The Segmentation Of Mammographic Image Based-on Wavelet Transform

Breast tumour is one of the most common diseases of women. Breast cancer is the top cause of women death. Researchers have been paying more and more attention on the diagnosis of breast tumour and have found that early finding, diagnosis and curing is very important to the caner.At present, the best way of detecting breast tumour is mammographic image. It is the only one method that can early detect breast cancer without any symptom.There are two characters to judge the beast cancer. One is the tumours, and the other is the calcification.The shapes of tumours are very different and the edge of tumours are very blur. The calcifications are also in deferent size and shape. So the segmentation of tumour and calcification are very difficult.Wavelet transform is a local time-frequency transform. It can extract useful information from signal, can analysis signal in multiscale, and just fulfils the segmentation demands of mammographic image.So in this dissertation, I designed a ID wavelet method to segment the tumours in mammographic image, and designed a 2D wavelet method to detect calcifications. The method used in tumour segmentation comes from the modifying of Olivo digital image segmentation method. First, calculating the histogram of the mammographicimage. Then running wavelet transform on the histogram, so as to find the thresholds of the image. Finally, segmenting the mammographic image using the thresholds. This method realized the purpose of segmenting mammographic image in deferent scales. I designed the program realizing the method, and through experiment, we get satisfying results.The method used in calcification detecting is a new method designed in this dissertation. This method transforms the 2D space field image signal into wavelet fleld signal. Then using the edge-detecting operator to deal with the wavelet coefficients. At last, using inverse wavelet transform to get the calcification in mammographic image. I designed the program realizing the method, and through experiment, we get a satisfying result.