Tremor assessment in people affected by movement disorders is conducted with the use of clinical scales. These evaluations suffer from subjectivity, depend on the rater’s ability, and lack objective measures. Approaches able to provide quantitative information are documented in the literature, but not adopted in the clinical practice as they do not adapt properly to the scenario. This work evaluates the instrumented assessment of 4 items of the TETRAS scale, the most commonly used clinical scale for the evaluation of essential tremor (ET), using an innovative smart ink pen. Such a tool, being used as a normal pen, allows performing all 4 items as it is usually done in the clinical examination. At the same time, it records kinematic and force signals from the embedded sensors, which can be analyzed to extract relevant indicators. Nine ET patients were recruited and asked to perform the 4 tasks (postural, spiral, line, phrase) using the smart ink pen. The protocol included test-retest acquisitions, both before and after the Focused UltraSound (FUS) intervention, a treatment aimed at reducing tremor symptoms. More than 150 indicators were computed from the data using already developed algorithms, separately for each task. The indicators quantify the performance in the domains of time, smoothness, force, tilt, tremor amplitude, frequency and regularity. Given that several implementations of the same measured quantity were present in the algorithms, the primary goal was to select the most suitable indicators among redundant ones for the characterization of each task. Such a goal was achieved by performing different analyses, based on indicators reliability and sensitivity to treatment, quantified by the Minimum Detectable Change. Given that the postural task signals showed a non-uniform behaviour throughout the acquisition, the first analysis aimed at understanding whether the indicators obtained from this task are influenced by the signal portion considered for their computation. We found out that the signal section does not have an impact on indicators extraction. The second analysis was devoted to choosing between two methods for indicator calculation in the spiral task: indicators can be computed on the entire signal or on strokes (single tracts in which the pen is continuously in contact with the paper). Ideally, a spiral should be completed in one stroke, but this condition was not always verified. For most of the indicators, the calculation on the entire signal showed more reliable results. A third analysis, for which the methods introduced above were used, consisted in selecting the most suitable indicators, deleting those which gave redundant information. Finally, a new indicator measuring tremor power spectral concentration in the angular velocity signal was developed. In the end, we were able to identify the most suitable pool of indicators for all analysed tasks. Future work should increase the sample size to improve the robustness of the conducted analysis and evaluate indicators validity by assessing their correlation with standardized clinical scales.
La valutazione del tremore in persone affette da disturbi del movimento è attualmente effettuata tramite l’utilizzo di scale cliniche. Queste valutazioni sono però soggettive e dipendono dall’abilità di coloro che le effettuano. In letteratura sono riportati approcci che includono informazioni quantitative, ma questi non sono attualmente adottati nella pratica clinica in quanto non adatti allo scenario. L’obiettivo di questo lavoro è la valutazione strumentale di 4 prove dalla scala TETRAS, la scala clinica più utilizzata per la valutazione del tremore essenziale, usando una penna sensorizzata. Essendo utilizzata come una normale penna, permette di svolgere le 4 prove così come vengono effettuate durante la valutazione clinica. Allo stesso tempo, grazie ai suoi sensori, la penna registra segnali cinematici e di forza che possono essere analizzati per estrarre indicatori rilevanti. Nove pazienti affetti da tremore essenziale sono stati coinvolti nello studio: è stato chiesto loro di eseguire le 4 prove (posturale, spirale, linea, frase) usando la penna sensorizzata. Il protocollo include un’acquisizione test-retest sia prima che dopo un intervento ad ultrasuoni focalizzati, il cui obiettivo è quello di ridurre i sintomi del tremore. Utilizzando algoritmi già esistenti, sono stati calcolati più di 150 indicatori per ogni prova. Gli indicatori quantificano le performance nel dominio del tempo, della fluidità, della forza, dell’ampiezza di tremore, della frequenza e della regolarità. Data la presenza di diverse implementazioni della stessa misura, l’obiettivo principale è stata la selezione degli indicatori migliori per la caratterizzazione di ogni prova. A tal fine, diverse analisi, basate sul calcolo dell’affidabilità degli indicatori e