Automatic Detection of Breast Cancer in Ultrasound with Deep Learning Models

Recently, the occurrence of the breast-cancer (BC) in women community is rising due to various causes and it is the reason for one of the cause for a countries' cancer burden. The clinical level detection of the BC is performed using images of a chosen modality and Breast Ultrasound Imaging (BUI) is one of the common and frequently used imaging scheme due to is safety and cost. The BC detection outcome of BUI can be considered to plan and execute the necessary treatment to cure the disease. This work proposed a Deep-Learning (DL) based method using ConvNeXt (CN) variants and the achieved results for SoftMax classifier is presented and evaluated. The stages in developed system includes; image collection and resizing, feature extraction with CN-model, and classification and performance evaluation using 3-fold cross validation. The experimental outcome of this study confirms that the CN-variants based detection of BC achieves an accuracy >91%.