在眾多類型的乳癌診斷程序中，超音波診斷是一種最為人接受且不會對人體產生副作用的科技，而且由於超音波掃描儀的低價位、便利性、廣泛性、及時性，使得超音波腫瘤影像成為臨床上醫生初次診斷乳癌的利器。然而，由於超音波影像中常含有大量的斑點、雜訊及組織紋理，容易因為操作超音波掃描儀的醫師訓練程度不同而產生不同的結果。所以在本論文中，利用超音波腫瘤影像的特性，採用分水嶺轉換(watershed transform)進行影像的切割，將腫瘤影像分割成不同屬性的區域，再利用區域影像屬性之差異，將相同屬性的區域加以整併，準確切割出腫瘤的輪廓，使找到的輪廓可以近似經驗豐富的醫師所認定的輪廓，由實驗結果得知，此方法描繪出的腫瘤輪廓與醫生手繪的腫瘤輪廓近似程度相當高。

In the many kind of breast tumor diagnosis method, medical ultrasound is a safety, convenience, and instance technology. Unfortunately, the digital ultrasonic image always comprises speckles, noise, and tissue-related textures, most traditional segmentation techniques for the ultrasonic images do not perform well. Automatic contour segmentation for the breast tumors in the ultrasonic images may assist physicians without experience in making a correct diagnosis. In this paper, we combined watershed transformation and region growing techniques to detect the contour of breast tumors in the ultrasonic images. Computer simulation results show that the proposed method always identified the similar contour with the manual sketch of the breast tumor in the ultrasonic images.