In this paper, we present a novel deep learning model built upon ResUNet++ for brain tumor segmentation. The objective of this study is to segment the tumor core (TC), whole tumor (WT), and enhancing tumor (ET) using the BRATS 2020. Brain tumor segmentation is important for frequently monitoring the tumor change over time, planning effective therapeutic approaches, and devising surgical maneuvers. We include additional convolution layers in the residual block which improves the basic ResUNet++ model. These additional layers capture deeper features of the images and allow for better flow of gradient through skip (residual) connections during backpropagation. To enable better segmentation, we need to better understand the dependence of spatially adjacent pixels which can be achieved by adding further convolutional layers. This in depth analysis of spatial data improves the segmentation process’s accuracy. The deep residual block is employed in both the decoder and encoder paths, boosting the overall performance of the network model. We integrate the Nadam optimizer and a binary loss function to manage noisy gradients and pixel-wise classification, respectively. The Dice score was employed to assess the effectiveness of the proposed model, revealing superior performance compared to existing state-of-the-art approaches.