characterized by sustained or intermittent muscle contractions that cause abnormal, often repetitive movements. Dystonic movements are typically stereotyped, torsional and can be characterized by tremor. Dystonia is often initiated or worsened by voluntary actions and is characterized by the activation of muscles that are not strictly functional to the motor gesture (overflow)’’ The most evident clinical manifestations of the pathology include abnormal postures, slow and not very fluid gestures (sometimes spasmodic and characterized by tremor), increased resistance to passive movement, activations of muscles not directly involved in the ongoing movement (T. D. Sanger et al., 2003) (T. D. Sanger et al., 2010). In terms of brain areas associated with the pathology, it should be emphasized that this is often correlated with lesions at the level of the basal ganglia, in particular putamen and globus pallidus; however, it is also possible to involve other, areas such as the cerebellum and the brainstem (V. K. Neychev et al., 2011) (A. Tewari et al., 2017). Sensory deficits have also been observed to be related to a deficit in the integration of afferent inputs at the central level: this negatively affects the generation of the corresponding motor output (A. Berardelli et al., 1998) (A. Quartarone and M. Hallett , 2013). In terms of etiology, dystonia is classified as primary when it is the most important feature of an idiopathic or an identified genetic disorder, while secondary dystonias are symptomatic disorders arising from another underlying disease, such as cerebral palsy (CP) or acquired brain injury. The physiopathology of dystonia is very variegate; there are evidences that subjects with primary dystonia do not show sensory deficits, whereas, subjects with secondary dystonia are characterized by sensory abnormalities entailing to motor impairments (F. M. Molloy et al., 2003) (T. D. Sanger and S. N. Kukke, 2007). Sensory deficits lead to a lack of improvement in complex motor skills despite repeated practice. The main treatment modalities range from physical and occupational therapy to the use of various types of drugs (botulinum toxin, anticholinergic drugs, dopaminergic drugs, ...) and finally to surgery (Deep Brain Stimulation, DBS): these are approaches exclusively based on the symptomatology presented by the subject and, often, only partially effective in controlling it (A. Albanese et al., 2006) (J. Jankovic, 2013) (M. Bertucco and T. D. Sanger, 2015). In this context, innovative and non-invasive treatment solutions that work at a higher level than the symptomatic level are needed, with the aim of favoring the correct motor learning of the subject. In general, biofeedback (BF) is a technique that consists in providing the subject with increased information of a physiological process, allowing him/her to increase awareness of the process itself and to acquire voluntary control over it (J. V. Basmajian, 1982). The subject is provided with information on the state of a physiological variable (EMG, EEG, strength, position, ...) through sensorial channels of different types (visual, auditory, tactile, ...). The first studies on the use of BF-based treatments for the improvement of motor learning have shown that the individual can acquire control at high levels on the neuromuscular system, even managing the single muscle (J. V. Basmajian, 1963). The neurological mechanisms underlying the BF training in dystonic subjects are not yet completely clear; it is however known that such patients are potentially characterized by inefficiencies in terms of sensory information: a hypothesis is that the use of BF can improve neural plasticity by increasing, thanks to the auxiliary information provided to the subject, the reliability of the incoming signal to sensorimotor areas; the effect is that of reinforcing the correct neural circuits. Rehabilitative tools that provide increased feedback, therefore, are potentially usable for improving motor control in subjects with movement disorders. In dystonic patients, the efficacy of the use of such a strategy has been demonstrated on several occasions (S. J. Young et al., 2011) (C. Casellato et al., 2013). However, to be effective a BF therapy should re-educate motor control during dynamic, functionally purpose-oriented movements (H. Huang et al., 2006): to satisfy this demand, Sanger and Bloom developed (2010), at SangerLab (University of Southern California (USC), Los Angeles, CA), a portable and silent device that provides the subject with a mechanical vibration proportional to the activation of the muscle on which it is positioned and which can therefore be easily used during daily life activities. The present thesis work is part of a multi-center cross over study that involves, besides the Politecnico of Milano, the Carlo Besta Neurological Institute, Milano (MI), the Scientific Institute Eugenio Medea, Bosisio Parini (LC), and the Children's Hospital, Los Angeles (CA). The research project has been approved and is supported by the United States Department of Health (NIH, National Institutes of Health). The coordinator of the entire study is Dr. T. D. Sanger, professor at the University of Southern California (USC, Los Angeles) and neurologist at the Children's Hospital of Los Angeles, who designed and built a battery-powered wearable electromyographic (EMG) feedback device. The objective of the present thesis work is to test the mechanism of action of this EMG-based vibro-tactile biofeedback device and to quantify the effect of increased sensory information during motor learning in children affected by primary and secondary dystonia while performing specific exercises in in the laboratory/clinic, and to compare the effect it produces on patients with its influence on healthy subjects. Subjects suffering from primary dystonia are recruited at Carlo Besta Neurological Institute and acquired at Laboratory of Neuroengineering and Medical Robotics (NearLab) of Politecnico di Milano, while the secondary patients are recruited and acquired at Scientific Institute Eugenio Medea. The aim is to prove that biofeedback improves motor function in children with secondary dystonia. On the other hand, children with primary dystonia (without sensory deficits) are analyzed as a control to show that the improvement is not due to a direct interaction between biofeedback and the motor abnormality in dystonia, and we compare to healthy control subjects to quantify the degree of normalization of function that can be achieved. Hence, the working hypothesis is that a positive effect should be recorded only on the secondary population, characterized by marked sensory deficits. In the case of subjects affected by primary dystonia, the use of a biofeedback strategy should have no clear role in improving performance. At least, the behavior of the tested primaries should be approximately consistent with the behavior of the healthy participants. Methods The acquisition protocol is aimed at testing the effectiveness of the BF device in the short term (1 week) in learning specific motor exercises to be performed in a controlled environment. It consists of two weeks of data acquisition, with a break of at least one month in between. The subject is asked to perform two defined motor exercises, the so-called Figure 8 Task and the Spoon Task, which simulate, respectively, the act of writing and the act of self-feeding. During this thesis work I participated in the entire experimental campaign, but the data analysis was concentrated on the evaluation of the Spoon Task. The subject is required to carry a marble in a spoon back and forth between two targets without dropping the marble. The depth of the spoon defines the difficulty of the excercise and the request is to perform 10 movements (forward and backward) at the maximum possible speed. Each week consists of two Testing sessions (day 1 and day5, which are, respectively, the baseline and the end of the training), in which the motor task are always performed with different degrees of difficulty, and three Training sessions, the intermediate days, during which the task is performed with only one difficulty level (the medium level). The subject wears the BF device (positioned on the dominant arm) only during the Training sessions of one of the two weeks and the order of use of the BF device is randomized between the two weeks. Healthy subjects followed a simplified and more practicable protocol. Their presence in the laboratory was required only during Testing days and they trained at home in the intermediate days. The protocol was approved by both Besta and Medea. However a distinction must be made. Unlike the PoliMi protocol, the Medea protocol provides four consecutive weeks of acquisition. In the first and third weeks the subject plays Figure 8 Task, during the second and fourth weeks the subject plays Spoon Task. The quantitative analysis of the motor exercises is performed through the evaluation of: • the electromyographic activity of eight muscles of the upper limb tested: Flexor Carpi Ulnaris (FCU), Extensor Carpi Radialis (ECR), Bicep (BIC), Tricep (TRIC), Anterior Deltoid (AD), Lateral Deltoid (LD), Posterior Deltoid (PD), Supraspinatus (SS); • the kinematics of the shoulder joint, the elbow joint, the wrist joint and the spoon. Figure 1 shows the experimental setup. Figure 1: Experimental setup The data analysis consisted of an initial pre-processing phase of the EMG signal