An ultra-wide band non-invasive system for health screening /
Minarul Islam,
An ultra-wide band non-invasive system for health screening / Minarul Islam - xiv, 115 pages : illustrations (some color) ; 30 cm. + 1 CD ROM
Faculty of Electrical and Electronics Engineering
Thesis (Master of Engineering (Electronics)) -- Universiti Malaysia Pahang – 2021
Includes bibliographical references
Based on its rapidly growing prevalence, diabetes is a major concern for health and is regarded as a world-wide problem in world health organizations (WHO). For a diabetic patient, it is important to maintain a normal level of blood glucose concentration level (BGCL) for a healthy lifestyle. Frequent testing of BGCL for type-1 diabetes is an important part of diabetic treatment. A lab or a self-test with a tool (e.g. glucometer) involves a painful needle to extract blood from a portion of the body. This excruciating process must be performed many times regularly in serious cases. A non-invasive (bloodless) and patient-friendly measurement approach is important in order to reduce this pain. Thus, a simple, easy, and economical non-invasive diabetes measurement method for end-users will mainly be built in the current research. An effective BGCL measurement system is proposed here using ultra-wideband (UWB) technology and deep learning-based enhanced artificial neural network (ANN) algorithm/module. It consists of two antennas; one is used for transmitting UWB signals from one side of an earlobe and the other one for receiving scattered signals from the opposite side. The earlobe is considered due to its BGCL measurement suitability as there is no bone, has saturated blood flow, and minimal thickness variation from person to person. The UWB antenna with wide bandwidth as well as low return loss is essential in this case and to design such an antenna is one of the objectives here. The development of a complete user-friendly and affordable software module for BGCL measurement system using an enhanced ANN algorithm with a graphical user interface (GUI) has also been focused due to scarce of such a module. A modified circular shaped microstrip patch compact antenna is designed and fabricated to transmit and receive the UWB signals. The bandwidth gain and directivity of the proposed antenna are 13.6 GHz (3.7 GHz-17.3 GHz), 5.38 dB and 6.79 dBi respectively, which shows 4.83 GHz bandwidth enhancement than other related existing antennas. Then, an ANN based deep learning signal processing algorithm, named, attention featured gate generator-long short-term memory (AFGG-LSTM) along with GUI are developed and integrated to measure the BGCL results easily. Besides, systolic blood pressure (SBP), diastolic blood pressure (DBP), pulse rate (PR), and body temperature (Temp) are also been considered as additional features to measure simultaneously, so, the AFGG-LSTM algorithm has been enhanced accordingly. The designed GUI and software module are then integrated with UWB antennas connected with transceivers to develop desired complete system for health status measurement. Total 1100 data samples are collected by the system and corresponding features are extracted, which fed into the AFGG-LSTM module to train, validate, and test the system performance. The obtained system accuracy is about 94.76%, 86.30%, 85.93%, 88.24%, and 98.16% in terms of BGCL, SBP, DBP, PR, and Temp respectively. The overall BGCL accuracy of proposed system is at least 12.76% enhanced compared to related existing systems (UWB system-based technique). This gives assurance of reliable BGCL measurement regularly at home in an affordable and easy operable way to save precious human life in near future.
THE0009286(Local)
Faculty of Electrical and Electronics Engineering--Dissertations
Universities and colleges--Dissertations
Theses
An ultra-wide band non-invasive system for health screening / Minarul Islam - xiv, 115 pages : illustrations (some color) ; 30 cm. + 1 CD ROM
Faculty of Electrical and Electronics Engineering
Thesis (Master of Engineering (Electronics)) -- Universiti Malaysia Pahang – 2021
Includes bibliographical references
Based on its rapidly growing prevalence, diabetes is a major concern for health and is regarded as a world-wide problem in world health organizations (WHO). For a diabetic patient, it is important to maintain a normal level of blood glucose concentration level (BGCL) for a healthy lifestyle. Frequent testing of BGCL for type-1 diabetes is an important part of diabetic treatment. A lab or a self-test with a tool (e.g. glucometer) involves a painful needle to extract blood from a portion of the body. This excruciating process must be performed many times regularly in serious cases. A non-invasive (bloodless) and patient-friendly measurement approach is important in order to reduce this pain. Thus, a simple, easy, and economical non-invasive diabetes measurement method for end-users will mainly be built in the current research. An effective BGCL measurement system is proposed here using ultra-wideband (UWB) technology and deep learning-based enhanced artificial neural network (ANN) algorithm/module. It consists of two antennas; one is used for transmitting UWB signals from one side of an earlobe and the other one for receiving scattered signals from the opposite side. The earlobe is considered due to its BGCL measurement suitability as there is no bone, has saturated blood flow, and minimal thickness variation from person to person. The UWB antenna with wide bandwidth as well as low return loss is essential in this case and to design such an antenna is one of the objectives here. The development of a complete user-friendly and affordable software module for BGCL measurement system using an enhanced ANN algorithm with a graphical user interface (GUI) has also been focused due to scarce of such a module. A modified circular shaped microstrip patch compact antenna is designed and fabricated to transmit and receive the UWB signals. The bandwidth gain and directivity of the proposed antenna are 13.6 GHz (3.7 GHz-17.3 GHz), 5.38 dB and 6.79 dBi respectively, which shows 4.83 GHz bandwidth enhancement than other related existing antennas. Then, an ANN based deep learning signal processing algorithm, named, attention featured gate generator-long short-term memory (AFGG-LSTM) along with GUI are developed and integrated to measure the BGCL results easily. Besides, systolic blood pressure (SBP), diastolic blood pressure (DBP), pulse rate (PR), and body temperature (Temp) are also been considered as additional features to measure simultaneously, so, the AFGG-LSTM algorithm has been enhanced accordingly. The designed GUI and software module are then integrated with UWB antennas connected with transceivers to develop desired complete system for health status measurement. Total 1100 data samples are collected by the system and corresponding features are extracted, which fed into the AFGG-LSTM module to train, validate, and test the system performance. The obtained system accuracy is about 94.76%, 86.30%, 85.93%, 88.24%, and 98.16% in terms of BGCL, SBP, DBP, PR, and Temp respectively. The overall BGCL accuracy of proposed system is at least 12.76% enhanced compared to related existing systems (UWB system-based technique). This gives assurance of reliable BGCL measurement regularly at home in an affordable and easy operable way to save precious human life in near future.
THE0009286(Local)
Faculty of Electrical and Electronics Engineering--Dissertations
Universities and colleges--Dissertations
Theses