Trenti Chiara, Ylipää Erik, Ebbers Tino, Carlhäll Carl-Johan, Engvall Jan, Dyverfeldt Petter
Department of Health, Medicine and Caring Sciences (HMV), Linköping University, Linköping, Sweden; Center for Medical Image Science and Visualization (CMIV), Linköping, Sweden.
Analytic Imaging Diagnostics Arena (AIDA), Linköping University, Linköping, Sweden.
J Cardiovasc Magn Reson. 2025 Jun 2;27(2):101920. doi: 10.1016/j.jocmr.2025.101920.
Despite its potential to improve the assessment of cardiovascular diseases, four-dimensional (4D) flow cardiovascular magnetic resonance (CMR) is hampered by long scan times. 4D flow CMR is conventionally acquired with three motion encodings and one reference encoding, as the three-dimensional velocity data are obtained by subtracting the phase of the reference from the phase of the motion encodings. In this study, we aim to use deep learning to predict the reference encoding from the three motion encodings for cardiovascular 4D flow.
A U-Net was trained with adversarial learning (U-Net) and with a velocity frequency-weighted loss function (U-Net) to predict the reference encoding from the three motion encodings obtained with a non-symmetric velocity-encoding scheme. Whole-heart 4D flow datasets from 126 patients with different types of cardiomyopathies were retrospectively included. The models were trained on 113 patients with a 5-fold cross-validation, and tested on 13 patients. Flow volumes in the aorta and pulmonary artery, mean and maximum velocity, total and maximum turbulent kinetic energy at peak systole in the cardiac chambers and main vessels were assessed.
Three-dimensional velocity data reconstructed with the reference encoding predicted by deep learning agreed well with the velocities obtained with the reference encoding acquired at the scanner for both models. U-Net performed more consistently throughout the cardiac cycle and across the test subjects, while U-Net performed better for systolic velocities. Comprehensively, the largest error for flow volumes, maximum and mean velocities was -6.031% for maximum velocities in the right ventricle for the U-Net, and -6.92% for mean velocities in the right ventricle for U-Net. For total turbulent kinetic energy, the highest errors were in the left ventricle (-77.17%) for the U-Net, and in the right ventricle (24.96%) for the U-Net, while for maximum turbulent kinetic energy were in the pulmonary artery for both models, with a value of -15.5% for U-Net and 15.38% for the U-Net.
Deep learning-enabled referenceless 4D flow CMR permits velocities and flow volumes quantification comparable to conventional 4D flow. Omitting the reference encoding reduces the amount of acquired data by 25%, thus allowing shorter scan times or improved resolution, which is valuable for utilization in the clinical routine.
尽管四维（4D）血流心血管磁共振成像（CMR）有改善心血管疾病评估的潜力，但扫描时间长限制了其应用。传统的4D血流CMR通过三次运动编码和一次参考编码来获取，因为三维速度数据是通过从运动编码的相位中减去参考相位得到的。在本研究中，我们旨在利用深度学习从三次运动编码中预测心血管4D血流的参考编码。
使用对抗学习训练一个U-Net（U-Net-AL）和一个速度频率加权损失函数训练的U-Net（U-Net-VFW），以从非对称速度编码方案获得的三次运动编码中预测参考编码。回顾性纳入了126例不同类型心肌病患者的全心4D血流数据集。模型在113例患者上进行5折交叉验证训练，并在13例患者上进行测试。评估了主动脉和肺动脉的血流量、平均和最大速度、心室和主血管收缩期末的总湍流动能和最大湍流动能。
对于两个模型，用深度学习预测的参考编码重建的三维速度数据与在扫描仪上获取的参考编码得到的速度非常吻合。U-Net-AL在整个心动周期和所有测试对象中表现更一致，而U-Net-VFW在收缩期速度方面表现更好。综合来看，U-Net-AL在右心室最大速度方面血流量、最大和平均速度的最大误差为-6.031%，U-Net-VFW在右心室平均速度方面的最大误差为-6.92%。对于总湍流动能，U-Net-AL在左心室的最高误差为-77.17%，U-Net-VFW在右心室的最高误差为24.96%，而对于最大湍流动能，两个模型在肺动脉中的误差最大，U-Net-AL为-15.5%，U-Net-VFW为15.38%。
基于深度学习的无参考4D血流CMR能够实现与传统4D血流相当的速度和血流量量化。省略参考编码可将采集的数据量减少25%，从而允许更短的扫描时间或更高的分辨率，这在临床常规应用中具有重要价值。