and of the kinematic signal, leading to obtain the EMG signal envelope for each of the 8 acquired muscles and the reprojected kinematics on the main components plane of each joint. Then, the evaluation of the subject motor performance was based on quantitative indices that describe various aspects of the task: 1) temporal indices: • movement time; • index of performance. Thus, deriving a speed-accuracy trade-off (SATO) is an important metric for assessing motor impairments in dystonia (F. Lunardini et al., 2015). The SATO was examined by changing the spoon size to create different difficulty levels, and the Index of Performance was computed, which reflects the efficiency of the nervous / motor system in facing tasks of increasing difficulty; 2) kinematic descriptors of the final output and of the movements of the entire kinematic chain: • movement speed; • smoothness of the movement in three different formulations, i.e. number of peaks in the velocity profile, the ratio between the mean velocity and the maximum velocity (H. Vikne et al., 2013), and a dimensionless form of the jerk, the Normalized Mean Squared Jerk (N. Hogan, D. Sternad, 2009); • movement linearity, an indirect measure of accuracy, calculated as the ratio between the amplitude of the trajectory generated by the subject and the linear distance between the starting and the ending point of the movement; • movement repeatability, which provides an indication of the subject's ability to perform repeatable movements throughout the trials; • range of motion of the elbow joint. It gives an indication of the subject's upper limb flexion-extension during the movement; • displacement of the shoulder joint, which measures the movement of the shoulder joini in the anterior-posterior direction; 3) indices of the electromyographic activity of three couples of agonist / antagonist muscles: Flexor Carpi Ulnaris and Extensor Carpi Ulnaris, Bicep and Tricep, Anterior Deltoid and Posterior Deltoid. In order to observe if the use of the biofeedback device during the Training impacted on these indices, compared to the simple practice without additional instrumental feedback, a statistical analysis was conducted: a subject-by-subject analysis was carried out to evaluate the effects of learning between day 1 and day 5 of the two weeks of treatment. The variability of each parameter was evaluated through a 2-way analysis of variance (ANOVA), including the condition (BF or noBF) and the day (day 1 and day 5) as fixed factors and the condition by day as interaction effect. A group-based statistical analysis was not performed due to the heterogeneity of the subjects in terms of age and level of impairment. Results and discussion A total of 15 patients were recruited: 9 primary dystonic subjects (5 boys and 4 girls from 7 to 19 years old); 6 secondary dystonic subjects (4 boys and 2 girls from 6 to 16 years old). A group of 12 age-matched healthy subjects were recruited and involved in the protocol: 5 boys and 7 girls from 6 to 19 years old. All participants gave informed written consent for participation. In case of minors, parents were asked to sign the informed consent and the authorization for use of protected health information, videos and images. Figure 2 shows the results of the analysis in terms of the percentage of the healthy, the primary and secondary subjects that showed learning characteristics with the practice, regardless of the use of the device, and better improvement when the BF device was used. Figure 2: Learning / learning with interaction. MT=Movement Time, PEAKS=speed peaks, SMOOTHNESS=smoothness, ML=Movement Linearity, ROM=elbow range of motion, ST=shoulder displacement, CC FCU / ECR=FCU / ECR co-contraction, CC BIC / TRIC=BIC / TRIC co-contraction, CC AD / PD=AD / PD co-contraction. DAY is the day effect, INTER is the interaction effect. 9 panels are reported, related to the outcomes for which the statystical analysis was performed. From left to right, the trend of the healthy (in blue), the primary (in black), the secondary (in red) subjects is shown. For each group, the bar on the left shows the percentage of subjects who learned the task with practice, the bar on the right shows the percentage of the subjects that showed a significant interaction effect in favour of the BF. 2 of the 12 the healthy subjects, 4 of the 9 primary subjects and 1 of the 6 secondary dystonic subjects reduced the movement time with practice. 5 of the 12 healthy subjects, 2 of the 9 primary subjects and 1 of the 6 secondary dystonic subjects showed better improvement when the BF device was used. 3 of the 12 the healthy subjects, 3 of the 9 primary subjects and 1 of the 6 secondary dystonic subjects reduced the number of the speed peaks with practice. 3 of the 12 healthy subjects, 2 of the 9 primary subjects and 1 of the 6 secondary dystonic subjects showed better improvement when the BF device was used. None of the 12 the healthy subjects, 1 of the 9 primary subjects and none of the 6 secondary dystonic subjects increased the smoothness of the movement with practice. 1 of the 12 healthy subjects, 1 of the 9 primary subjects and 1 of the 6 secondary dystonic subjects showed better improvement when the BF device was used. 2 of the 12 the healthy subjects, 3 of the 9 primary subjects and 1 of the 6 secondary dystonic subjects reduced the movement linearity with practice. 2 of the 12 healthy subjects, 1 of the 9 primary subjects and 1 of the 6 secondary dystonic subjects showed better improvement when the BF device was used. For what concerns the elbow ROM, a complete analysis was performed on 5 of the 6 secondary subjects, since kinematic data concerning the spatial coordinates of the shoulder of the second testing day of the second week were missing for one secondary dystonic subject. 4 of the 12 the healthy subjects, none of the 9 primary subjects and 2 of the 5 secondary dystonic subjects increased the elbow ROM with practice. 4 of the 12 healthy subjects, 3 of the 9 primary subjects and 1 of the 5 secondary dystonic subjects showed better improvement when the BF device was used. Even considering the shoulder displacement, a complete analysis was performed on 5 of the 6 secondary subjects, since kinematic data concerning the spatial coordinates of the shoulder of the second testing day of the second week were missing for one secondary dystonic subject. 3 of the 12 the healthy subjects, 1 of the 9 primary subjects and 1 of the 5 secondary dystonic subjects reduced the shoulder displacement during the movement with practice. 4 of the 12 healthy subjects, 3 of the 9 primary subjects and 2 of the 5 secondary dystonic subjects showed better improvement when the BF device was used. For what concerns EMG parameters, a complete analysiy was performed on 4 of the 6 secondary subjects, because all the EMG data were missing, in one subject relating to the second testing day of the first week, in the other secondary dystonic subject related to the first week. 1 of the 12 the healthy subjects, none of the 9 primary subjects and none of the 4 secondary dystonic subjects reduced the FCU/ECR co-contraction with practice. 3 of the 12 healthy subjects, 2 of the 9 primary subjects and 1 of the 4 secondary dystonic subjects showed better improvement when the BF device was used.4 of the 12 the healthy subjects, 2 of the 9 primary subjects and none of the 4 secondary dystonic subjects reduced the BIC/TRIC co-contraction with practice. 3 of the 12 healthy subjects, 3 of the 9 primary subjects and 2 of the 4 secondary dystonic subjects showed better improvement when the BF device was used. 1 of the 12 the healthy subjects, 1 of the 9 primary subjects and none of the 4 secondary dystonic subjects reduced the AD/PD co-contraction with practice. None of the 12 healthy subjects, 4 of the 9 primary subjects and 2 of the 4 secondary dystonic subjects showed better improvement when the BF device was used. For what concerns time and speed of execution, the practice with the BF device influenced the behavior of the primary subjects and the secondary patients almost in the same way, in fact the percentage of the primary subjects for whom a significant interaction effect was found is about the same with respect to the percentage of the secondary dystonic. It can be stated that for the movement time we see learning trend mainly for the primary dystonic subjects. However, considering also the group of healthy subjects, a great learning effect is not visible: less than half of people had a significant day effect. The improvement trend observed looking at the IP is equal to that of the movement time. Thus, it can be stated that the device affected the processing capabilities of the primary patients and the secondary more or less in the same way. Even looking at the smoothness of the movement there’s no evidence to confirm that the exercise performed with the BF produced better results on the secondary group: the effect of the device on the task learning was comparable between patients; learning induced by the practice was higher for the primary patients, but, overall, learning was not so evident, since only one subjects over all the recruited subjects showed learning with the practice. The analysis on the movement linearity confirms what has been said about the movement time and the smoothness. Considering the elbow ROM and the displacement of the shoulder, the results show something different with respect to the final output kinematic