| Summary: | 碩士 === 國立交通大學 === 生醫工程研究所 === 108 === Hypertension is a major risk of cardiovascular disease morbidity and mortality. Therefore, blood pressure (BP) is an important physiological parameter in medicine. The BP measurement methods of the using the auscultatory technique and oscillometric technique are commonly used in hospitals. These methods require trained people to make these measurements and this parameter cannot be measured continuously. The long press of the oscillometric technique makes the subject feel uncomfortable. Therefore, the demand of continuous cuffless BP monitor is presented. In order to meet this demand, the pulse wave velocity (PWV) method is considered as a potential technology. Because individual differences lead to poor accuracy, parameters of the blood vessels become important. Therefore, it is indispensable to obtain the parameters of blood vessels with tomographic images.
Electrical impedance tomography (EIT) is a non-invasive and non-radiative medical imaging technique based on detecting the inhomogeneous electrical properties of the tissue. The inverse problem of EIT is a highly nonlinear ill-posed problem, which is the main reason that affects image quality. Our goal is to solve the EIT inverse problem using the nonlinear mapping properties of artificial neural networks (ANNs) and convolutional neural networks (CNNs). In this thesis, the Adam optimization method and mean-squared-error (MSE) function are used to train an ANN to solve the inverse problem and a CNN to process the ANN image. The networks are trained on datasets of simulated data, and tested on datasets of simulated data and experimental data. Results for time-difference EIT (td-EIT) images are presented using simulated EIT data from EIDORS and experimental EIT data from our developed EIT systems. The results are used to compare the proposed method with the one-step Gauss–Newton linear method. The proposed method offers improved resolution (RES), low position error (PE) and excellent artefact removal compared to the existing methods. The phantom results are shown that can improve the RES by 50 to 70 percent and reduce the PE by 60 to 70 percent using our method. The improvements in RES and processing speed are essential for clinical EIT measurement of dynamic physiological processes.
|
|---|
