Regression forests for efficient anatomy detection and localization in computed tomography scans
Regression forests for efficient anatomy detection and localization in computed tomography scans
复制标题
DOI：
10.1016/j.media.2013.01.001
复制
发表时间：
2013-12-01
影响因子：
10.9
通讯作者：
Siddiqui, K.
中科院分区：
文献类型：
--
作者：
Criminisi, A.;Robertson, D.;Siddiqui, K.
提出了一种在三维CT扫描中对多个解剖结构进行高效、自动检测和定位的新算法。应用包括从PACS系统中选择性地检索患者图像、语义视觉导航和跟踪随时间变化的辐射剂量。这项工作的主要贡献是解剖定位问题的一种新的、连续的参数化，这使得该问题可以通过多DOSS随机回归森林来有效地解决。回归森林类似于更流行的分类森林，但经过训练以预测连续的、多变量的产出，其中训练的重点是最大限度地提高产出预测的置信度。我们的概率算法的一次通过实现了从体素到器官位置和大小的直接映射。定量验证是在400个高可变CT扫描的数据库上执行的。实验结果表明，该方法比基于高效多图谱配准和基于模板的最近邻检测方法具有更高的精确度和鲁棒性。由于递归器的上下文丰富的视觉特征的简单性和算法的并行性，这些结果在仅类似于传统单核机器上的4s的典型运行时间内获得。(C)2013爱思唯尔B.V.保留所有权利。
This paper proposes a new algorithm for the efficient, automatic detection and localization of multiple anatomical structures within three-dimensional computed tomography (CT) scans. Applications include selective retrieval of patients images from PACS systems, semantic visual navigation and tracking radiation dose over time.The main contribution of this work is a new, continuous parametrization of the anatomy localization problem, which allows it to be addressed effectively by multi-doss random regression forests. Regression forests are similar to the more popular classification forests, but trained to predict continuous, multivariate outputs, where the training focuses on maximizing the confidence of output predictions. A single pass of our probabilistic algorithm enables the direct mapping from voxels to organ location and size.Quantitative validation is performed on a database of 400 highly variable CT scans. We show that the proposed method is more accurate and robust than techniques based on efficient multi-atlas registration and template-based nearest-neighbor detection. Due to the simplicity of the regressor's context-rich visual features and the algorithm's parallelism, these results are achieved in typical run-times of only similar to 4s on a conventional single-core machine. (C) 2013 Elsevier B.V. All rights reserved.