parameters. For what concerns the elbow ROM, there was no learning for the primary dystonic patients, who, instead, improved their performance when the BF was used, while, for the secondary patients, a greater percentage of them learned the task with the practice. For what concerns the shoulder displacement, the behavior of the three groups was very similar: both the healthy subjects and the patients (both primary dystonic and secondary dystonic) had better improvement when the BF device was used. However, it is likely that these two parameters do not identify so well the behavior of the subjects, who adopted different strategies that translated into a great inter-sample variability in the movements of the entire kinematic chain: since the position of the wooden block on the support table was not fixed, there were subjects who managed to perform the exercise remaining firmly on the chair and positioning themselves closer to the block, and there were those who, instead, preferred to stay less firmly on the chair and accompanied the movement with the trunk. Initial values of the movement repeatability were more or less equal between primary distonic subjects and secondary distonic patients. There was no clear trend towards learning among the primary subjects and the healthy subjects. It is probably necessary to consider that the range of improvement was limited by the already good performance on the first day of testing. The same type of analysis can be conducted on the secondaries, which did not clearly reveal whether the device was actually effective. However, we must consider that only in one case (S001BoLu) we there was an isolated learning in the week of non-use of the device. Even looking at the electromyographic parameters, there is no clear tendency in favour of the initial hypotheses: indeed, for what concerns both the FCU/ECR muscles co-contraction and the AD/PD muscles co-contraction, better improvement was evident when the BF device was used for both the primary subjects and the secondary subjects; learning induced by the practice was null for the primary patients and visible for only one secondary subjects. Focusing on the co-contraction of the muscles primarly involved in the prono-supination of the wrist, the behavior of the three groups was similar: the healthy subjects and the patients (primary subjects and secondary subjects) better improved their performance when the sensory feedback was provided, while learning with the practice was shown by only one subject (one healthy subject). The trend of the healthy subjects was to increase the FCU/ECR co-contraction with the practice. It is not wrong to assume that only a greater simultaneous activation of theese muscles reflects into greater stability of the wrist joint, and therefore greater control on the spoon. This explains why, if the expected result was a decrease in the co-contraction levels, low levels of learning were found. Finally, considering the BIC/TRIC co-contraction, the secondary subjects did not reveal better improvement while using the BF. Rather, the practice induced a learning of the trained task, but this was not influenced by the use of the device; moreover, for the primary subjects, the BF device influenced the task learning more than the practice did. In general, it can not be stated that an effect related to the use of the device is more pronounced in the group of secondary dystonic: better improvement with the device was recorded for both the primary subjects and the secondary subjects. Even for the healthy subjects the device accelerated learning processes. This was particularly evident considering the time-related parameters and the FCU/ECR co-contraction. For what concerns the temporal paramaters and the parameters related to the final output kinematics, learning in primary dystonic was not significantly stressed by the use of the device, but this trend was not visible considering the parameters linked to the entire kinematic chain (ROM and ST) and the EMG parameters, for which the device accelerated learning. Hence, the results do not allow us to support the hypotheses the study is based on: even individuals affected by secondary dystonia, which are characterized by marked sensory deficits that prevent their improvement in the performance of complex motor tasks (despite repeated training), seem not to be so much favored by using such a strategy; on the other hand, for the subjects affected by primary dystonia and for the healthy subjects, in which there are no sensory deficits and the physiological learning mechanisms linked to the training of a particular motor exercise are intact, the use of a biofeedback strategy seems to have a role in improving performance. However, we are not even in the position to refuse the initial hypotheses. Some limitations still need to be overcome. The first limitation of the study lies in the fact that it does not currently take into consideration cognitive variables that could lead to behavioral disorders and that can prevent the secondaries from accessing sufficient sources for a correct learning. A limitation derives from the lack of some fundamental data for the analysis of some secondary subjects. Furthermore, the lack of a group statistic analysis and the scattered results do not allow for definitive conclusions. This problem is added to the limited number of subjects, which make the sample power limited and the comparison not always statistically robust. However, the ongoing multi-center clinical trial will overcome this limitation and more dystonic subjects will be recruited. Furthermore, the long-term effect of the use of the biofeedback device during daily life activities (wearing the device at least 5 hours a day for one month) will be investigated. Future developments of this work could be: • definition of a set of criteria to guide the operator in choosing the most suitable muscles on which to place the biofeedback; these could be selected, for example, after a clinical and quantitative evaluation that clearly indicates the most compromised ones; • test of simultaneous use of multiple feedback modalities; • test of the effects different sensory channels through which providing the feedback.
Sommario Introduzione Nel campo della riabilitazione neuromotoria, lo scopo di questa tesi è stato di valutare l'efficacia di un dispositivo di biofeedback vibro-tattile basato sull'elettromiografia, nel miglioramento dell’ apprendimento motorio in bambini e adolescenti con distonia. Nel 2013, un comitato internazionale formato da diverse personalità con anni di esperienza nel campo della distonia si è incontrato per chiarire gli aspetti terminologici e di classificazione relativi a questa patologia, che è stata definita come segue (A. Albanese et al., 2013): "La distonia è un disturbo del movimento caratterizzato da contrazioni muscolari prolungate o intermittenti che causano movimenti anormali, spesso ripetitivi. I movimenti distonici sono tipicamente stereotipati, torsionali e possono essere caratterizzati da tremore. La distonia è spesso iniziata o aggravata da azioni volontarie ed è caratterizzata dall'attivazione di muscoli che non sono strettamente funzionali al gesto motorio (overflow) ". Le manifestazioni cliniche più evidenti della patologia includono posture anormali, gesti lenti e non molto fluidi (a volte spasmodici e caratterizzati da tremore), maggiore resistenza al movimento passivo, attivazione di muscoli non direttamente coinvolti nel movimento (TD Sanger et al., 2003) (TD Sanger et al., 2010). In termini di aree cerebrali associate, va sottolineato che questa patologia è spesso correlata a lesioni a livello dei gangli della base, in particolare putamen e globo pallido; tuttavia, è anche possibile riscontrare un coinvolgimento di altre aree, quali il cervelletto e il tronco cerebrale (V. K. Neychev et al., 2011) (A. Tewari et al., 2017). È stato osservato che i deficit sensoriali sono correlati a un deficit nell'integrazione di input afferenti a livello centrale: ciò influenza negativamente la generazione della corrispondente output motorio (A. Berardelli et al., 1998) (A. Quartarone e M. Hallett, 2013). In termini di eziologia, la distonia viene classificata come primaria quando è la caratteristica più importante di una malattia genetica idiopatica, come secondaria quando deriva da cause pregresse, ad esempio paralisi cerebrale (CP). Alcune evidenze provano che i soggetti affetti da distonia primaria non mostrano deficit sensoriali; i soggetti affetti da distonia secondaria, invece, sono caratterizzati da anomalie sensoriali che comportano alterazioni motorie (F. Molloy et al., 2003) (T. D. Sanger e S. N. Kukke, 2007). Tali deficit sensoriali portano ad una mancanza di miglioramento nelle complesse abilità motorie nonostante la pratica ripetuta. Le principali modalità di trattamento spaziano dalle terapie fisica ed occupazionale all'uso di vari tipi di farmaci (tossina botulinica, farmaci anticolinergici, farmaci dopaminergici, ...) e al trattamento chirurgico (Deep Brain Stimulation, DBS): si tratta di approcci basati esclusivamente sulla sintomatologia presentata dal soggetto e, spesso, solo parzialmente efficaci (A. Albanese et al., 2006) (J. Jankovic, 2013) (M. Bertucco e TD Sanger, 2015). In questo contesto, sono necessarie soluzioni di trattamento innovative e non invasive che agiscano ad un livello superiore rispetto al livello sintomatico, con l'obiettivo di favorire il corretto apprendimento