Zhang Yuting, Zhou Wenjun, Huang Lijie, Shao Yongjie, Luo Anguo, Luo Jianwen, Peng Bo
IEEE Trans Ultrason Ferroelectr Freq Control. 2024 Dec;71(12: Breaking the Resolution Barrier in Ultrasound):1714-1734. doi: 10.1109/TUFFC.2024.3424955. Epub 2025 Jan 8.
Ultrasound localization microscopy (ULM), an emerging medical imaging technique, effectively resolves the classical tradeoff between resolution and penetration inherent in traditional ultrasound imaging, opening up new avenues for noninvasive observation of the microvascular system. However, traditional microbubble tracking methods encounter various practical challenges. These methods typically entail multiple processing stages, including intricate steps such as pairwise correlation and trajectory optimization, rendering real-time applications unfeasible. Furthermore, existing deep learning-based tracking techniques neglect the temporal aspects of microbubble motion, leading to ineffective modeling of their dynamic behavior. To address these limitations, this study introduces a novel approach called the gated recurrent unit-based multitasking temporal neural network (GRU-MT). GRU-MT is designed to simultaneously handle microbubble trajectory tracking and trajectory optimization tasks. In addition, we enhance the nonlinear motion model initially proposed by Piepenbrock et al. to better encapsulate the nonlinear motion characteristics of microbubbles, thereby improving trajectory tracking accuracy. In this study, we perform a series of experiments involving network layer replacements to systematically evaluate the performance of various temporal neural networks, including recurrent neural network (RNN), long short-term memory network (LSTM), GRU, Transformer, and its bidirectional counterparts, on the microbubble trajectory tracking task. Concurrently, the proposed method undergoes qualitative and quantitative comparisons with traditional microbubble tracking techniques. The experimental results demonstrate that GRU-MT exhibits superior nonlinear modeling capabilities and robustness, both in simulation and in vivo dataset. In addition, it achieves reduced trajectory tracking errors in shorter time intervals, underscoring its potential for efficient microbubble trajectory tracking. The model code is open-sourced at https://github.com/zyt-Lib/GRU-MT.
超声定位显微镜（ULM）是一种新兴的医学成像技术，它有效地解决了传统超声成像中分辨率和穿透深度之间固有的经典权衡问题，为微血管系统的无创观察开辟了新途径。然而，传统的微泡跟踪方法面临各种实际挑战。这些方法通常需要多个处理阶段，包括成对相关和轨迹优化等复杂步骤，使得实时应用变得不可行。此外，现有的基于深度学习的跟踪技术忽略了微泡运动的时间方面，导致对其动态行为的建模无效。为了解决这些限制，本研究引入了一种名为基于门控循环单元的多任务时间神经网络（GRU-MT）的新方法。GRU-MT旨在同时处理微泡轨迹跟踪和轨迹优化任务。此外，我们改进了最初由皮彭布罗克等人提出的非线性运动模型，以更好地封装微泡的非线性运动特征，从而提高轨迹跟踪精度。在本研究中，我们进行了一系列涉及网络层替换的实验，以系统地评估各种时间神经网络，包括递归神经网络（RNN）、长短期记忆网络（LSTM）、GRU、Transformer及其双向对应网络，在微泡轨迹跟踪任务上的性能。同时，将所提出的方法与传统微泡跟踪技术进行定性和定量比较。实验结果表明，GRU-MT在模拟数据集和体内数据集上均表现出卓越的非线性建模能力和鲁棒性。此外，它在更短的时间间隔内实现了轨迹跟踪误差的降低，突出了其在高效微泡轨迹跟踪方面的潜力。该模型代码可在https://github.com/zyt-Lib/GRU-MT上开源获取。
