| According to the latest global cancer statistics report published by CA :A Cancer Journal for Clinicians in 2018,breast cancer has become the cancer with the highest morbidity and mortality among women all over the world.Early detection and timely treatment can improve the survival rate of breast cancer patients effectively.We need a non-destructive,simple and convenient breast cancer detection method for early screening of breast cancer urgently.The microwave near-field breast cancer detection came into being,our team has done a lot of research work on it and has proposed the concept of BRATUMASS(Breast Tumor Microwave Sensor System).The main content of this thesis is the study of microwave image optimization based on BRATUMASS,it's also part of the project “Microwave Near-field Breast Cancer Image Inversion Method and Early Diagnosis of Breast Cancer Research”(project number: 08JC14026),which is sponsored by Shanghai Science and Technology Development Fund.The main work of this thesis is as follows:First of all,this thesis introduces the conventional breast cancer detection methods and their limitations.The use of microwave near-field detection for early breast cancer and its advantages are introduced.Based on the research results of our team,the physical basis and imaging principle of BRATUMASS for early breast cancer detection is introduced in detail.In this thesis,the composition of BRATUMASS is summarized.The use of BRATUMASS inversion software and the meaning of microwave image are also introduced in detail.Based on the characteristics of BRATUMASS microwave image--the larger the gray value is,the more likely it is the lesion area,this thesis proposes the processing method of power law transformation for microwave image through the study and comparison of image processing algorithms.By stretching the high gray level of the image,the power law transformation can emphasize the high gray area and its details so that we can locate the tumor area visually and clearly.Compared with the method of marking tumor area with fixed threshold,this method not only highlights the tumor area,but also keeps the difference information between pixels in the tumor area,thus this method can help locate breast tumors accurately.Using some actual cases BRATUMASS microwave image can get the results of these two methods.The analysis and comparison of the two methods' results show that: power law transformation is feasible and advantageous.On this basis,this thesis proposes the use of FCM(Fuzzy C-means)to segment the microwave image to remove suspicious lesions and get the tumor region automatically,combining with the characteristics of microwave image gray histogram.The experiment and result analysis are carried out with three real cases,which proves the feasibility and correctness of automatic segmentation based on FCM.Finally,combining with the distribution characteristics of breast tumors and the function of manually adjusting the upper and lower limits in the inversion software,this thesis proposes an algorithm to determine the upper and lower limits automatically based on convex hull.In this thesis,MATLAB simulation experiment and C++ programming of this algorithm are done and a button to realize this automatic function is added by embedding it into the inversion software.The operator can be better assisted in determining the tumor position by automatically adjusting the upper and lower limits instead of manually.Three groups of real cases were used to test this function,and the results are analyzed and illustrated to prove its feasibility and accuracy.In summary,this thesis mainly studies the tumor region display and automatic segmentation of BRATUMASS microwave image based on our team research work,designs and implements the algorithm for automatically determining the upper and lower limits for the inversion software. |
PDF Full Text Request