motorio del soggetto. In generale, il biofeedback (BF) è una tecnica che consiste nel fornire al soggetto una maggiore informazione di un processo fisiologico, consentendogli di aumentare la consapevolezza del processo stesso e di acquisire il controllo volontario su di esso (J. V. Basmajian, 1982). Al soggetto vengono fornite informazioni sullo stato di una variabile fisiologica (EMG, EEG, forza, posizione, ...) attraverso canali sensoriali di diverso tipo (visivo, uditivo, tattile, ...). I primi studi sull'uso di trattamenti basati sul BF per il miglioramento dell'apprendimento motorio hanno dimostrato che l'individuo può acquisire il controllo ad alti livelli sul sistema neuromuscolare, gestendo anche il singolo muscolo (J.Ve Basmajian, 1963). I meccanismi neurologici alla base dell'allenamento con il BF nei soggetti distonici non sono ancora del tutto chiari. Si ritiene che l'uso del BF possa migliorare la plasticità neuronale, aumentando, grazie alle informazioni ausiliarie fornite al soggetto, l'affidabilità del segnale afferente, con l'effetto di rinforzare i circuiti neurali. Gli strumenti riabilitativi che forniscono un feedback sensoriale aumentato, quindi, sono potenzialmente utilizzabili per migliorare il controllo motorio in soggetti con disturbi del movimento. Nei pazienti distonici, l'efficacia dell'uso di tale strategia è stata dimostrata in diverse occasioni (S. J. Young et al., 2011) (C. Casellato et al., 2013). Tuttavia, per essere efficace, una terapia basata sull’uso del BF dovrebbe rieducare il controllo motorio durante movimenti dinamici, orientati allo scopo (H. Huang et al., 2006): per soddisfare questa necessità, Sanger e Bloom hanno sviluppato (2010), presso il SangerLab (University of Southern California (USC), Los Angeles, CA), un dispositivo di BF portatile e silenzioso che fornisce al soggetto una vibrazione meccanica proporzionale all'attivazione del muscolo su cui è posizionato e che può quindi essere facilmente utilizzato durante le attività quotidiane. Metodi Il protocollo di acquisizione è finalizzato a testare l'efficacia del dispositivo di BF nel breve periodo (una settimana) durante l’apprendimento di specifici esercizi motori da eseguire in un ambiente controllato. Consiste in due settimane di acquisizione dati, con un'interruzione di almeno un mese. Al soggetto viene chiesto di eseguire due esercizi motori ben definiti, il cosiddetto Figure 8 Task e lo Spoon Task, che simulano, rispettivamente, l'atto della scrittura e l'atto del cibarsi. Durante questo lavoro di tesi ho partecipato a tutta la campagna sperimentale, ma l'analisi dei dati si è concentrata sulla valutazione del solo Spoon Task. Al soggetto è chiesto di trasportare una biglia in un cucchiaio avanti e indietro rispetto a un blocco di legno senza far cadere la pallina. La profondità del cucchiaio definisce la difficoltà dell'esercizio e la richiesta è di eseguire 10 movimenti (avanti e indietro) alla massima velocità possibile. Ogni settimana consiste in due sessioni di test (giorno 1 e giorno 5, che sono rispettivamente la baseline e la fine dell'allenamento), in cui il compito motorio viene eseguito con diversi gradi di difficoltà, e tre sessioni di allenamento, i giorni intermedi, durante i quali il task viene eseguito con un solo livello di difficoltà (il livello medio). Il soggetto indossa il dispositivo di BF (posizionato sul braccio dominante, se affetto) solo durante le sessioni di allenamento di una delle due settimane e l'ordine di utilizzo del dispositivo è randomizzato tra le due settimane. I soggetti sani hanno seguito un protocollo semplificato e più praticabile. La loro presenza in laboratorio era richiesta solo durante i giorni di test e si allenavano a casa nei giorni intermedi. Il protocollo è stato approvato da Besta e Medea. Tuttavia una distinzione deve essere fatta. A differenza del protocollo PoliMi, il protocollo Medea prevede quattro settimane consecutive di acquisizione. Nella prima e nella terza settimana il soggetto svolge il Figure 8 Task, durante la seconda e la quarta settimana il soggetto svolge lo Spoon Task. L'analisi quantitativa degli esercizi motori viene eseguita attraverso la valutazione: • dell'attività elettromiografica di otto muscoli dell'arto superiore testato: Flessore Ulnare del Carpo (FCU), Estensore Radiale del Carpo (ECR), Bicipite (BIC), Tricipite (TRIC), Deltoide Anteriore (AD), Deltoide Laterale (LD), Deltoide Posteriore (PD), Sopraspinato (SS); • della cinematica dell'articolazione della spalla, dell'articolazione del gomito, dell'articolazione del polso e del cucchiaio. La Figura 1 mostra il setup sperimentale. Figura 1: Setup sperimentale L'analisi dei dati è consistita in una fase iniziale di pre-processing sia del segnale EMG sia del segnale cinematico, con l’obiettivo di ricavare l'inviluppo del segnale EMG per ciascuno degli 8 muscoli acquisiti e la cinematica riproiettata sul piano delle componenti principali di ciascuna articolazione. Quindi, la valutazione delle prestazioni del soggetto si è basata su indici quantitativi che descrivono vari aspetti del gesto motorio: 1) indici temporali: • tempo di movimento; • Indice di Prestazione. Derivare un trade-off velocità-accuratezza (SATO) è importante per la valutazione delle anomalie motorie nella distonia (F. Lunardini et al., 2015). Esso è stato esaminato cambiando le dimensioni del cucchiaio per creare diversi livelli di difficoltà, e l'Indice di Prestazione è stato calcolato. Esso riflette quindi l'efficienza del sistema nervoso/motorio nell'affrontare compiti di difficoltà crescente; 2) descrittori cinematici dell'output finale e dei movimenti dell'intera catena cinematica: • velocità di movimento; • fluidità del movimento in tre diverse formulazioni, ovvero il numero di picchi nel profilo di velocità, il rapporto tra la velocità media e la velocità massima (H. Vikne et al., 2013), e una forma adimensionale del jerk, il Normalized Mean Squared Jerk (N. Hogan, D. Sternad, 2009); • linearità del movimento, una misura indiretta di precisione, calcolata come rapporto tra l'ampiezza della traiettoria generata dal soggetto e la distanza lineare tra i punti iniziale e finale del movimento; • ripetibilità del movimento, che fornisce un'indicazione circa la capacità del soggetto di eseguire movimenti ripetibili durante le prove; • range di movimento dell'articolazione del gomito. Fornisce un'indicazione dell'estensione e della flessione dell'arto superiore del soggetto durante il movimento; • spostamento dell'articolazione della spalla, che misura il movimento della stessa nella direzione antero-posteriore; 3) indici dell'attività elettromiografica di tre coppie di muscoli agonisti / antagonisti: Flessore Ulnare del Carpo ed Estensore Radiale del Carpo, Bicipite e Tricipite, Deltoide anteriore e Deltoide posteriore. Al fine di valutare se l'uso del dispositivo di biofeedback, rispetto alla sola pratica senza ulteriori feedback strumentali, avesse impattato sugli indici poco fa descritti, è stata condotta un'analisi statistica: un'analisi soggetto per soggetto è stata effettuata per valutare gli effetti di apprendimento tra il giorno 1 e il giorno 5 delle due settimane di trattamento. La variabilità di ciascun parametro è stata valutata mediante un’analisi della varianza a 2 vie (ANOVA), scegliendo la condizione (BF o noBF) e il giorno (giorno 1 e giorno 5) come fattori fissi. E’ anche stato valutato l’effetto interazione condizione-giorno. Un'analisi statistica di gruppo non è stata invece eseguita a causa dell'eterogeneità dei soggetti in termini di età e livello di compromissione. Risultati e discussione Sono stati reclutati 15 pazienti in totale: 9 soggetti distonici primari (5 maschi e 4 femmine dai 7 ai 19 anni) e 6 soggetti distonici secondari (4 maschi e 2 femmine dai 6 ai 16 anni). Un gruppo di 12 soggetti sani di pari età è stato reclutato e coinvolto nel protocollo: 5 ragazzi e 7 ragazze dai 6 ai 19 anni. Tutti i partecipanti hanno dato il consenso informato per la partecipazione. In caso di minori, i genitori sono stati invitati a firmare il consenso informato e l'autorizzazione per l'uso di informazioni, video e immagini sanitarie protette. La figura 2 mostra i risultati dell'analisi in termini di percentuale dei soggetti sani, primari e secondari che hanno mostrato apprendimento con la pratica, indipendentemente dall'uso del dispositivo, e un maggior apprendimento con l’utilizzo dispositivo. Figura 2: Apprendimento / apprendimento con interazione. MT = Tempo di Movimento, PEAKS = picchi di velocità, SMOOTHNESS = fluidità, ML = Linearità del Movimento, ROM = ROM del gomito, ST = Spostamento della Spalla, CC FCU / ECR = co-contrazione dei muscoli FCU / ECR, CC BIC / TRIC = Co-contrazione dei muscoli BIC / TRIC, CC AD / PD = co-contrazione dei muscoli AD / PD. DAY indica l'effetto giorno, INTER indica, invece, l'effetto di interazione. In figura sono riportati 9 pannelli, relativi a ciascun parametro su cui è stata eseguita l'analisi statistica. Da sinistra a destra, viene mostrata la tendenza dei soggetti sani (in blu), dei primari (in nero), dei secondari (in rosso). Per ogni gruppo, la barra di sinistra mostra la percentuale di soggetti che hanno appreso il task con la pratica, la barra di destra mostra la percentuale di soggetti che hanno mostrato un'interazione significativa a beneficio della condizione BF. 2 dei 12 soggetti sani, 4 dei 9 soggetti primari e 1 dei 6 soggetti distonici secondari hanno ridotto il tempo di movimento con la pratica. 