Millard Charles, Chiew Mark
the Wellcome Centre for Integrative Neuroimaging, FMRIB, Nuffield Department of Clinical Neurosciences, University of Oxford, OX3 9DU Oxford, U.K.
the Wellcome Centre for Integrative Neuroimaging, FMRIB, Nuffield Department of Clinical Neurosciences, University of Oxford, OX3 9DU Oxford, U.K., and with the Department of Medical Biophysics, University of Toronto, Toronto, ON M5S 1A1, Canada, and also with the Canada and Physical Sciences, Sunnybrook Research Institute, Toronto, ON M4N 3M5, Canada.
IEEE Trans Comput Imaging. 2023 Jul 26;9:707-720. doi: 10.1109/TCI.2023.3299212.
In recent years, there has been attention on leveraging the statistical modeling capabilities of neural networks for reconstructing sub-sampled Magnetic Resonance Imaging (MRI) data. Most proposed methods assume the existence of a representative fully-sampled dataset and use fully-supervised training. However, for many applications, fully sampled training data is not available, and may be highly impractical to acquire. The development and understanding of self-supervised methods, which use only sub-sampled data for training, are therefore highly desirable. This work extends the Noisier2Noise framework, which was originally constructed for self-supervised denoising tasks, to variable density sub-sampled MRI data. We use the Noisier2Noise framework to analytically explain the performance of Self-Supervised Learning via Data Undersampling (SSDU), a recently proposed method that performs well in practice but until now lacked theoretical justification. Further, we propose two modifications of SSDU that arise as a consequence of the theoretical developments. Firstly, we propose partitioning the sampling set so that the subsets have the same type of distribution as the original sampling mask. Secondly, we propose a loss weighting that compensates for the sampling and partitioning densities. On the fastMRI dataset we show that these changes significantly improve SSDU's image restoration quality and robustness to the partitioning parameters.
近年来，人们开始关注利用神经网络的统计建模能力来重建欠采样的磁共振成像（MRI）数据。大多数提出的方法都假设存在一个具有代表性的全采样数据集，并使用全监督训练。然而，对于许多应用来说，全采样训练数据是不可用的，而且获取起来可能非常不切实际。因此，非常需要开发和理解仅使用欠采样数据进行训练的自监督方法。这项工作将最初为自监督去噪任务构建的Noisier2Noise框架扩展到可变密度欠采样的MRI数据。我们使用Noisier2Noise框架来分析性地解释通过数据欠采样进行自监督学习（SSDU）的性能，SSDU是一种最近提出的方法，在实践中表现良好，但到目前为止缺乏理论依据。此外，我们提出了SSDU的两种改进，这是理论发展的结果。首先，我们建议对采样集进行划分，以便子集具有与原始采样掩码相同类型的分布。其次，我们提出一种损失加权方法，以补偿采样和划分密度。在fastMRI数据集上，我们表明这些变化显著提高了SSDU的图像恢复质量和对划分参数的鲁棒性。
