Characterization of Breast Fibroadenoma in U/S Images Using Image Texture Analysis Techniques

A new approach to texture characterization from breast ultrasound scan is presented. This study was aimed to use the texture analysis and classification methods to characterize the breast Fibroadenoma, cyst and normal breast regions in U/S images using image processing program (IDL, interactive data language). An analytical case control study of Breast US images of 130 adult subjects with Fibroadenoma, cyst and normal images that in which used as entry data. Tiff format was created as IDL variables and then using 3x3 window the image was scanned and based on the image histogram the selected feature also called FOS was calculated using this window. Linear discriminant analysis was used for the tissue classification. The study found that the Fibroadenoma texture reveal a different underlying pattern compared to the cyst and normal breast tissues with classification sensitivity and specificity 100 % and 92.5 % respectively for fibroadenoma, and the combination of the texture features throughout the different U/S image phases provide the highest predictive overall accuracy of 95 % using linear discriminant analysis technique.