5 dei 12 soggetti sani, 2 dei 9 soggetti primari e 1 dei 6 soggetti distonici secondari hanno mostrato un maggior apprendimento grazie all’uso del dispositivo di BF. 3 dei 12 soggetti sani, 3 dei 9 soggetti primari e 1 dei 6 soggetti distonici secondari hanno ridotto il numero dei picchi di velocità con la pratica. 3 dei 12 soggetti sani, 2 dei 9 soggetti principali e 1 dei 6 soggetti distonici secondari hanno mostrato un maggior apprendimento grazie all’uso del dispositivo di BF. Nessuno dei 12 soggetti sani, 1 dei 9 soggetti principali e nessuno dei 6 soggetti distonici secondari ha aumentato la fluidità del movimento con la pratica. 1 dei 12 soggetti sani, 1 dei 9 soggetti primari e 1 dei 6 soggetti distonici secondari ha mostrato un maggior apprendimento grazie all’uso del dispositivo di BF. 2 dei 12 soggetti sani, 3 dei 9 soggetti principali e 1 dei 6 soggetti distonici secondari hanno ridotto la linearità del movimento con la pratica. 2 dei 12 soggetti sani, 1 dei 9 soggetti primari e 1 dei 6 soggetti distonici secondari hanno mostrato un maggior miglioramento grazie all’uso del dispositivo di BF. Per quanto riguarda il ROM del gomito e lo spostamento della spalla, è stata eseguita un'analisi completa su 5 dei 6 soggetti secondari, poiché mancavano dati cinematici relativi alle coordinate spaziali della spalla del secondo giorno di test della seconda settimana per un soggetto distonico secondario. 4 dei 12 soggetti sani, nessuno dei 9 soggetti principali e 2 dei 5 soggetti distonici secondari hanno aumentato la ROM del gomito con la pratica. 4 dei 12 soggetti sani, 3 dei 9 soggetti primari e 1 dei 5 soggetti distonici secondari hanno mostrato un un maggior apprendimento grazie all’uso del dispositivo di BF. 3 dei 12 soggetti sani, 1 dei 9 soggetti primari e 1 dei 5 soggetti distonici secondari hanno ridotto lo spostamento della spalla durante il movimento con la pratica. 4 dei 12 soggetti sani, 3 dei 9 soggetti primari e 2 dei 5 soggetti distonici secondari hanno mostrato un maggior apprendimento grazie all’uso del dispositivo di BF. Per quanto riguarda i parametri EMG, è stata eseguita un'analisi completa su 4 dei 6 soggetti secondari, poiché mancavano tutti i dati EMG di due soggetti: in un soggetto quelli del secondo giorno di test della prima settimana, nell'altro soggetto distonico secondario della prima settimana di acquisizioni. 1 dei 12 soggetti sani, nessuno dei 9 soggetti principali e nessuno dei 4 soggetti distonici secondari ha ridotto la co-contrazione dei muscoli FCU/ECR con la pratica. 3 dei 12 soggetti sani, 2 dei 9 soggetti primari e 1 dei 4 soggetti distonici secondari hanno mostrato un maggior apprendimento grazie all’uso del dispositivo di BF. 4 dei 12 soggetti sani, 2 dei 9 soggetti primari e nessuno dei 4 i soggetti distonici secondari riducevano la co-contrazione dei muscoli BIC/TRIC con la pratica. 3 dei 12 soggetti sani, 3 dei 9 soggetti primari e 2 dei 4 soggetti distonici secondari hanno mostrato un maggior apprendimento grazie all’uso del dispositivo di BF. 1 dei 12 soggetti sani, 1 dei 9 soggetti primari e nessuno dei 4 soggetti distonici secondari ha ridotto la co-contrazione AD / PD con la pratica. Nessuno dei 12 soggetti sani, 4 dei 9 soggetti primari e 2 dei 4 soggetti distonici secondari hanno mostrato un maggior apprendimento grazie all’uso del dispositivo di BF. Per quanto riguarda il tempo e la velocità di esecuzione, la pratica con il dispositivo ha influenzato il comportamento dei soggetti primari e dei pazienti secondari quasi nello stesso modo, infatti la percentuale dei soggetti primari per i quali è stato trovato un significativo effetto di interazione è circa equivalente alla percentuale dei primari. Si può affermare che per il tempo di movimento ci sia stata una tendenza all'apprendimento principalmente nei soggetti distonici primari. Tuttavia, considerando anche il gruppo dei soggetti sani, non è risultato un così evidente apprendimento: meno della metà dei soggetti reclutati ha mostrato un effetto giorno significativo. La tendenza al miglioramento osservata per l'IP è uguale a quella del tempo di movimento. Pertanto, si può affermare che il dispositivo abbia influenzato le capacità di elaborazione delle informazioni dei pazienti primari e dei secondari allo stesso modo. Anche guardando alla fluidità del movimento non ci sono prove che confermino come l'esercizio eseguito con il BF possa aver prodotto risultati migliori sul gruppo dei soggetti secondari. L'effetto del dispositivo sull'apprendimento del compito è stato paragonabile tra i pazienti; l'apprendimento indotto dalla pratica è stato maggiore per i pazienti primari, ma, nel complesso, esso non è stato così evidente, dal momento che solo uno tra tutti i soggetti reclutati ha dimostrato di apprendere il task motorio con la pratica. L'analisi sulla linearità del movimento conferma quanto è stato detto sul tempo di movimento e sulla fluidità. Considerando il ROM del gomito e lo spostamento della spalla, i risultati mostrano qualcosa di diverso rispetto ai parametri cinematici dell'output finale. Per quanto riguarda il ROM del gomito, non è stato riscontato alcun apprendimento nei pazienti distonici primari, che, invece, hanno migliorato le loro prestazioni con l’uso del BF; considerando i secondari, una percentuale maggiore ha imparato con la sola pratica. Per quanto riguarda lo spostamento della spalla, il comportamento dei tre gruppi è stato molto simile: sia i soggetti sani sia i pazienti (distonici primari e distonici secondari) hanno avuto un apprendimento migliore quando è stato utilizzato il dispositivo di BF. Tuttavia, è probabile che questi due parametri non identifichino così bene il comportamento dei soggetti, che hanno adottato strategie diverse che si sono tradotte in una grande variabilità inter-campione nei movimenti dell'intera catena cinematica: poiché la posizione del blocco di legno sul tavolo di supporto non era fissa, alcuni soggetti riuscivano a eseguire l'esercizio rimanendo fermamente sulla sedia e posizionandosi più vicino al blocco, altri soggetti, invece, preferivano stare meno fermi sulla sedia e accompagnavano il movimento con il busto. I valori iniziali di ripetibilità del movimento sono più o meno uguali tra distonici primari e distonici secondari. Non c’è stata una chiara tendenza verso l'apprendimento nè tra i soggetti primari né tra i sani. Probabilmente è necessario considerare che le possibilità di miglioramento fossero già limitate in partenza, date le buone prestazioni al primo giorno di test. Lo stesso tipo di analisi può essere condotto sui secondari, che non hanno rivelato se il dispositivo fosse effettivamente efficace. Comunque, va considerato che solo in un caso (S001BoLu) è stato riscontrato un apprendimento isolato nella settimana di non utilizzo del dispositivo. Anche guardando ai parametri elettromiografici, non vi è una chiara tendenza a favore delle ipotesi iniziali: infatti, per quanto riguarda sia la co-contrazione dei muscoli FCU/ECR che la co-contrazione dei muscoli AD/PD, è stato evidente un maggior apprendimento quando il BF è stato utilizzato sia per i soggetti primari sia per soggetti secondari; l'apprendimento indotto dalla pratica è stato nullo per i pazienti primari e visibile solo per un soggetto secondario. Concentrandosi sulla co-contrazione dei muscoli primariamente coinvolti nella prono-supinazione del polso, il comportamento dei tre gruppi è stato simile: i soggetti sani e i pazienti (soggetti primari e soggetti secondari) hanno migliorato le loro prestazioni quando il feedback sensoriale è stato fornito, mentre l'apprendimento con la pratica è stato mostrato da un solo soggetto (un soggetto sano). La tendenza dei soggetti sani è stata quella di aumentare la co-contrazione dei muscoli FCU/ECR con la pratica. Non è sbagliato ipotizzare che solo una maggiore attivazione simultanea di questi muscoli si rifletta in una maggiore stabilità dell'articolazione del polso, e quindi un maggiore controllo sul cucchiaio. Questo spiega perché, se il risultato atteso era una diminuzione dei livelli di co-contrazione, sono stati trovati bassi livelli di apprendimento. Infine, considerando la co-contrazione dei muscoli BIC/TRIC, i soggetti secondari non hanno rivelato maggior apprendimento durante l'utilizzo del BF. Piuttosto, la pratica ha indotto un apprendimento del compito, ma questo non è stato influenzato dall'uso del dispositivo; inoltre, per i soggetti primari, il dispositivo di BF ha influenzato l'apprendimento del compito più di quanto abbia fatto la pratica. In generale, non si può affermare che l’effetto correlato all'uso del dispositivo sia più pronunciato nel gruppo di distonici secondari: è stato riscontrato un apprendimento accelerato dall’uso del dispositivo sia nei soggetti primari sia nei pazienti secondari. Anche per i soggetti sani il BF ha accelerato i processi di apprendimento. Ciò è stato particolarmente evidente considerando i parametri temporali e la co-contrazione dei muscoli FCU/ECR. Per quanto riguarda i parametri temporali e i parametri relativi alla cinematica dell'output finale, l'apprendimento nei distonici primari non è stato significativamente influenzato dall'uso del dispositivo, ma questa tendenza non è stata confermata considerando i parametri legati all'intera catena cinematica (ROM e ST) e i parametri EMG, per i quali il dispositivo ha accelerato l'apprendimento. Quindi, i risultati non ci permettono di supportare le