Study of the impact of image preprocessing approaches on the segmentation and classification of breast lesions on ultrasound
Arturo Rodríguez-Cristerna, C. P. Guerrero-Cedillo, G. A. Donati-Olvera, Wilfrido Gómez‐Flores, Wagner Coelho de Albuquerque Pereira · 2017
This paper presents a study of the impact of image preprocessing techniques on the segmentation and classification of breast lesions on ultrasound. Commonly, image preprocessing performs contrast enhancement and speckle reduction. In this sense, five contrast enhancement techniques and four despeckling methods were combined to generate 20 different image preprocessing schemes. The experiments considered 1,021 breast ultrasound images (766 benign and 255 malignant lesions). The results revealed that fuzzy enhancement followed by the interference-based speckle filter produced adequate lesion segmentation quality, in terms of the Jaccard index (0.85 ± 0.05), and acceptable classification performance in benign and malignant classes, in terms of the area under the ROC curve (0.82 ± 0.02).