Time Reversal Algorithm for Brain Cancer Detection Aims

Time Reversal Algorithm for Brain Cancer Detection Aims & Objectives Background This project aims to evaluate the performance of the Time Reversal algorithm for detection of brain cancer. Time Reversal techniques exploit the natural phenomenon that if an array of transducers re-emit a signal they have received in reverse, it will converge back towards its source. The algorithm was first demonstrated by Mathias Fink and his team in 1989 and he stated: Objectives: • Understand affects of changing shape, size and location of tumour. • Perform simulations with accurate electrical permittivity and conductivity parameters of the tumour. • Understand how media parameters of the tumour relate to real-world development and growth of brain tumours. • Evaluate how time reversal algorithm performs with changing parameters by performing data analysis and assessment on the limitations of the time reversal algorithm and the minimum and maximum parameters detectable. “A reflective target whose position is unknown can be focused through complicated inhomogenous medium and the most reflective one is focused with re-iterating the process in the prescence of several targets”. This is particularly applicable to brain cancer detection as due to significant differences in electrical permittivity and conductivity, brain tumours are highly reflective. Motivation Cancer causes almost 15% of all human deaths, brain cancer is a cancer that even now still has very high mortality rates; 60% within a year and 81% within 5 years of diagnosis. Mortalities from brain cancer have halved within the last 50 years, but it is evident that early detection is key and there are still huge advancements to be made in the field of brain cancer detection and diagnosis. Figure 2 This project focuses on gliomas in the frontal lobe, temporal lobes and occipital lobe as these are very hazardous and account for the majority with over 57% of all brain tumours. Figure 1 shows the anatomic site distribution of gliomas in an axial projection of the brain, the shading represents the number of gliomas in each 1 x 1 cm square with smoothing based on adjacent squares. The inset to the right shows the section plane. Progress A significant amount of data has been obtained from the literature review and research into the media parameters of the brain tumour, with usable values for electrical permittivity and conductivity and a good idea of where to focus simulations due to where gliomas are likely to manifest and the sizes they are currently only detected at. The below simulation results (Figure 3) show distinct accuracy and reasonable precision in focusing on the modelled brain tumour within the digital human phantom. Accurate values for electrical permittivity, conductivity and size of the tumour have been implemented, however, more work still needs to be done on the relevant location of the tumour and if possible speculation as to what shape should be modelled. Figure 1 Problems Encountered The location and number of antennas used in the simulation has been problematic, while a greater number of antennas should always yield greater resolution, too many antennas prevents the simulation from being able to complete. It has also been discovered that the location of the antennas from the simulated tumour only yields reasonable results within a fairly small definable range. The media parameters of gliomas have also provided difficulties. While research has yielded adequate information about the electrical permittivity, conductivity, location and to some extent their size; data encompassing the modal shapes of gliomas has been much harder to come by and it is doubtable that within the scope of this project tumours with accurate shape will be simulated. Iteration 0 Iteration 500 Iteration 1000 Iteration 2500 Figure 3 William John Holdoway Electrical & Electronic Engineering School of Electrical and Electronic Engineering The University of Manchester William. holdoway@student. manchester. ac. uk Supervisor: Fumie Costen Fumie. costen@manchester. ac. uk