AI-Based Breast Cancer X-Ray Image Detection Using Generative Adversarial Attacks
V. S. R. K. Raju Dandu, R. N. V. Jagan Mohan, M. Chandra Naik · 2024
Breast cancer is one type of cancer that disproportionately affects women. Mammograms are X-ray scans that doctors use to identify breast cancer. Even though AI is quite good at identifying false photographs, some of them can be so convincing that they lead to the wrong diagnosis of cancer. AI-powered technologies have the potential to improve the accuracy of cancer detection. Increasing the resilience of AI models to harmful attacks is critical. Models are trained to identify and steer clear of purposefully antagonistic false pictures using adversarial training. A study found that simulated attacks can confuse both AI systems for detecting breast cancer and human radiologists, putting medical AI at risk. It is critical to investigate how AI models respond to hostile attacks in order to guarantee security and robustness. By leveraging mammography imaging data, the study developed a deep learning strategy for breast cancer identification, improving AI’s response to intricate adversarial attacks. The system constructed accurate images of benign and malignant illnesses using generative adversarial networks (GANs). This experimental research maps intricate relationships, records long-term temporal linkages, and creates synthetic time-series data for healthcare cancer datasets using GANs. In order to find patterns in the data, it also uses mode collapse and data analysis. Principle Component Analysis (PCA) and other data visualisation techniques are crucial for improving understanding of the relationships between the variables.