Fractal Analysis For Ultrasound Image Of Breastcancer

Objective:As a graphica lanalysis tool,fractal analysis has been applied in medica limage processing.Explore the diagnostic value of fractal analysis applied to diffe rential diagnosis of benignand malignant breast tumors by comparing the fractal dimension of ultrasound images of patients with breast tumor.Methods:We retrospectively analysed 64 cases of ultrasound images of patients with breast tumor in our hospital from February 2014 to February 2015,includin g 32 cases of breast cancer,32 cases of patients with breast fibroadenoma.We cal culate the fractal dimension of lesion areas.The fractal dimension is the object of fractal geometry,and fractal geometry is a irregular geometry for the study of g eometry.It had already been identified that many systems,phenomenons and the pr ocess of nature had the fractal geometry.For example,the nebula’sdistribution,the s hape of the coastline,the human vascular system and lung membrane structure etc.Because of the geometric structure has self-similarity,so the complexity of its stat e can be used with exponential function D(the fractal dimension).A lot of experi ments in foreign countries had proved that morphology of tumor has fractal char acteristics,so this paper introduces the method of fractal to distinguish the benign and malignant breast tumors and make the fractal dimension as the index.As a s cience and engineering software,MATLAB is very powerful.It can combine the n umerical analysis,image processing and programming technology together.And it i s very useful for binaryzation images analysis.The image was preprocessed using graphics processing function of MATLAB software,including gray-scale transforma tion,binaryzation processing,image counter,morphological treatment,region growing method to extrac tumor boundaries.We calculate the fractal dimension ofl esionar eas using the fractal analysis function of fraclab toolbox.T-test were used to analyze the fractal dimension of the lesionin SPSS. When the P value is less than 0.05,it is statistically significant.Results:The average of fractal dimension in ultrasound images of breast cancer p atients was:1.96±0.01;The average of fractal dimension in ultrasound images of breast fibroadenoma patients was:1.93±0.02;The different was statistically signific ant(P<0.05).Conclusion:1.With the characteristics of simple and quick,we can use the functio n of MATLAB software to extract tumor margins,however it was limited in the t umors with massive adhesion of surrounding tissues. 2.The fractal dimension in ultrasound images of breast cancer patients was higher than that in breast fibroadenoma patients.It can visually reflex the future of histo logical texture and therefore we can distinguish benighand malignant tumor by o bserving the fractal dimension of its ultrasound images.