ipotesi su cui si basa lo studio: anche individui affetti da distonia secondaria, che sono caratterizzati da marcati deficit sensoriali che impediscono il loro miglioramento nell'esecuzione di compiti motori complessi (nonostante l'allenamento ripetuto), sembrano non essere così favoriti dall’uso di tale strategia; d'altra parte, per i soggetti affetti da distonia primaria e per i sani, in cui non ci sono deficit sensoriali e i meccanismi di apprendimento fisiologico legati alla formazione di un particolare esercizio motorio sono intatti, l'uso di una strategia di biofeedback sembra avere un ruolo nel miglioramento delle prestazioni . Tuttavia, non siamo neanche nella posizione di rifiutare le ipotesi iniziali. Alcune limitazioni devono ancora essere superate. La prima limitazione dello studio risiede nel fatto che attualmente esso non prende in considerazione variabili cognitive che potrebbero portare a disturbi comportamentali e che potrebbero impedire ai secondari di accedere a risorse sufficienti per un apprendimento corretto. Una limitazione deriva dalla mancanza di alcuni dati fondamentali per l'analisi di alcuni pazienti secondari. Inoltre, la mancanza di un'analisi statistica di gruppo e i risultati molto variegati non consentono conclusioni definitive. Questo problema si aggiunge al numero limitato di soggetti, che rende la potenza del campione limitata e il confronto non sempre statisticamente robusto. Tuttavia, lo studio clinico multicentrico in corso supererà questa limitazione ed è previsto il reclutamento di ulteriori soggetti distonici. Inoltre, sarà studiato l'effetto a lungo termine dell'uso del dispositivo di biofeedback durante le attività quotidiane (indossando il dispositivo almeno 5 ore al giorno per un mese). I futuri sviluppi di questo lavoro potrebbero essere: • definizione di una serie di criteri per guidare l'operatore nella scelta dei muscoli più adatti su cui collocare il biofeedback; questi potrebbero essere selezionati, ad esempio, dopo una valutazione clinica e quantitativa che indica chiaramente i più compromessi; • test di utilizzo simultaneo di più modalità di feedback; • test degli effetti di diversi canali sensoriali attraverso i quali fornire il feedback.
Effect of an EMG-based vibro-tactile biofeedback training on motor learning in children and adolescents with dystonia
SANTAMBROGIO, ALESSIO
2017/2018
Abstract
Abstract Introduction Within the field of neuromotor rehabilitation, the aim of this thesis was to evaluate the efficacy of a vibro-tactile biofeedback device based on electromyography, in improving motor learing in children and adolescents with dystonia. In 2013, an international committee formed by different personalities with years of experience in the field of dystonia met to clarify terminological and classification aspects related to this pathology, which was defined as follows (A. Albanese et al., 2013): ‘’Dystonia is a movement disorder characterized by sustained or intermittent muscle contractions that cause abnormal, often repetitive movements. Dystonic movements are typically stereotyped, torsional and can be characterized by tremor. Dystonia is often initiated or worsened by voluntary actions and is characterized by the activation of muscles that are not strictly functional to the motor gesture (overflow)’’ The most evident clinical manifestations of the pathology include abnormal postures, slow and not very fluid gestures (sometimes spasmodic and characterized by tremor), increased resistance to passive movement, activations of muscles not directly involved in the ongoing movement (T. D. Sanger et al., 2003) (T. D. Sanger et al., 2010). In terms of brain areas associated with the pathology, it should be emphasized that this is often correlated with lesions at the level of the basal ganglia, in particular putamen and globus pallidus; however, it is also possible to involve other, areas such as the cerebellum and the brainstem (V. K. Neychev et al., 2011) (A. Tewari et al., 2017). Sensory deficits have also been observed to be related to a deficit in the integration of afferent inputs at the central level: this negatively affects the generation of the corresponding motor output (A. Berardelli et al., 1998) (A. Quartarone and M. Hallett , 2013). In terms of etiology, dystonia is classified as primary when it is the most important feature of an idiopathic or an identified genetic disorder, while secondary dystonias are symptomatic disorders arising from another underlying disease, such as cerebral palsy (CP) or acquired brain injury. The physiopathology of dystonia is very variegate; there are evidences that subjects with primary dystonia do not show sensory deficits, whereas, subjects with secondary dystonia are characterized by sensory abnormalities entailing to motor impairments (F. M. Molloy et al., 2003) (T. D. Sanger and S. N. Kukke, 2007). Sensory deficits lead to a lack of improvement in complex motor skills despite repeated practice. The main treatment modalities range from physical and occupational therapy to the use of various types of drugs (botulinum toxin, anticholinergic drugs, dopaminergic drugs, ...) and finally to surgery (Deep Brain Stimulation, DBS): these are approaches exclusively based on the symptomatology presented by the subject and, often, only partially effective in controlling it (A. Albanese et al., 2006) (J. Jankovic, 2013) (M. Bertucco and T. D. Sanger, 2015). In this context, innovative and non-invasive treatment solutions that work at a higher level than the symptomatic level are needed, with the aim of favoring the correct motor learning of the subject. In general, biofeedback (BF) is a technique that consists in providing the subject with increased information of a physiological process, allowing him/her to increase awareness of the process itself and to acquire voluntary control over it (J. V. Basmajian, 1982). The subject is provided with information on the state of a physiological variable (EMG, EEG, strength, position, ...) through sensorial channels of different types (visual, auditory, tactile, ...). The first studies on the use of BF-based treatments for the improvement of motor learning have shown that the individual can acquire control at high levels on the neuromuscular system, even managing the single muscle (J. V. Basmajian, 1963). The neurological mechanisms underlying the BF training in dystonic subjects are not yet completely clear; it is however known that such patients are potentially characterized by inefficiencies in terms of sensory information: a hypothesis is that the use of BF can improve neural plasticity by increasing, thanks to the auxiliary information provided to the subject, the reliability of the incoming signal to sensorimotor areas; the effect is that of reinforcing the correct neural circuits. Rehabilitative tools that provide increased feedback, therefore, are potentially usable for improving motor control in subjects with movement disorders. In dystonic patients, the efficacy of the use of such a strategy has been demonstrated on several occasions (S. J. Young et al., 2011) (C. Casellato et al., 2013). However, to be effective a BF therapy should re-educate motor control during dynamic, functionally purpose-oriented movements (H. Huang et al., 2006): to satisfy this demand, Sanger and Bloom developed (2010), at SangerLab (University of Southern California (USC), Los Angeles, CA), a portable and silent device that provides the subject with a mechanical vibration proportional to the activation of the muscle on which it is positioned and which can therefore be easily used during daily life activities. The present thesis work is part of a multi-center cross over study that involves, besides the Politecnico of Milano, the Carlo Besta Neurological Institute, Milano (MI), the Scientific Institute Eugenio Medea, Bosisio Parini (LC), and the Children's Hospital, Los Angeles (CA). The research project has been approved and is supported by the United States Department of Health (NIH, National Institutes of Health). The coordinator of the entire study is Dr. T. D. Sanger, professor at the University of Southern California (USC, Los Angeles) and neurologist at the Children's Hospital of Los Angeles, who designed and built a battery-powered wearable electromyographic (EMG) feedback device. The objective of the present thesis work is to test the mechanism of action of this EMG-based vibro-tactile biofeedback device and to quantify the effect of increased sensory information during motor learning in children affected by primary and secondary dystonia while performing specific exercises in in the laboratory/clinic, and to compare the effect it produces on patients with its influence on healthy subjects. Subjects suffering from primary dystonia are recruited at Carlo Besta Neurological Institute and acquired at Laboratory of Neuroengineering and Medical Robotics (NearLab) of Politecnico di Milano, while the secondary patients are recruited and acquired at Scientific Institute Eugenio Medea. The aim is to prove that biofeedback improves motor function in children with secondary dystonia. On the other hand, children with primary dystonia (without sensory deficits) are analyzed as a control to show that the improvement is not due to a direct interaction between biofeedback and the motor abnormality in dystonia, and we compare to healthy control subjects to quantify the degree of normalization of function that can be achieved. Hence, the