Axial disorders negatively affect the quality of life of those people affected by Parkinson’s disease (PD). Indeed, gait alterations and postural abnormalities are one of the most invalidating motor symptoms of the pathology. Assessments to evaluate PD-related motor impairments are based on the subjective observation of an individual physician delivering clinical scales, which might lack of intra-session and inter-operator repeatability. In the last couple of decades, clinical gait analysis has become a useful instrument to characterize motor alterations induced by the disease, and leaded to quantify a reduction in gait speed, step length and lower-limb joints range of motion. In addition to this, the adoption of surface electromyography (sEMG) allowed to determine abnormalities in lower-limb muscle activity in PD when compared to healthy controls, both in terms of magnitude and timing of the activation profile. Even though those are established tools in the clinical practice, they do not provide information on the causal relationship between the altered neural activation and the disrupted joint kinematics in people suffering from PD in vivo. In this direction, data-driven neuromusculoskeletal modeling (NMSM) approaches enable to understand how neural command is translated into mechanical output. These simulations provide sets of variables which can be used to implement new quantitative objective metrics for assessing individuals’ neural motor control allowing to gain insights on the disease progression. These models, informed via the subject’s true muscle activity, can be calibrated and tuned to provide a higher degree of personalization. Yet, the translation into clinical environments of NMSM techniques is still hampered due to practical limitations. First of all there is no direct measurement of muscle force in vivo to validate the model outcomes. However, it is generally accepted in the common practice to use sEMG profile to validate patterns of muscular force. Second, in laboratory conditions, sEMG signals to be used as input for the modeling pipeline are normalized with respect to their associated maximum voluntary contraction (MVC), which is unattainable in neurological populations due to the reduced muscle voluntary selectivity. In addition to this, a high number of sEMG signals are necessary in order to obtain a full characterization of the motor functions, making the subject’s preparation time long and cumbersome, causing discomfort and fatigue. Finally, due to their simplicity of use, static optimization (SO) techniques, that do not require any a priori knowledge on the muscle activation, are usually preferred over sEMG-driven approaches when solving the muscle redundancy problem. Nonetheless, SO modeling based approaches assume identical neuromuscular control strategies between individuals and tasks in the muscle force distribution algorithm, thus not allowing a tailored personalization of the subject’s motor control. Therefore, the present doctoral project is a first attempt to propose a reliable sEMG-informed modeling pipeline suitable for clinical application aiming to overcome the experimental setup limitations (i.e., the need for MVC collection and the requirement of a high number of sEMG signals) while maintaining a proper degree of model personalization in the context of people with PD. Results linked with this project might provide quantitative and repeatable metric defining a new set of possible biomechanical biomarkers which can objectively report on the motor capacity of the patient, reducing the possible sources of variability affecting the clinical scales. Moreover, the model-based estimation of in vivo variables may provide additional information to clinicians that could be used to plan intervention treatments aiming to restore and improve muscle forces.
I disordini assiali impattano negativamente la qualità della vita delle persone affette da Morbo di Parkinson (PD). Infatti, alterazioni del normale ciclo del passo e anomalie posturali sono uno dei sintomi motori più invalidanti di questa malattia. I metodi per la valutazione dei disordini motori indotti dal PD si basano sull’osservazione soggettiva di un singolo specialista che somministra scale cliniche, le quali possono mancare di ripetibilit`a inter-operatore e intra-sessione. Negli ultimi decenni, l’analisi del movimento in clinica si è affermata come un utile strumento per la caratterizzazione delle alterazioni motorie indotte dalla patologia e ha permesso di quantificare una riduzione nella velocit`a del passo, nella sua lunghezza e nel range of motion articolare nell’arto inferiore. In aggiunta a questo, l’elettromiografia superficiale (sEMG) ha permesso di determinare alterazioni nell’attività elettrica dei muscoli dell’arto inferiore nel PD se confrontata rispetto a una popolazione di controllo, sia in termini di intensit`a che di timing di attivazione. Sebbene questi siano metodi stabiliti nella pratica clinica, risulta impossibile fornire informazioni riguardanti la relazione causale tra l’alterata attivazione neurale e il deficit motorio nelle persone affette da PD in vivo. In questa direzione, approcci data-driven di modellazione neuro-muscolo-scheletrica (NMSM) consentono di capire come il comando neurale sia tradotto in output meccanico. Queste simulazioni forniscono un set di variabili che possono essere utilizzate per l’implementazione nuove metriche quantitative ed oggettive per la valutazione del controllo motorio dell’individuo, e permettono di approfondire la comprensione nella progressione della malattia. Questi modelli possono essere calibrati e settati per fornire un ulteriore grado di personalizzazione tramite l’incorporazione dell’attivit`a muscolare sperimentale del soggetto nel flusso di elaborazione dei dati. Tuttavia, il trasferimento di queste tecniche in pratica clinica è ostacolato da limitazioni pratiche. Innanzitutto, non esiste una misura diretta della forza muscolare in vivo per la validazione dei risultati del modello. Ciononostante, è generalmente accettato nella pratica utilizzare