الفهرس | Only 14 pages are availabe for public view |
Abstract Remote healthcare and telemedicine technology have witnessed a large and rapid development in the last decade with the large development of information and communication technology (ICT). Smart portable products can now be used for monitoring of different medical signals of individuals to track the human general health and to detect abnormalities within some organ functionality, and also in the pre-diagnosis of various diseases. Certain diseases are usually associated with changes in some physiological parameters in the human body such as heart rate, oxygen saturation, respiration rate, respiration pattern, body temperature, and blood pressure. The diagnosis of such diseases involves making some checks in the hospital to measure how the physiological parameter readings are close to the normal rates, and then determine the presence or absence of those diseases. With remote health monitoring, while sitting in the comfortable home, a patient can either monitor his own health, or pass the readings to a specialist to see if certain precautions or actions should be taken. There are two modes for health monitoring: local mode, where the patient himself or some relatives can monitor the biomedical signs and related measurements, and they can call the specialist in case of some readings outside the normal ranges; and remote mode, which provides accessibility for the medical specialist to monitor the patient’s readings, remotely. This provides fast response and interactions from the health organization in case of emergency. The main challenges encountered in healthcare systems are the necessity to capture diverse signals for a complete picture regarding the human health, the need for an efficient monitoring algorithm with minimum time complexity, the need to secure the measured signals with appropriate encryption schemes to keep the security and integrity for the patient health records, the need to transmit the signals to a central point for processing or storage while keeping the encrypted versions of the signals unaffected by the communication channel, the need for efficient schemes to process the received encrypted medical signals for information retrieval, and the need for powerful computational resources that can support the required computations. In this thesis, we present different implementations for efficient, accurate, and secure health monitoring schemes, adopting both local and remote health monitoring modes. The proposed schemes provide reliable solutions to monitor and track several biomedical aspects. This may help in early prediction or diagnosis of different diseases associated with abnormalities of those aspects, and also can be used in emergency cases, where the transmission of critical patient data can make a significant impact on the patient life. In the first approach, we propose a contactless, inexpensive, and fast breathing rate and respiration pattern monitoring algorithm using video analysis and computer vision techniques. In the proposed algorithm, the chest region, which exhibits the intended region of interest (ROI), is automatically detected, and then, a number of dominant corner points within the ROI are extracted and tracked to adjust the boundaries of the ROI during frame processing. The breathing pattern is extracted from the integral form of frames, which simplifies and speeds up the calculations. The proposed algorithm is tested on 28 videos of sleeping-simulated positions, and the results are compared with the manual visual inspection values. In linear regression results, the determination coefficient (R2) is 0.961, which demonstrates high agreement with reference measurements. In addition, the Bland-Altman plot shows that almost all data points are within the 95% limits of agreement. Moreover, the time complexity of the proposed algorithm, which involves taking just a single point value from the integral form of the image, is lower than that of traditional methods that circulate over a large number of points. In other words, the proposed algorithm achieves O(1) fixed time complexity compared to O(N2) for traditional methods. The average speed of processing is enhanced by about 17.4%. Consequently, the proposed approach is promising towards contactless, low-cost, and fast respiration rate monitoring. The purpose of the second proposal is to present a new multi-function portable health monitoring device that can help in the pre-diagnosis of various diseases. It can be used to keep an eye on people we need to care about, while keeping them in their normal daily life. In addition, the proposed device presents a framework for securing the measured signals by adopting the advanced encryption standard (AES) algorithm, and transmitting them over the communication channel with WiFi technology to either a mobile application in a local mode or to the cloud storage through an access point. A medical specialist can visualize the health records in real time only after providing decryption credentials. The proposed device can be used as a stand-alone medical device for all people. It consists mainly of different types of medical sensors and a programmable microprocessor. Because of its small size, it can be easily carried anywhere and used personally to monitor the health of the person at any time. It is not restricted to be used at home only, but can be carried easily in any place: the office, the school, the club, and others. Consequently, the proposed device is expected to perform measurement of heart rate (HR), measurement of blood oxygen saturation (SpO2), acquisition of photoplethysmograms (PPG), acquisition of electrocardiograms (ECG), measurement of body temperature, measurement of air temperature, measurement of air humidity, encryption of measured signals using AES algorithm, and transmission of measured values and the encrypted signals to a mobile or access point for further processing. Moreover, it can send immediate notifications (Email messages) to caregivers if some thresholds on measurements are exceeded. Furthermore, the proposed system measurements are compared with commercially available High Care medical product measurements. Results demonstrate that measurements of the proposed system are within the 95% confidence interval, and the determination coefficient (R2) is 0.983, which demonstrates the high accuracy of the proposed system. One of the biological signals captured from the proposed device is the PPG signal, that is a biological signal, which describes the volumetric change of blood flow in peripherals with heartbeats. We study the possibility of using the PPG signal as a biometric to identify people based on the way their heart beats. Two strategies are adopted in this thesis for human identification using the PPG signals. The first strategy adopts the Mel-Frequency Cepstral Coefficients (MFCCs) algorithm for feature extraction, and artificial neural networks (ANN) for classification of persons. A dataset of PPG signals for 35 healthy persons was collected to test the performance of the proposed approach. Experimental results demonstrate 100% and 98.07% accuracy levels using the hold-out method and the 10-fold crossvalidation method, respectively. Another innovative PPG-based human identification approach based on deep learning is proposed. A convolutional neural network (CNN) is specifically designed for the purpose of PPG signal classification. The proposed identification approach is applied on images with and without an additive white Gaussian noise (AWGN) effect. The simulation results reveal that the proposed approach achieves an accuracy of 99.5% with spectrogram representations and 89.8% with just a 2D image representation, in the absence of noise. In addition, this thesis discusses the efficiency of denoising techniques such as wavelet, Savitzky-Golay and Kalman filters when involved with the proposed approach. The simulation results prove that wavelet denoising gives better performance among the discussed denoising techniques. |