A Comparative Experimental Study of DenseNet121, EfficientNetB7, MobileNetV2, and ConvNeXTV2 for Breast Cancer Classification

Breast cancer remains a significant global health challenge, necessitating advanced diagnostic tools to improve early detection and patient outcomes. This study conducts a comparative experimental analysis of four convolutional neural network (CNN) architectures—DenseNet121, EfficientNetB7, MobileNetV2, and ConvNeXTV2—for classifying breast cancer as benign or malignant using mammographic images. Leveraging the Digital Database for Screening Mammography (DDSM), we evaluate these models based on accuracy, precision, recall, F1-score, and computational efficiency. EfficientNetB7 achieved the highest accuracy (94.2%), while MobileNetV2 offered the best trade-off between performance and efficiency, with an accuracy of 90.5% and the lowest inference time (12.1 ms). DenseNet121 and ConvNeXTV2 provided intermediate results, with accuracies of 91.8% and 93.0%, respectively. These findings highlight the strengths and limitations of each model, offering insights into their applicability in clinical settings.