Learning image context for segmentation of prostate in CT-guided radiotherapy.
Learning image context for segmentation of prostate in CT-guided radiotherapy.
复制标题
DOI：
10.1007/978-3-642-23626-6_70
复制
发表时间：
2011
期刊：
影响因子：
--
通讯作者：
Shen D
中科院分区：
文献类型：
--
作者：
Li W;Liao S;Feng Q;Chen W;Shen D
在前列腺癌的外照射治疗中，前列腺的分割是非常重要的。然而，由于图像对比度低，前列腺运动，以及前列腺周围的膀胱和直肠的强度和形状的变化，在CT图像中定位前列腺癌是一项具有挑战性的工作。本文提出了一种基于位置自适应图像上下文的在线学习和患者特定分类方法来精确分割CT图像中的前列腺。具体地，将两组位置自适应分类器分别沿两个坐标方向放置，并与先前分割的治疗图像进一步训练，共同进行前列腺分割。特别是，每个位置自适应分类器使用在不同尺度和方向上收集的不同图像上下文递归地训练，以更好地识别每个前列腺区域。所提出的基于学习的前列腺分割方法已经在大量的患者身上进行了广泛的评估，取得了非常有希望的结果。
Segmentation of prostate is highly important in the external beam radiotherapy of prostate cancer. However, it is challenging to localize prostate in the CT images due to low image contrast, prostate motion, and both intensity and shape changes of bladder and rectum around the prostate. In this paper, an online learning and patient-specific classification method based on location-adaptive image context is proposed to precisely segment prostate in the CT image. Specifically, two sets of position-adaptive classifiers are respectively placed along the two coordinate directions, and further trained with the previous segmented treatment images to jointly perform the prostate segmentation. In particular, each location-adaptive classifier is recursively trained with different image context collected at different scales and orientations for better identification of each prostate region. The proposed learning-based prostate segmentation method has been extensively evaluated on a large set of patients, achieving very promising results.