Comparative Analysis of Inception V3 and Xception Models for Breast Ultrasound Image Classification

Breast cancer is one of the most prevalent cancers globally and a leading cause of death among women. Early detection is crucial in reducing mortality rates, and ultrasound screening is particularly beneficial for women under 35, where mammograms can be less effective due to dense breast tissue. This study applied deep learning models, specifically Inception V3 and Xception, to classify ultrasound breast images into three categories: malignant, benign, and normal. The results unveiled that the Inception V3 model outperformed Xception in terms of both training and testing accuracy. Inception V3 achieved an average accuracy of 92% with a precision of 84%, a sensitivity of 95%, a specificity of 91%, and an F-score of 89%. In comparison, Xception demonstrated lower performance, with an average accuracy of 76%. These findings suggest that Inception V3 is a more reliable model for classifying breast cancer in ultrasound images, potentially aiding in earlier and more accurate diagnoses.