working hypothesis is that a positive effect should be recorded only on the secondary population, characterized by marked sensory deficits. In the case of subjects affected by primary dystonia, the use of a biofeedback strategy should have no clear role in improving performance. At least, the behavior of the tested primaries should be approximately consistent with the behavior of the healthy participants. Methods The acquisition protocol is aimed at testing the effectiveness of the BF device in the short term (1 week) in learning specific motor exercises to be performed in a controlled environment. It consists of two weeks of data acquisition, with a break of at least one month in between. The subject is asked to perform two defined motor exercises, the so-called Figure 8 Task and the Spoon Task, which simulate, respectively, the act of writing and the act of self-feeding. During this thesis work I participated in the entire experimental campaign, but the data analysis was concentrated on the evaluation of the Spoon Task. The subject is required to carry a marble in a spoon back and forth between two targets without dropping the marble. The depth of the spoon defines the difficulty of the excercise and the request is to perform 10 movements (forward and backward) at the maximum possible speed. Each week consists of two Testing sessions (day 1 and day5, which are, respectively, the baseline and the end of the training), in which the motor task are always performed with different degrees of difficulty, and three Training sessions, the intermediate days, during which the task is performed with only one difficulty level (the medium level). The subject wears the BF device (positioned on the dominant arm) only during the Training sessions of one of the two weeks and the order of use of the BF device is randomized between the two weeks. Healthy subjects followed a simplified and more practicable protocol. Their presence in the laboratory was required only during Testing days and they trained at home in the intermediate days. The protocol was approved by both Besta and Medea. However a distinction must be made. Unlike the PoliMi protocol, the Medea protocol provides four consecutive weeks of acquisition. In the first and third weeks the subject plays Figure 8 Task, during the second and fourth weeks the subject plays Spoon Task. The quantitative analysis of the motor exercises is performed through the evaluation of: • the electromyographic activity of eight muscles of the upper limb tested: Flexor Carpi Ulnaris (FCU), Extensor Carpi Radialis (ECR), Bicep (BIC), Tricep (TRIC), Anterior Deltoid (AD), Lateral Deltoid (LD), Posterior Deltoid (PD), Supraspinatus (SS); • the kinematics of the shoulder joint, the elbow joint, the wrist joint and the spoon. Figure 1 shows the experimental setup. Figure 1: Experimental setup The data analysis consisted of an initial pre-processing phase of the EMG signal and of the kinematic signal, leading to obtain the EMG signal envelope for each of the 8 acquired muscles and the reprojected kinematics on the main components plane of each joint. Then, the evaluation of the subject motor performance was based on quantitative indices that describe various aspects of the task: 1) temporal indices: • movement time; • index of performance. Thus, deriving a speed-accuracy trade-off (SATO) is an important metric for assessing motor impairments in dystonia (F. Lunardini et al., 2015). The SATO was examined by changing the spoon size to create different difficulty levels, and the Index of Performance was computed, which reflects the efficiency of the nervous / motor system in facing tasks of increasing difficulty; 2) kinematic descriptors of the final output and of the movements of the entire kinematic chain: • movement speed; • smoothness of the movement in three different formulations, i.e. number of peaks in the velocity profile, the ratio between the mean velocity and the maximum velocity (H. Vikne et al., 2013), and a dimensionless form of the jerk, the Normalized Mean Squared Jerk (N. Hogan, D. Sternad, 2009); • movement linearity, an indirect measure of accuracy, calculated as the ratio between the amplitude of the trajectory generated by the subject and the linear distance between the starting and the ending point of the movement; • movement repeatability, which provides an indication of the subject's ability to perform repeatable movements throughout the trials; • range of motion of the elbow joint. It gives an indication of the subject's upper limb flexion-extension during the movement; • displacement of the shoulder joint, which measures the movement of the shoulder joini in the anterior-posterior direction; 3) indices of the electromyographic activity of three couples of agonist / antagonist muscles: Flexor Carpi Ulnaris and Extensor Carpi Ulnaris, Bicep and Tricep, Anterior Deltoid and Posterior Deltoid. In order to observe if the use of the biofeedback device during the Training impacted on these indices, compared to the simple practice without additional instrumental feedback, a statistical analysis was conducted: a subject-by-subject analysis was carried out to evaluate the effects of learning between day 1 and day 5 of the two weeks of treatment. The variability of each parameter was evaluated through a 2-way analysis of variance (ANOVA), including the condition (BF or noBF) and the day (day 1 and day 5) as fixed factors and the condition by day as interaction effect. A group-based statistical analysis was not performed due to the heterogeneity of the subjects in terms of age and level of impairment. Results and discussion A total of 15 patients were recruited: 9 primary dystonic subjects (5 boys and 4 girls from 7 to 19 years old); 6 secondary dystonic subjects (4 boys and 2 girls from 6 to 16 years old). A group of 12 age-matched healthy subjects were recruited and involved in the protocol: 5 boys and 7 girls from 6 to 19 years old. All participants gave informed written consent for participation. In case of minors, parents were asked to sign the informed consent and the authorization for use of protected health information, videos and images. Figure 2 shows the results of the analysis in terms of the percentage of the healthy, the primary and secondary subjects that showed learning characteristics with the practice, regardless of the use of the device, and better improvement when the BF device was used. Figure 2: Learning / learning with interaction. MT=Movement Time, PEAKS=speed peaks, SMOOTHNESS=smoothness, ML=Movement Linearity, ROM=elbow range of motion, ST=shoulder displacement, CC FCU / ECR=FCU / ECR co-contraction, CC BIC / TRIC=BIC / TRIC co-contraction, CC AD / PD=AD / PD co-contraction. DAY is the day effect, INTER is the interaction effect. 9 panels are reported, related to the outcomes for which the statystical analysis was performed. From left to right, the trend of the healthy (in blue), the primary (in black), the secondary (in red) subjects is shown. For each group, the bar on the left shows the percentage of subjects who learned the task with practice, the bar on the right shows the percentage of the subjects that showed a significant interaction effect in favour of the BF. 2 of the 12 the healthy subjects, 4 of the 9 primary subjects and 1 of the 6 secondary dystonic subjects reduced the movement time with practice. 5 of the 12 healthy subjects, 2 of the 9 primary subjects and 1 of the 6 secondary dystonic subjects showed better improvement when the BF device was used. 3 of the 12 the healthy subjects, 3 of the 9 primary subjects and 1 of the 6 secondary dystonic subjects reduced the number of the speed peaks with practice. 3 of the 12 healthy subjects, 2 of the 9 primary subjects and 1 of the 6 secondary dystonic subjects showed better improvement when the BF device was used. None of the 12 the healthy subjects, 1 of the 9 primary subjects and none of the 6 secondary dystonic subjects increased the smoothness of the movement with practice. 1 of the 12 healthy subjects, 1 of the 9 primary subjects and 1 of the 6 secondary dystonic subjects showed better improvement when the BF device was used. 2 of the 12 the healthy subjects, 3 of the 9 primary subjects and 1 of the 6 secondary dystonic subjects reduced the movement linearity with practice. 2 of the 12 healthy subjects, 1 of the 9 primary subjects and 1 of the 6 secondary dystonic subjects showed better improvement when the BF device was used. For what concerns the elbow ROM, a complete analysis was performed on 5 of the 6 secondary subjects, since kinematic data concerning the spatial coordinates of the shoulder of the second testing day of the second week were missing for one secondary dystonic subject. 4 of the 12 the healthy subjects, none of the 9 primary subjects and 2 of the 5 secondary dystonic subjects increased the elbow ROM with practice. 4 of the 12 healthy subjects, 3 of the 9 primary subjects and 1 of the 5 secondary dystonic subjects showed better improvement when the BF device was used. Even considering the shoulder displacement, a complete analysis was performed on 5 of the 6 secondary subjects, since kinematic data concerning the spatial coordinates of the shoulder of the second testing day of the second week were missing for one secondary dystonic subject. 3 of the 12 the healthy subjects, 1 of the 9 primary subjects and 1 of the 5 secondary dystonic subjects reduced the shoulder displacement during the movement with practice. 