High Efficient Reconstruction of Single-Shot Magnetic Resonance T_2 Mapping Through Overlapping Echo Detachment and DenseNet

摘要

Rapid and quantitative magnetic resonance T_2 imaging plays an important role in medical imaging field. However, the existing quantitative T_2 mapping method are usually time-consuming and sensitive to motion artifacts. Recently, a novel single-shot quantitative parameter mapping method based on overlapped-echo detachment technique has been proposed by us, but an efficient reconstruction algorithm is necessary. In this paper, a multi-stage DenseNet was utilized to reconstruct single-shot T_2 mapping efficiently. The contributions of the paper mainly include the following aspects. First, an end-to-end neural network is proposed, which can directly obtain the reconstructed images without any secondary processing. Second, DenseNet was introduced into the reconstruction network to better reuse the features. Third, a weighted Euclidean loss function is proposed, which can be better used for image reconstruction.