Medical Imagery for classification has over the years been completely dependent on existing datasets which are certainly limited or not enough for obtaining the best results. Therefore, the traditional approach for increasing the volume of the dataset has been Data Augmentation which can be limited in its ability to create new data points and may introduce bias into the training dataset. Impact of Generative Adversarial Networks (GANs) in medical imaging has been extremely promising, especially generation, segmentation, and classification. This project employs Generative Adversarial Networks to improve the performance of classification models by creating synthetic datasets for the dataset of interest. GANs can overcome the limitations of data augmentation by generating new data points that are more realistic, complex, and representative of the original dataset. In our solution, the classification models are trained on the generated synthetic data using GANs to detect whether the medical image provided is malignant or benign. Through this effort we additionally want to substantiate utilization of synthetic data results in better classification.