Research On The Method Of The Extraction And Quantification Of Parkinsonian Tremor Signal

Parkinson’s Disease is one of the current incidence increased year by year, which is a nervous system degenerative disease. With the acceleration of population aging in our country in the future, a number of Parkinson’s Disease patients will be very large, and this will bring the family great physical and mental pressure, and heavy financial burden. Due to the particularity of Parkinson’s disease, clinical doctors according to the development of the patient’s medical history, physical examination, symptoms(slow, resting tremor, etc.) and unified Parkinson disease rating scale to the patient condition to judge, but this kind of judgment has great subjectivity. Therefore, the study on an objective evaluation method of the course is necessary.There are currently condition judgment method based on electrophysiological changes in surface EMG, EEG signal, but collection methods of such signals are invasive type, the patient’s mood will be impact, leading to the large measurement deflection. Resting tremor of Parkinson’s disease is more common first symptom, it can be detected by non-contact detection, which can eliminate the detecting way bring negative emotions to the patient, and reduce the deviation between the measurement results and real tremor. Currently, there are many deficiencies analysis and study of Parkinson’s tremor: firstly, the inherent physiological tremor of human body is not taken into account in the collected signal; secondly the studies only classified Parkinson tremor and other tremors, but the severity of Parkinson’s tremor was not quantified. In view of the above problem, this paper uses methods about the Parkinson’s tremor extraction and quantitative.In this paper, based on the non-contact Parkinson’s tremor detector measured signal as the research object, characteristics of physiological tremor and Parkinson’s tremors is deeply studied.The de-noising of original signal is achieved. And to obtain the Parkinson tremor signal the extraction method based on the adaptive filtering structure is adopted. The characteristics of the Parkinson’s tremor signals is analyzed, and a quantitative model to Parkinson’s tremor signals is established. The correlation coefficient between quantitative results of tremor severity of Parkinson’s patients and the results of clinicians’ judgements is achieved. The rationality and validity of the model is verified. The main work and innovations of this paper are summarized as follows:(1) The characteristics of physiological tremor signals and Parkinson’s tremors signals is in-depth studied. The Parkinson patients with tremor signal acquisition based on the detector of three-axis accelerometer and gyroscope and position sensitive detector is researched. The composition and characteristics of noise in the tremor acquisition process is analyzed.(2) According to the characteristics of each component noise in the tremor signal, the noise filtering methods outside the tremor signal’s band is researched. The band-pass filter(removed noise outside the signal band tremor) is used. The improved stationary wavelet thresholding algorithm(removed noise within the frequency band) is put forward. The Fourier- linear combiner(Fourier linear combiner, FLC) based on adaptive methods and its derivatives combiners(modeling of physiological tremor) are adopted. The adaptive filtering structure(get Parkinson’s tremor signal) is used. Then the realistic basis for Parkinson’s tremor’s quantification is provided.(3) The characteristic parameters of Parkinson tremor signal is researched. The dominant frequency of the tremor signal obtained by power spectrum estimation as a supplementary basis for judging whether the disease was affected is used. The Parkinson tremor signal amplitude is regarded as the object of quantification. A multiple parameters quantitative model to calculate the data outputted from the detectors measured data is builded. The correlation coefficient between quantitative results of tremor severity of Parkinson’s patients and the results of clinicians’ judgements is obtained.Experimental data shows that Parkinson’s tremor signal which comes from the detection signal after de-noising and extracted, is analyzed by PSD analysis. Then the prediction score calculated by the multiple parameters quantitative model is compared to the UPDRS rating scores. The results show that the predicted score and clinical rating UPDRS score have a very high correlation. The corresponding correlation about the detection device based gyroscope and position sensitive detector is higher than the acceleration sensor’s.