CRIS: Cross-Plane Self-Supervised Isotropic Restoration for Multi-Modality Volumetric Imaging

Abstract: Anisotropic volumetric acquisitions are a significant bottleneck in clinical and biological imaging, where hardware constraints often lead to blurred through-plane resolution and structural artifacts. This work introduces CRIS, a self-supervised cross-plane framework that restores isotropic volumes from anisotropic acquisitions without requiring paired high-resolution training data. Extensive evaluations demonstrate that CRIS currently sets a new benchmark in the field, consistently outperforming all existing state-of-the-art methods across multiple performance metrics. CRIS provides a robust solution for high-quality volumetric restoration across diverse modalities, including MRI and Volume Electron Microscopy.