Microwave Breast Cancer Imaging And Fast Recognition

Breast cancer is becoming a big threat to women’s health because of the rising incidence all over the world.Regular screening is most effective in detecting early stage breast tumor,and the earlier therapy start,the better prognosis is.The main clinical screening devices mainly include mammography and ultrasonography.However,patients often suffer radiation exposure risk and uncomfortable pressure on breast.Recently,radar-based ultra-wideband(UWB)microwave breast tumor imaging has been proven to be a very promising technique for new screening tool development due to its non-ionization,low cost,miniaturization and high accuracy.The physical basis of the new technique is the detectable difference in dielectric properties of breast lesions and normal tissues.Radar near-flied imaging system can capture these difference and display tumor locations on screen.The researches mainly focus on pre-feasibility study problems,such as anatomically realistic numerical model,accurate imaging technique,artifact removal and fast recognition of breast lesions.Experimental system is also built to make some preliminary researches on tumor imaging.Firstly,anatomically realistic numerical breast model is developed based on magnetic resonance images(MRI).The technique for the first time adds accurate intrathoracic region into three-dimensional(3D)breast model based on hybrid contour detection method.The numerical model is compatible with finite-different time-domain(FDTD)grid and is capable of simulating the case that tumor is close to chest wall.Then a more accurate MRI-aided tumor imaging method is proposed based on gland modeling to accurately calculate signal arrival time in dense breast,which makes the new imaging method feasible for breasts of various densities.An efficient artifact removal method is developed for breast imaging in supine position using movable linear antenna array.The method can greatly suppress the early time content of received signal and is capable of highlight tumor voxels in small tumor cases.The method is useful for developing planar antenna array real time operation monitor.A pole characteristics-based matching pursuit and likelihood ratio test(MPLRT)method is proposed for fast breast lesion identification.The dielectric properties of the lesion can be estimated efficiently with the best matched reference tumor poles dictionary.This method can be quickly implemented with one or two antennas.Moreover,an energy-based artifact removal is invented using singularity expansion method(SEM),which ensures poles can be directly extracted from received signal instead of pure tumor response.This identification technique is promising to be utilized for developing a practical clinical device in future.A primary experimental system is constructed for tumor imaging.An MRI-derived breast phantom is manufactured using 3D printer.The breast phantom has a cavity filled with lipoid liquid with a tumor phantom hang into it.A dual port vector network analyzer connected with two antennas is used to measure and store data.Confocal microwave imaging(CMI)is used to show tumor location.