Introduction: Chen et al. proposed a two-stage randomized basket design, which tests l=1 treatment vs control across various cancer indications, We expand this design to evaluate l>1 treatments vs a common control. Methods: We extended Chen's Type I error rates and power calculations. Our design allocates samples evenly to each basket, with additional samples added strategically at the second stage. A basket is pruned only if none of its treatments are effective at a predefined alpha level during the interim. We incorporated the Dunnett test to allow for the pruning of underperforming treatments while pooling effective ones, adjusting the covariance matrix based on Dunnett's findings. Simulations were used to validate our calculation results. Results: Our validation results show that for l=1 treatment, our result of type I error is very similar to Chen’s design. For l>1, we derived a reasonable result using a generalized computational equation with validated through simulation.