The task is to perform regression on the simulated tumor growth dataset, obtained from the Department of Mathematics at UC, Irvine. The dataset contains a set of simulations each associated with a pair of targets that jointly control the growth of tumors. Two different approaches were chosen in extracting the feature set from the images. The first approach is extracting various types of features i.e., image features and shape features. The second approach is similar to the Bag of Words approach where a subset of shapes is computationally chosen that is sufficient to describe all the shapes we observe in the dataset. Thus, for a given simulation, the first approach gives a vector that contains the values of each feature extracted; where as the second approach gives a vector of counts that specifies the number of shapes in the simulation that are similar to the shapes we have chosen in the subset. We then use various available regression models to predict the two targets separately. Finally, to account for the joint effect of the targets in the growth, we treat the regression as a multi-task problem with the a-priori that there could be a non-zero correlation across the tasks. In this effort, we use Multi-task Gaussian Process to perform joint-regression. We have been fairly successful in our efforts and we look forward to working with real data from the NIH as the project's next phase.