• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>SPIE Medical Imaging Conference >Localization and Labeling of Cervical Vertebral Bones in the Micro-CT Images of Rabbit Fetuses Using a 3D Deep Convolutional Neural Network
【24h】

Localization and Labeling of Cervical Vertebral Bones in the Micro-CT Images of Rabbit Fetuses Using a 3D Deep Convolutional Neural Network

机译：使用3D深卷积神经网络的兔胎儿微型CT图像中颈椎骨骼的定位和标记

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

In developmental and reproductive toxicology (DART) studies, high-throughput micro-CT imaging of Dutch-Beltedrabbit fetuses has been used as a method for the assessment of compound-induced skeletal abnormalities. Sinceperforming visual inspection of the micro-CT images by the DART scientists is a time- and resource-intensive task, anautomatic strategy was proposed to localize, segment out, label, and evaluate each bone on the skeleton in a testingenvironment. However, due to the lack of robustness in this bone localization approach, failures on localizing certainbones on the critical path while traversing the skeleton, e.g., the cervical vertebral bones, could lead to localization errorsfor other bones downstream. Herein an approach based on deep convolutional neural networks (CNN) is proposed toautomatically localize each cervical vertebral bone represented by its center. For each center, a 3D probability map withGaussian decay is computed with the center itself being the maximum. From cervical vertebrae C1 to C7, the 7 volumesof distance transforms are stacked in order to form a 4-dimensional array. The deep CNN with a 3D U-Net architectureis used to estimate the probability maps for vertebral bone centers from the CT images as the input. A post-processingscheme is then applied to find all the regions with positive response, eliminate the false ones using a point-basedregistration method, and provide the locations and labels for the 7 cervical vertebral bones. Experiments were carried outon a dataset of 345 rabbit fetus micro-CT volumes. The images were randomly divided into training/validation/testingsets at an 80/10/10 ratio. Results demonstrated a 94.3% success rate for localization and labeling on the testing dataset of35 images, and for all the successful cases the average bone-by-bone localization error was at 0.84 voxel.
机译：在发育和生殖毒理学（DART）研究中，荷兰腰带的高通量微型CT成像兔胎儿已被用作评估复合骨骼异常的方法。自从通过飞镖科学家对微型CT图像进行目视检查是一项时间和资源密集型任务建议在测试中本地化，分段，标签和评估骨骼上的每个骨骼的自动策略环境。但是，由于这种骨质定位方法缺乏稳健性，对某些情况的失败在临界路径上弯曲的骨骼，例如颈椎骨骼，可能导致局部化错误对于下游的其他骨头。这里提出了一种基于深卷积神经网络（CNN）的方法自动定位由其中心表示的每个颈椎骨骼。对于每个中心，3D概率图高斯衰变使用中心本身进行了最大值。从颈椎C1到C7，7个容量距离变换堆叠以形成4维阵列。具有3D U-Net架构的深层CNN用于估计从CT图像作为输入的椎骨骨中心的概率图。后处理然后应用方案来查找具有肯定响应的所有区域，使用基于点的错误消除错误的区域注册方法，并为7个颈椎骨骼提供位置和标签。进行实验在345兔胎儿微型CT卷的数据集上。图像随机分为培训/验证/测试设置为80/10/10比率。结果显示了94.3％的定位成功率和标签测试数据集35图像，并且对于所有成功情况下，平均骨骨定位误差为0.84 voxel。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号