Inter-Slice Super-Resolution of Magnetic Resonance Images by Pre-Training and Self-Supervised Fine-Tuning

In clinical practice, 2D magnetic resonance (MR) sequences are widely adopted. While individual 2D slices can be stacked to form a 3D volume, the relatively large slice spacing can pose challenges for both image visualization and subsequent analysis tasks, which often require isotropic voxel spacing. To reduce slice spacing, deep-learning-based super-resolution techniques are widely investigated. However, most current solutions require a substantial number of paired high-resolution and low-resolution images for supervised training, which are typically unavailable in real-world scenarios. In this work, we propose a self-supervised super-resolution framework for inter-slice super-resolution of MR images. Our framework is first featured by pre-training on video dataset, as temporal correlation of videos is found beneficial for modeling the spatial relation among MR slices. Then, we use public high-quality MR dataset to fine-tune our pre-trained model, for enhancing awareness of our model to medical data. Finally, given a target dataset at hand, we utilize self-supervised fine-tuning to further ensure our model works well with user-specific super-resolution tasks. The proposed method demonstrates superior performance compared to other self-supervised methods and also holds the potential to benefit various downstream applications.

1. INTRODUCTION

Magnetic resonance (MR) imaging is widely used for its non-invasive and detailed visualization of the human body. In clinical MR imaging, 2D sequences are widely employed for faster image acquisition, where multiple 2D slices can be stacked to create a 3D volume. Such volume typically has large inter-slice spacing, in contrast to the fine-grained intra-slice spacing [1]. The anisotropic voxel spacing in 3D can pose significant challenges for many automatic image processing software requiring isotropic voxel spacing as input.