Fehrenbach Jérôme, Masmoudi Mohamed, Melodelima David
Institut de Mathématiques de Toulouse, 31062 Toulouse cedex 9, France.
Ultrasonics. 2010 Jan;50(1):44-51. doi: 10.1016/j.ultras.2009.07.009. Epub 2009 Jul 24.
This study presents a contribution to the tracking of a moving target during high-intensity focused ultrasound (HIFU) treatment. Indeed, HIFU has proved to be highly efficient in inducing homogeneous and reproducible tumor destruction by thermal coagulation necrosis. However, accurate targeting of human abdominal tumors is difficult to maintain due to the motion induced by breathing. An algorithm is presented to track a region of interest of fixed size in a sequence of images. This algorithm was evaluated on synthetic data and on in vivo sequences of ultrasound liver images acquired using 12 MHz ultrasound imaging probe at a rate of 16 frames/s. The algorithm presented here was derived from the non-linear constant brightness assumption. Since the motion was smooth it was possible to reduce the space of admissible displacements; hence the number of unknown parameters was small compared with the size of the data. The optimal displacement was estimated by a Gauss-Newton method, and the matrix required at each step was assembled by reading the data only once. This algorithm was applied to simulated data, where the true displacement was known and a precise evaluation was possible. The relative error was about 2%. The algorithm was also applied to a video sequence of sonograms acquired during in vivo experiments. These trials were conducted on porcine liver since its size and physiology are similar to humans. Movements were induced by breathing and heart-beating. Two particular frequencies representing breathing (0.26 Hz) and heart beat (1.14 Hz) were identified in the estimated displacement and were correlated with the monitored breathing (0.27 Hz) and electrocardiograms (1.28 Hz). In addition, a region of interest (ROI) modeling the focal zone of a HIFU transducer was tracked along time. Therefore this study provides a mean of determining the location of the targeted region in vivo during HIFU treatments. This can be applied to correct the location of the focal zone accordingly. This method can preferentially be applied to the liver or to any other moving organ.
本研究为高强度聚焦超声（HIFU）治疗期间移动目标的跟踪做出了贡献。事实上，HIFU已被证明在通过热凝固坏死诱导均匀且可重复的肿瘤破坏方面非常有效。然而，由于呼吸引起的运动，难以保持对人体腹部肿瘤的精确靶向。提出了一种算法来跟踪图像序列中固定大小的感兴趣区域。该算法在合成数据以及使用12 MHz超声成像探头以16帧/秒的速率采集的肝脏超声体内序列上进行了评估。这里提出的算法源自非线性恒定亮度假设。由于运动是平滑的，所以可以减少允许位移的空间；因此，与数据大小相比，未知参数的数量较少。通过高斯 - 牛顿法估计最佳位移，并且在每一步所需的矩阵仅通过读取一次数据来组装。该算法应用于模拟数据，其中真实位移是已知的并且可以进行精确评估。相对误差约为2％。该算法还应用于体内实验期间采集的超声图像视频序列。这些试验在猪肝上进行，因为其大小和生理机能与人类相似。运动由呼吸和心跳引起。在估计的位移中识别出代表呼吸（0.26 Hz）和心跳（1.14 Hz）的两个特定频率，并与监测到的呼吸（0.27 Hz）和心电图（1.28 Hz）相关。此外，对模拟HIFU换能器焦点区域的感兴趣区域（ROI）进行了随时间的跟踪。因此，本研究提供了一种在HIFU治疗期间确定体内目标区域位置的方法。这可用于相应地校正焦点区域的位置。该方法可优先应用于肝脏或任何其他移动器官。
