Comparative Analysis of ResNet50V2 and ConvNeXt-Tiny for Breast Cancer Histopathological Image Classification

Breast cancer is one of the primary causes of cancer-related deaths among women globally. Timely and accurate classification of histopathological images is critical for improving patient outcomes. This study conducts a comparative evaluation of two deep learning architectures—ResNet50V2 and ConvNeXt-Tiny—for binary classification of breast histopathology images into benign and malignant categories. The models were trained and evaluated on a large, balanced dataset containing over 570,000 images, divided into training, validation, and test sets. Performance was assessed using standard metrics including accuracy, precision, recall, F1-score, ROC-AUC, and PR-AUC. To enhance interpretability, Grad-CAM visualizations were employed to highlight class-discriminative regions. Results indicate that although both models reach similar accuracy (~91%), ConvNeXt-Tiny exhibits slightly better precision, while ResNet50V2 demonstrates marginally improved recall, reflecting their different sensitivities. These insights may help guide model selection in clinical contexts where the balance between false positives and false negatives is critical.