della loro sensitività al trattamento, quest’ultima quantificata con il calcolo del Minimum Detectable Change (MDC). Dato il comportamento non uniforme nel tempo mostrato dai segnali ottenuti nella prova posturale, la prima analisi si è focalizzata sullo studio dell’influenza della porzione di segnale considerata per il loro calcolo degli indicatori. È risultato che la sezione di segnale non ha un impatto sull’estrazione degli indicatori. La seconda analisi è stata invece eseguita per scegliere uno tra i due metodi disponibili per il calcolo degli indicatori della prova spirale: gli indicatori possono essere calcolati sul segnale intero o sugli stroke (tratti di segnale in cui la penna è continuamente a contatto con il foglio). Idealmente, una spirale dovrebbe contenere un solo stroke, ma ciò non è sempre verificato. Il calcolo sul segnale intero ha generato risultati più affidabili. Una terza analisi è stata invece eseguita per selezionare gli indicatori effettivamente utili alla valutazione del tremore, eliminando quelli ridondanti. Infine, è stato creato un nuovo indicatore capace di misurare il tremore a partire dallo spettro della velocità angolare. In futuro, per migliorare i risultati ottenuti, si potrebbe aumentare il numero di partecipanti e valutare la validità degli indicatori comparandoli alle scale cliniche attualmente utilizzate.
Reliability of writing and tremor indicators obtained from a smart ink pen during TETRAS scale assessment in essential tremor patients
CANDIDO, SARA
2021/2022
Abstract
Tremor assessment in people affected by movement disorders is conducted with the use of clinical scales. These evaluations suffer from subjectivity, depend on the rater’s ability, and lack objective measures. Approaches able to provide quantitative information are documented in the literature, but not adopted in the clinical practice as they do not adapt properly to the scenario. This work evaluates the instrumented assessment of 4 items of the TETRAS scale, the most commonly used clinical scale for the evaluation of essential tremor (ET), using an innovative smart ink pen. Such a tool, being used as a normal pen, allows performing all 4 items as it is usually done in the clinical examination. At the same time, it records kinematic and force signals from the embedded sensors, which can be analyzed to extract relevant indicators. Nine ET patients were recruited and asked to perform the 4 tasks (postural, spiral, line, phrase) using the smart ink pen. The protocol included test-retest acquisitions, both before and after the Focused UltraSound (FUS) intervention, a treatment aimed at reducing tremor symptoms. More than 150 indicators were computed from the data using already developed algorithms, separately for each task. The indicators quantify the performance in the domains of time, smoothness, force, tilt, tremor amplitude, frequency and regularity. Given that several implementations of the same measured quantity were present in the algorithms, the primary goal was to select the most suitable indicators among redundant ones for the characterization of each task. Such a goal was achieved by performing different analyses, based on indicators reliability and sensitivity to treatment, quantified by the Minimum Detectable Change. Given that the postural task signals showed a non-uniform behaviour throughout the acquisition, the first analysis aimed at understanding whether the indicators obtained from this task are influenced by the signal portion considered for their computation. We found out that the signal section does not have an impact on indicators extraction. The second analysis was devoted to choosing between two methods for indicator calculation in the spiral task: indicators can be computed on the entire signal or on strokes (single tracts in which the pen is continuously in contact with the paper). Ideally, a spiral should be completed in one stroke, but this condition was not always verified. For most of the indicators, the calculation on the entire signal showed more reliable results. A third analysis, for which the methods introduced above were used, consisted in selecting the most suitable indicators, deleting those which gave redundant information. Finally, a new indicator measuring tremor power spectral concentration in the angular velocity signal was developed. In the end, we were able to identify the most suitable pool of indicators for all analysed tasks. Future work should increase the sample size to improve the robustness of the conducted analysis and evaluate indicators validity by assessing their correlation with standardized clinical scales.| File | Dimensione | Formato | |
|---|---|---|---|
|
2022_Candido.pdf
accessibile in internet solo dagli utenti autorizzati
Dimensione
2.22 MB
Formato
Adobe PDF
|
2.22 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/10589/196535