Revolutionizing Breast Cancer Diagnosis: VGG16's Breakthrough in Histopathological Image Classification
Eshika Jain, Amanveer Singh · 2024
Breast cancer is still the leading cause of cancer mortality among ladies, thereby demanding reliable and effective diagnostic tools. Conventionally, these tools of diagnosis include mammography, ultrasound, and histopathology, which have been quite valuable but are often limited by a number of problems such as variability in interpretation and dependence on specialized expertise. Deep learning has recently had some exciting developments that could provide very promising alternatives through automated or even objective analysis of medical images, especially in the field of computer vision. This research study presents the application of the VGG16 architecture in one of the well-established CNNs for classifying breast cancer histopathological images. In this study, the VGG16 model that was pre-trained on the ImageNet dataset and fine-tuned with data augmentation techniques has huge potential to distinguish between benign and malignant cases of breast cancer. The model delivered an ROC-AUC score of 0.87337 and an accuracy of 0.82942, thus performing well in classifying breast cancer images. Meanwhile, the number of false negatives was as high as 107, indicating that more efforts are needed to improve the detection rate further and reduce the missing rate. These results suggest that the VGG16-based approach has possible application as an automated tool for the diagnosis of breast cancer, and its eventual integration into such clinical workflows and tasks will be provided to enhance its diagnostic accuracy and patient outcomes. Future research will be directed at optimization of model parameters and further testing on more diversified datasets to improve clinical applicability.