Performance Evaluation of Efficientnetv2 Models on the Classification of Histopathological Benign Breast Cancer Images

In the field of breast cancer diagnosis, the precise classification of benign images plays a pivotal role in ensuring effective patient care. This research undertakes a detailed examination of EfficientNetV2 models, specifically focusing on their ability to discern benign histopathology breast cancer images. The dataset were carefully curated to include diverse benign cases such as adenosis, fibroadenoma, phyllodesₜumor, and tubularₐdenoma of image level for 40X magnification factor underwent thorough preprocessing before being divided into training and testing sets. Various variants of the EfficientNetV2 model—EfficientNetV2B0, EfficientNetV2B1, EfficientNetV2B2, EfficientNetV2B3, EfficientNetV2S, EfficientNetV2M, and EfficientNetV2L—were trained on the designated dataset. The performance evaluation shows the intricacies of the efficiency of each model. Notably, EfficientNetV2L emerged as a standout performer, boasting impressive metrics such as Accuracy (0. 97), Precision (0. 97), Recall (0. 97), F1-score (0. 97). These findings underscore the potential of EfficientNetV2L as a robust tool for accurately discerning benign histopathology breast cancer images. This study contributes significant insights to the field of breast cancer diagnostics, particularly addressing the critical task of classifying benign cases accurately. The gained insights pave the way for improved decision-making in assessments, ultimately enhancing the overall efficacy of breast cancer diagnosis.