Multivariate Time Series, that is, repeated measuring of data at equal intervals, is a widely used technique in the medical field for treating ailments such as Brain tumors. Capturing anomalies in multivariate time series data is, however, tough. It is equally important to detect a brain tumor in its early stage before it reaches a state of no return. The tumor grows speedily in the brain & its size doubles in just 25 days. Thus, a novel approach has been proposed to detect and locate anomalies in Brain MR Images through Cross-Correlation Encoder Decoder GAN (CC-EDGAN). This methodology combines Generative Adversarial Networks (GAN) & Autoencoders and can gather correlation and time based features of Multivariate Time Series. To date, this method has only been implemented on Industrial data, which makes its implementation in the healthcare sector a novelty. Different performance metrics have been chosen to test validity of the algorithm that has been proposed. The results show that the proposed CC EDGEAN model outperforms the traditional F-AnoGAN model on all the performance metrics to evaluate the proposed work, including Peak Signal-to-Noise Ratio (PSNR), Mean Absolute Error (MAE), Structural Similarity Index (SSIR), Matthews Correlation Coefficient (MCC), and the F1-Score when test sets are compared. Specifically, the proposed system achieves a percentage improvement of 26.61% on MAE, 15.40% on PSNR, 3.13% on SSIR, 20.35% on MCC, and 22.81% on F1 score when test set is compared, which indicates its potential to improve patient outcomes by enabling early diagnosis and treatment.