Detection and injury extent of neonatal hypoxic-ischemic encephalopathy (HIE) is a major challenge in neonatal medicine all across the world. This paper explores the applications of 3D based deep learning (DL) models on high-resolution diffusion based MRI images on the Boston Neonatal Brain Injury Dataset for Hypoxic Ischemic Encephalopathy (BONBID-HIE). We propose an ensemble method along with a hybrid loss function that consists of combining six different variants of UNet architecture and three different loss functions respectively. Our findings reveal that the ensemble approach, combining different architectures, outperforms single models, leading to improved evaluation metrics. Specifically, in this case, 0.5800 ± 0.2557 dice score was achieved using the ensemble approach. These results show the importance of tailored DL techniques in precisely segmenting HIE lesions revealing its extent and lay groundwork for future work to fine- tune models as well as the proposed ensemble approach.