4 of the 12 healthy subjects, 3 of the 9 primary subjects and 2 of the 5 secondary dystonic subjects showed better improvement when the BF device was used. For what concerns EMG parameters, a complete analysiy was performed on 4 of the 6 secondary subjects, because all the EMG data were missing, in one subject relating to the second testing day of the first week, in the other secondary dystonic subject related to the first week. 1 of the 12 the healthy subjects, none of the 9 primary subjects and none of the 4 secondary dystonic subjects reduced the FCU/ECR co-contraction with practice. 3 of the 12 healthy subjects, 2 of the 9 primary subjects and 1 of the 4 secondary dystonic subjects showed better improvement when the BF device was used.4 of the 12 the healthy subjects, 2 of the 9 primary subjects and none of the 4 secondary dystonic subjects reduced the BIC/TRIC co-contraction with practice. 3 of the 12 healthy subjects, 3 of the 9 primary subjects and 2 of the 4 secondary dystonic subjects showed better improvement when the BF device was used. 1 of the 12 the healthy subjects, 1 of the 9 primary subjects and none of the 4 secondary dystonic subjects reduced the AD/PD co-contraction with practice. None of the 12 healthy subjects, 4 of the 9 primary subjects and 2 of the 4 secondary dystonic subjects showed better improvement when the BF device was used. For what concerns time and speed of execution, the practice with the BF device influenced the behavior of the primary subjects and the secondary patients almost in the same way, in fact the percentage of the primary subjects for whom a significant interaction effect was found is about the same with respect to the percentage of the secondary dystonic. It can be stated that for the movement time we see learning trend mainly for the primary dystonic subjects. However, considering also the group of healthy subjects, a great learning effect is not visible: less than half of people had a significant day effect. The improvement trend observed looking at the IP is equal to that of the movement time. Thus, it can be stated that the device affected the processing capabilities of the primary patients and the secondary more or less in the same way. Even looking at the smoothness of the movement there’s no evidence to confirm that the exercise performed with the BF produced better results on the secondary group: the effect of the device on the task learning was comparable between patients; learning induced by the practice was higher for the primary patients, but, overall, learning was not so evident, since only one subjects over all the recruited subjects showed learning with the practice. The analysis on the movement linearity confirms what has been said about the movement time and the smoothness. Considering the elbow ROM and the displacement of the shoulder, the results show something different with respect to the final output kinematic parameters. For what concerns the elbow ROM, there was no learning for the primary dystonic patients, who, instead, improved their performance when the BF was used, while, for the secondary patients, a greater percentage of them learned the task with the practice. For what concerns the shoulder displacement, the behavior of the three groups was very similar: both the healthy subjects and the patients (both primary dystonic and secondary dystonic) had better improvement when the BF device was used. However, it is likely that these two parameters do not identify so well the behavior of the subjects, who adopted different strategies that translated into a great inter-sample variability in the movements of the entire kinematic chain: since the position of the wooden block on the support table was not fixed, there were subjects who managed to perform the exercise remaining firmly on the chair and positioning themselves closer to the block, and there were those who, instead, preferred to stay less firmly on the chair and accompanied the movement with the trunk. Initial values of the movement repeatability were more or less equal between primary distonic subjects and secondary distonic patients. There was no clear trend towards learning among the primary subjects and the healthy subjects. It is probably necessary to consider that the range of improvement was limited by the already good performance on the first day of testing. The same type of analysis can be conducted on the secondaries, which did not clearly reveal whether the device was actually effective. However, we must consider that only in one case (S001BoLu) we there was an isolated learning in the week of non-use of the device. Even looking at the electromyographic parameters, there is no clear tendency in favour of the initial hypotheses: indeed, for what concerns both the FCU/ECR muscles co-contraction and the AD/PD muscles co-contraction, better improvement was evident when the BF device was used for both the primary subjects and the secondary subjects; learning induced by the practice was null for the primary patients and visible for only one secondary subjects. Focusing on the co-contraction of the muscles primarly involved in the prono-supination of the wrist, the behavior of the three groups was similar: the healthy subjects and the patients (primary subjects and secondary subjects) better improved their performance when the sensory feedback was provided, while learning with the practice was shown by only one subject (one healthy subject). The trend of the healthy subjects was to increase the FCU/ECR co-contraction with the practice. It is not wrong to assume that only a greater simultaneous activation of theese muscles reflects into greater stability of the wrist joint, and therefore greater control on the spoon. This explains why, if the expected result was a decrease in the co-contraction levels, low levels of learning were found. Finally, considering the BIC/TRIC co-contraction, the secondary subjects did not reveal better improvement while using the BF. Rather, the practice induced a learning of the trained task, but this was not influenced by the use of the device; moreover, for the primary subjects, the BF device influenced the task learning more than the practice did. In general, it can not be stated that an effect related to the use of the device is more pronounced in the group of secondary dystonic: better improvement with the device was recorded for both the primary subjects and the secondary subjects. Even for the healthy subjects the device accelerated learning processes. This was particularly evident considering the time-related parameters and the FCU/ECR co-contraction. For what concerns the temporal paramaters and the parameters related to the final output kinematics, learning in primary dystonic was not significantly stressed by the use of the device, but this trend was not visible considering the parameters linked to the entire kinematic chain (ROM and ST) and the EMG parameters, for which the device accelerated learning. Hence, the results do not allow us to support the hypotheses the study is based on: even individuals affected by secondary dystonia, which are characterized by marked sensory deficits that prevent their improvement in the performance of complex motor tasks (despite repeated training), seem not to be so much favored by using such a strategy; on the other hand, for the subjects affected by primary dystonia and for the healthy subjects, in which there are no sensory deficits and the physiological learning mechanisms linked to the training of a particular motor exercise are intact, the use of a biofeedback strategy seems to have a role in improving performance. However, we are not even in the position to refuse the initial hypotheses. Some limitations still need to be overcome. The first limitation of the study lies in the fact that it does not currently take into consideration cognitive variables that could lead to behavioral disorders and that can prevent the secondaries from accessing sufficient sources for a correct learning. A limitation derives from the lack of some fundamental data for the analysis of some secondary subjects. Furthermore, the lack of a group statistic analysis and the scattered results do not allow for definitive conclusions. This problem is added to the limited number of subjects, which make the sample power limited and the comparison not always statistically robust. However, the ongoing multi-center clinical trial will overcome this limitation and more dystonic subjects will be recruited. Furthermore, the long-term effect of the use of the biofeedback device during daily life activities (wearing the device at least 5 hours a day for one month) will be investigated. Future developments of this work could be: • definition of a set of criteria to guide the operator in choosing the most suitable muscles on which to place the biofeedback; these could be selected, for example, after a clinical and quantitative evaluation that clearly indicates the most compromised ones; • test of simultaneous use of multiple feedback modalities; • test of the effects different sensory channels through which providing the feedback.File | Dimensione | Formato | |
---|---|---|---|
Effects of an EMG-based vibro-tactile biofeedback training on motor learning in children and adolescents with dystonia.pdf
accessibile in internet per tutti
Descrizione: Titolo della tesi
Dimensione
8.34 MB
Formato
Adobe PDF
|
8.34 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/146166