| Abstract | Brain stroke is globally one of the most widespread sorts of brain abnormalities. There are common symptoms between the transient ischemic attack (TIA), strokes and generic medical conditions like fainting, migraine, heart problems and seizures. Therefore, the other health conditions should not be misdiagnosed with stroke. It is well known that providing immediate medical attention for a patient with a brain injury is of vital importance. Every second, from the moment of brain injury, millions of brain cells die, leading to irreparable and permanent damage or even death. Thus, if medical staff diagnose stroke, and perform an appropriate drug treatment within a few hours of the symptoms onset, they play a crucial role in saving a patient’s life.
The key factor in treatment is to reliably diagnose the stroke immediately. Hence, a portable diagnosic system is pivotal on the spot for rapid diagnosis of brain injuries. Initially, a clinical examination using a neurological assessment is performed by a general practitioner (GP). Compared to CT and MRI scanners, microwave imaging (MWI) can provide a portable detection system, and allow initial diagnosis of various emergency, life-threatening circumstances such as strokes due to brain injury, whilst patients are still being taken by ambulance to hospital,
and saving critical time. In recent years, MWI has emerged as a promising non-ionising and non-invasive technology for a range of applications, particularly medical applications. In the current thesis, radar-based MWI is proposed as a procedure for brain haemorrhagic stroke detection. This imaging procedure has also more advantages such as low cost, being portable, fast, and easy to use with a good potential for brain haemorrhage detection. In MWI, the imaging of different human head tissues relies on their different response (i.e., electric contrast) to an applied microwave radiation. MWI is a screening technology for detection and monitoring of haemorrhagic stroke, tumours and cancerous cells, based on the significant contrast in the dielectric properties at microwave frequencies of normal and abnormal tissues. This thesis deals with the use and validation of an innovative low complexity MWI procedure for brain imaging,
where antennas operate in free space. In particular, we employ only two microstrip antennas, operating between 1 and 2 GHz for successful detection of the haemorrhagic stroke. Detection is achieved using both simulation and experimental measurements.
I. In the first stage, a wideband (WB) microstrip antenna with fractal ground plane is proposed, simulated and fabricated for brain haemorrhage detection. The designed antennas exhibit a WB working frequency between 1-2 GHz. This band has demonstrated to be ideal and optimal to do brain imaging; in addition, it is obviously emphasised that WB can
enhance performance in lesion detection. The simulations have been performed applying an anthropomorphic human head model where a haemorrhagic stroke has been inserted (using CST Microwave studio). The simulation results concluded that the emulated brain haemorrhagic stroke can be distinguished at four different positions of 0◦, 5◦, 40◦, and 45◦.
II. The second stage of this study presents a hemi-ellipsoidal human head phantom with a millimetric cylindrically-shaped inclusion to emulate brain haemorrhage (suitable to be used inside the anechoic chamber) and a human head phantom (suitable to be applied in MWI device). The process has been performed based on the following procedures:
- In the second, stage, first, multi-biostatic frequency-domain measurements have been performed to collect the transfer function (S21) between two proposed mono-static radar system based antennas inside an anechoic chamber using a multi-layered phantom mimicking a human head. This procedure is used to measure the received signal (S21). A
Vector Network Analyser (VNA) is linked to the mentioned antennas, and the measured (S21) are recorded when they changed the position to every new observation position.
Subsequently, the measured (S21) are post-processed in order to generate microwave images with emphasising the object (e.g. the tumour or the stroke). In this stage, on the basis of the measurement results, it is concluded that the object (brain haemorrhagic stroke phantom) can be successfully detected at four different positions of 0◦, 90◦, 180◦ and 270◦.
- Secondly, since the results coming from measurements inside the anechoic chamber are not as realistic as clinical trials reports and also there is a medical requirement for a brain
stroke portable imaging device, we have come to a decision on applying different signal pre-processing methods to the imaging results collected from a portable MWI device for brain haemorrhage imaging. A portable MWI device, which operates in free space with two azimuthally-rotating antennas, has been used for brain haemorrhage detection.
Measurements are performed by recording the complex (S21) in a multi-bistatic fashion, i.e. for each transmitting position the receiving antenna is moved to measure the received signal every 4.5◦, leading to a total of 80 receiving points. In conclusion, based on the results of the MWI device, the inclusion emulating the brain haemorrhage may be detected at four different positions of 0◦, 90◦, 180◦ and 270◦.
In this thesis, all images have been obtained through Huygens Principle (HP). To reconstruct the image, signal pre-processing techniques are used to reduce artefacts (which may be due to the direct fields and the fields reflected by the first layer). Subtraction artefact removal method between the data of a healthy head and the data of a head with stroke has been initially employed in simulation and measurements. Accordingly, an "Ideal" image would be generated using this artefact removal method to prove the concept of the technology. This would mean that the "Ideal" image performed as a reference for the comparison with the resulting image
from using other artefact removal methods. It is important to point out that, for the purposes of real scenario, there is no possibility of applying this artefact removal method to medical imaging, where the ideal response is not calculated or known. Hence, in clinical trials this artefact removal method cannot be helpful. In addition to the subtraction artefact removal method, in this research, four more methods have been introduced and investigated. These methods consist
of rotation subtraction, average subtraction, differential symmetric receiver type, and summed symmetric differential. The subtraction and rotation subtraction artefact removal methods have been used both in simulations and measurements. It has been verified that all artefact removal procedures allow detection. Subsequently, 6 dedicated image quantification procedures have been implemented in order to assess the detection capability. These procedures comprise area difference, centroid difference, signal-to-noise ratio, structural similarity index metric, image quality index, and signal-to-clutter ratio. Validation of the techniques through both simulation and experimental measurements have been performed and presented, illustrating the effectiveness of the methods. |
|---|