A Study on the Combination of Image Preprocessing Method Based on Texture Feature and Segmentation Algorithm for Breast Ultrasound Images

Breast cancer is the most common cancer in women. Obtaining the tumor part of breast ultrasound image is of great significance for medical assistance. Breast ultrasound images have the characteristics of variable tumor morphology, more shadows, and blurred borders. Therefore, image preprocessing is usually required before segmentation. However, the traditional image preprocessing method is difficult to effectively distinguish the tumor area and tissue shadow in the ultrasound image, which affects the segmentation result of the tumor part. Therefore, this paper proposes an ultrasound image preprocessing method based on texture features. First, extract the different texture feature images of the breast ultrasound image, and then concatenate the original image and two different texture feature images together to form a new 3-channel RGB image. In the experiment, combining different preprocessing methods and different segmentation methods, the segmentation results of breast ultrasound images are evaluated. Compared with traditional preprocessing methods, the preprocessing methods proposed in this paper have improved in all segmentation evaluation indexes. The experimental results show that the Intersection-over-Union (IoU) and Dice-Similarity-Coefficient (DSC) increased to 0.6022 and 0.7554 respectively.