Detecting Anatomical Landmarks From Limited Medical Imaging Data Using Two-Stage Task-Oriented Deep Neural Networks.
Detecting Anatomical Landmarks From Limited Medical Imaging Data Using Two-Stage Task-Oriented Deep Neural Networks.
复制标题
DOI：
10.1109/tip.2017.2721106
复制
发表时间：
2017-10
期刊：
影响因子：
--
通讯作者：
Shen D
中科院分区：
文献类型：
--
作者：
Zhang J;Liu M;Shen D
基于深度神经网络的解剖标志检测的主要挑战之一是用于网络学习的医学成像数据的有限可用性。为了解决这个问题，我们提出了一个两阶段的面向任务的深度学习（T2DL）方法，使用有限的训练数据，在真实的时间内同时检测大规模解剖标志。具体来说，我们的方法由两个深度卷积神经网络（CNN）组成，每个网络专注于一个特定的任务。具体来说，为了缓解训练数据有限的问题，在第一阶段，我们提出了一个基于CNN的回归模型，使用数百万个图像块作为输入，旨在学习局部图像块与目标解剖标志之间的内在关联。为了进一步对图像块之间的相关性进行建模，在第二阶段中，我们开发了另一种CNN模型，该模型包括a）与第一阶段中使用的CNN共享相同架构和网络权重的全卷积网络（FCN），以及B）几个额外的层，以联合预测多个解剖标志的坐标。重要的是，我们的方法可以联合检测大规模（例如，数千个）地标。我们已经进行了各种实验，从700名受试者的3D T1加权磁共振（MR）图像中检测1200个脑标志，并从73名受试者的3D计算机断层扫描（CT）图像中检测7个前列腺标志。实验结果表明，我们的方法在解剖标志检测的准确性和效率的有效性。
One of the major challenges in anatomical landmark detection, based on deep neural networks, is the limited availability of medical imaging data for network learning. To address this problem, we present a two-stage task-oriented deep learning (T2DL) method to detect large-scale anatomical landmarks simultaneously in real time, using limited training data. Specifically, our method consists of two deep convolutional neural networks (CNN), with each focusing on one specific task. Specifically, to alleviate the problem of limited training data, in the first stage, we propose a CNN based regression model using millions of image patches as input, aiming to learn inherent associations between local image patches and target anatomical landmarks. To further model the correlations among image patches, in the second stage, we develop another CNN model, which includes a) a fully convolutional network (FCN) that shares the same architecture and network weights as the CNN used in the first stage and also b) several extra layers to jointly predict coordinates of multiple anatomical landmarks. Importantly, our method can jointly detect large-scale (e.g., thousands of) landmarks in real time. We have conducted various experiments for detecting 1200 brain landmarks from the 3D T1-weighted magnetic resonance (MR) images of 700 subjects, and also 7 prostate landmarks from the 3D computed tomography (CT) images of 73 subjects. The experimental results show the effectiveness of our method regarding both accuracy and efficiency in the anatomical landmark detection.