Ultra-high-field MRI (UHF-MRI) at 7T is growing increasingly important for clinical diagnostics and research. However, UHF-MRI faces more challenges from susceptibility artifacts and B0-inhomogeneities, impacting segmentation procedures that are typically optimized for lower field strengths as UHF-MRI is not yet widely available across radiological sites. In this study, we assessed therefore, whether MONAI (The Medical Open Network for Artificial Intelligence), an open source framework for AI-based medical image segmentation, could be employed for accurately segmenting also UHF-MRI data as MONAI is trained on 3T datasets exhibiting different contrast behavior than 7T. For testing, we used a publicly accessible 7T-brain MRI dataset and compared the AI-based segmentations with segmentations yielded by the framework statistical parametric mapping (SPM). The segmentation accuracy was assessed through comparison of the mean balanced Hausdorff-Distance (bHD) with expert-level segmented ground truths. Statistical analysis (left-sided Wilcoxon signed-rank test) indicated that cortical and deep gray matter were more accurately segmented by MONAI than by SPM (mean bHD MONAI: 2.65x10-4 ; mean bHD SPM: 3.37x10-4). The findings demonstrate MONAI's ability to segment data more effectively than the standard functional MRI analysis tool SPM. This makes newest technology available for researchers not trained specifically in AI-based image segmentation.