In this study, we apply nnU-Net, a U-Net based neural network approach, to segment the lesions from brain magnetic resonance imaging (MRI). The proposed approach segments lesions that occur in neonatal patients with hypoxic ischemic encephalopathy (HIE). The nnU-Net is trained using the skull-stripped apparent diffusion coefficient (ss-ADC) MRI and z-score apparent diffusion coefficient (Z-ADC) maps provided by the BONBID-HIE2023 segmentation challenge organizers. A total of 85 pairs of ADC and Z-ADC image volumes were used for the training, and no datasets other than the ones shared by challenge organizers were used. We used the default configuration of the nnU-Net with removing the header information from the image datasets prior to training. The trained neural network was evaluated using the online platform by the challenge organizers on validation and test sets consisting of 4 and 44 datasets, respectively. The proposed method yielded Dice scores of \(0.6735 \pm 0.2169\) and \( 0.4998 \pm 0.2809\) for validation and test sets, respectively.