Title: Machine Learning aided k-t SENSE for fast reconstruction of highly accelerated PCMR data Show Chinese title

Title: 基于机器学习的k-t SENSE方法用于快速重建高度加速的PCMR数据

Abstract: Purpose: We implemented the Machine Learning (ML) aided k-t SENSE reconstruction to enable high resolution quantitative real-time phase contrast MR (PCMR). Methods: A residual U-net and our U-net M were used to generate the high resolution x-f space estimate for k-t SENSE regularisation prior. The networks were judged on their ability to generalise to real undersampled data. The in-vivo validation was done on 20 real-time 18x prospectively undersmapled GASperturbed PCMR data. The ML aided k-t SENSE reconstruction results were compared against the free-breathing Cartesian retrospectively gated sequence and the compressed sensing (CS) reconstruction of the same data. Results: In general, the ML aided k-t SENSE generated flow curves that were visually sharper than those produced using CS. In two exceptional cases, U-net M predictions exhibited blurring which propagated to the extracted velocity curves. However, there were no statistical differences in the measured peak velocities and stroke volumes between the tested methods. The ML aided k-t SENSE was estimated to be ~3.6x faster in processing than CS. Conclusion: The ML aided k-t SENSE reconstruction enables artefact suppression on a par with CS with no significant differences in quantitative measures. The timing results suggest the on-line implementation could deliver a substantial increase in clinical throughput. Show Chinese abstract

Abstract: 目的：我们实现了机器学习（ML）辅助的k-t SENSE重建，以实现高分辨率定量实时相位对比磁共振（PCMR）。 方法：使用残差U-net和我们的U-net M来生成k-t SENSE正则化之前的高分辨率x-f空间估计。 这些网络根据其对真实欠采样数据的泛化能力进行评估。 体内验证是在20个实时18x前瞻性欠采样GAS扰动PCMR数据上进行的。 将ML辅助的k-t SENSE重建结果与自由呼吸的笛卡尔回顾性门控序列和相同数据的压缩感知（CS）重建进行比较。 结果：总体而言，ML辅助的k-t SENSE生成的流量曲线在视觉上比使用CS生成的更清晰。 在两个例外情况下，U-net M的预测出现了模糊，并传播到了提取的速度曲线中。 然而，测试方法之间的测量峰值速度和搏出量没有统计学差异。 估计ML辅助的k-t SENSE处理速度比CS快约3.6倍。 结论：ML辅助的k-t SENSE重建在伪影抑制方面与CS相当，定量测量没有显著差异。 时间结果表明，在线实现可以显著提高临床吞吐量。