Uwb microwave radar for early detection of head stroke

Head stroke compromises cerebral blood flowing, leading to brain injuries. It is one of the main causes of disability and death worldwide and it requires an immediate diagnostic investigation and therapeutic solution. It would be extremely important to detect as soon as possible any abnormality, by using a device that can be carried on site by first response paramedical teams or in the ambulance. This is not possible with traditional diagnostics tools, such as CT (Computed Tomography) or MRI (Magnetic Resonance Imaging), which can be used only in hospital, under specific conditions. Furthermore, the response time might be long (the CT exam takes about 10 minutes, while the MRI takes about 30 minutes). The objective of this work is the design of a diagnostic device for early detection of cerebral vascular diseases in adults, which is accurate and reliable, low cost, low complex, non-invasive for patients, able to provide a fast response (about few minutes) and transportable. Such a device could be used by first responders or on the ambulance. In such a way, upon arrival at the hospital, physicians already know critical patients that need further diagnostic exams. In this way, unnecessary costly diagnostic exams (and in some cases, invasive) can be avoided and the management of the emergency situations (where there are more injured patients) could be greatly improved. Moreover, the device could be also used in hospital for daily monitoring of patients at risk. For the purpose of this work, Microwave Imaging (MWI) is emerging as an interesting solution. MWI is not invasive and allows to design low cost and low complexity devices. In this work, we focused on one of the possible MWI approaches, the microwave UWB radar. The backscattered signals acquired are first pre-processed by artifacts removal algorithms. Then, those cleaned signals are processed by beamforming algorithms. This work reveals that not all of the state-of-the-art artifacts removal algorithms, mainly proposed for breast cancer detection, are able to achieve good performance. Some algorithms which show good performance with noiseless signals conversely provide poor results in noisy conditions. Therefore, novel algorithms for both artifacts removal and beamforming phase have been proposed, showing very reliable and good 2 performance, in terms of accuracy of stroke localization, amount of artifacts not completely removed and computational complexity, even in noisy conditions. The thesis is organized as follows. In Chapter 1 the background of the proposed system is presented. The traumatic brain injury is defined. Microwave imaging approaches, namely microwave tomography and UWB radar, are presented. We then focus on UWB radar imaging devices for stroke detection. Finally, the analysis of the state-of-the-art supports motivation and objectives of the thesis. In Chapter 2 the main assumptions of the system model are presented, with the description of the antennas system and the head model. Chapter 3 presents the main algorithms proposed in literature both for artifacts removal and beamforming, outlining advantages and drawbacks and identifying the directions in which improvements are needed. Chapter 4 presents novel solutions both for the artifacts removal and the beamforming, showing, with extensive comparison results, the effectiveness of the proposed solutions with respect to the state-of-the art algorithms. Finally, the conclusions and future perspectives are discussed in Chapter 5.