il profilo di segnale sEMG per validare i pattern di forza muscolare. Secondo, in condizioni di laboratorio, i segnali sEMG da utilizzare come input della pipeline di modellazione vengono normalizzati rispetto alle loro contrazioni massimali volontarie (MVC), che sono per`o difficilmente ottenibili in popolazioni affette da disordini neurologici a causa della ridotta selettività manifestata nelle contrazioni muscolari. E’ inoltre necessario un gran numero di segnali sEMG per ottenere una completa caratterizzazione del profilo motorio del soggetto: questo pu`o rendere la preparazione dispendiosa in termini di tempo e pu`o provocare disagio e affaticamento in chi viene sottoposto alla valutazione. Infine, a causa della semplicità del loro utilizzo, tecniche di ottimizzazione statica (SO), che non richiedono alcuna conoscenza a priori sull’attivazione muscolare, sono solitamente preferite agli approcci guidati ad sEMG nella risoluzione del problema di ridondanza muscolare. Tuttavia, i modelli basati su SO assumono una strategia di controllo motorio identica tra diversi individui nell’algoritmo di distribuzione della forza, e quindi non permettendo una personalizzazione del controllo motorio adattata al soggetto. Il presente progetto di dottorato è un primo tentativo di proporre una pipeline di modellazione informata con sEMG che sia affidabile e adatta per l’applicazione clinica e che miri a oltrepassare le limitazioni dovute al setup sperimentale (come la necessit`a di acquisire MVC e un elevato numero di canali sEMG) mantenendo un grado di personalizzazione accettabile nel contesto di soggetti affetti da PD.
Characterization of motor control in Parkinson’s disease: a translational electromyography-informed modeling approach / Romanato, Marco. - (2023 Mar 17).
Characterization of motor control in Parkinson’s disease: a translational electromyography-informed modeling approach
ROMANATO, MARCO
2023
Abstract
Axial disorders negatively affect the quality of life of those people affected by Parkinson’s disease (PD). Indeed, gait alterations and postural abnormalities are one of the most invalidating motor symptoms of the pathology. Assessments to evaluate PD-related motor impairments are based on the subjective observation of an individual physician delivering clinical scales, which might lack of intra-session and inter-operator repeatability. In the last couple of decades, clinical gait analysis has become a useful instrument to characterize motor alterations induced by the disease, and leaded to quantify a reduction in gait speed, step length and lower-limb joints range of motion. In addition to this, the adoption of surface electromyography (sEMG) allowed to determine abnormalities in lower-limb muscle activity in PD when compared to healthy controls, both in terms of magnitude and timing of the activation profile. Even though those are established tools in the clinical practice, they do not provide information on the causal relationship between the altered neural activation and the disrupted joint kinematics in people suffering from PD in vivo. In this direction, data-driven neuromusculoskeletal modeling (NMSM) approaches enable to understand how neural command is translated into mechanical output. These simulations provide sets of variables which can be used to implement new quantitative objective metrics for assessing individuals’ neural motor control allowing to gain insights on the disease progression. These models, informed via the subject’s true muscle activity, can be calibrated and tuned to provide a higher degree of personalization. Yet, the translation into clinical environments of NMSM techniques is still hampered due to practical limitations. First of all there is no direct measurement of muscle force in vivo to validate the model outcomes. However, it is generally accepted in the common practice to use sEMG profile to validate patterns of muscular force. Second, in laboratory conditions, sEMG signals to be used as input for the modeling pipeline are normalized with respect to their associated maximum voluntary contraction (MVC), which is unattainable in neurological populations due to the reduced muscle voluntary selectivity. In addition to this, a high number of sEMG signals are necessary in order to obtain a full characterization of the motor functions, making the subject’s preparation time long and cumbersome, causing discomfort and fatigue. Finally, due to their simplicity of use, static optimization (SO) techniques, that do not require any a priori knowledge on the muscle activation, are usually preferred over sEMG-driven approaches when solving the muscle redundancy problem. Nonetheless, SO modeling based approaches assume identical neuromuscular control strategies between individuals and tasks in the muscle force distribution algorithm, thus not allowing a tailored personalization of the subject’s motor control. Therefore, the present doctoral project is a first attempt to propose a reliable sEMG-informed modeling pipeline suitable for clinical application aiming to overcome the experimental setup limitations (i.e., the need for MVC collection and the requirement of a high number of sEMG signals) while maintaining a proper degree of model personalization in the context of people with PD. Results linked with this project might provide quantitative and repeatable metric defining a new set of possible biomechanical biomarkers which can objectively report on the motor capacity of the patient, reducing the possible sources of variability affecting the clinical scales. Moreover, the model-based estimation of in vivo variables may provide additional information to clinicians that could be used to plan intervention treatments aiming to restore and improve muscle forces.File | Dimensione | Formato | |
---|---|---|---|
Tesi_Marco_Romanato.pdf
embargo fino al 16/03/2026
Descrizione: Tesi Marco Romanato
Tipologia:
Tesi di dottorato
Dimensione
17.72 MB
Formato
Adobe PDF
|
17.72 MB | Adobe PDF | Visualizza/Apri Richiedi una copia |
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.