Gait Analysis of Patients Affected by Parkinson’s Disease using Inertial Sensors

Parkinson's disease is a chronic irreversible disease that worsens over time, the clinical picture is extremely complex and varied. In fact, the diagnosis of the disease is not based on diagnostic tests but consists of a clinical investigation that analyses the patient's symptoms and medical history. The motor symptoms are considered the element most characteristic of Parkinson’s disease, these include, gait disorder, bradykinesia, tremor, rigidity and in additional freezing, postural deformation and instability. Several rating scales are used to diagnose the disorder. Currently there are pharmacological and surgical treatments that are used to mitigate the effects of this disease. Gait is a relevant clinical evaluation tool because changes in gait affect the total health of the patient. Measurement has always been a difficult process done only in a clinical setting by a specialist largely through subjective testing. The spread of MEMS technologies has made it possible to carry out objective evaluations at a low price. The work carried out in this thesis sees the use of wearable inertial sensors, used for monitoring and diagnosing Parkinson's disease through gait analysis. The purpose of this thesis is to obtain temporal parameters, characteristics of the gait, through an algorithm that analyses the signals extracted from the sensors worn by each patient, and then carry out a check that allows you to understand the reliability of this analysis compared to the subjective analysis carried out by the specialist. The dataset includes 15 patients affected by Parkinson's disease, the acquisition of the signals was carried out through inertial sensors positioned at the height of the ankles during the execution of various motor tasks. The subjects involved had been diagnosed with Parkinson’s disease for at least a couple of years. For these measurements were used two wearable sensors attached to the ankles of the patients and were taken into account the signals from tri-axis accelerometer and gyroscope. The recording of the signals of interest was conducted during the patient's motor inspection. The extracted parameters are: step time, swing time, stance time, stride time, strike time, double support and mid swing; these parameters were calculated for each task of each patient. After this computation comparisons have been done, in particular, the Wilcoxon Signed-Rank test was performed on the tasks of the same patient, the characteristic parameters of the gait recorded in each task were compared. Second step of this work was to evaluate the mean trend of the tasks performed by each patient. At the end was performed the Pearson correlation between the parameters obtained from the task in which the patients walked at regime along a corridor and the value of Clinical scales. The results obtained from this study can be defined good despite the restricted dataset, in addition other studies in the literature are to support the use of wearable sensors for monitoring and diagnostics of Parkinson’s disease.