Title: Heart Segmentation From MRI Scans Using Convolutional Neural Network Show Chinese title

Title: 基于卷积神经网络的MRI扫描心脏分割

Abstract: Heart is one of the vital organs of human body. A minor dysfunction of heart even for a short time interval can be fatal, therefore, efficient monitoring of its physiological state is essential for the patients with cardiovascular diseases. In the recent past, various computer assisted medical imaging systems have been proposed for the segmentation of the organ of interest. However, for the segmentation of heart using MRI, only few methods have been proposed each with its own merits and demerits. For further advancement in this area of research, we analyze automated heart segmentation methods for magnetic resonance images. The analysis are based on deep learning methods that processes a full MR scan in a slice by slice fashion to predict desired mask for heart region. We design two encoder decoder type fully convolutional neural network models Show Chinese abstract

Abstract: 心脏是人体的重要器官之一。 即使短时间内出现轻微的心脏功能障碍也可能致命，因此，对心血管疾病患者的心脏生理状态进行有效监测至关重要。 在最近的过去，已经提出了各种计算机辅助医学影像系统，用于感兴趣器官的分割。 然而，对于使用MRI进行心脏分割，只有少数方法被提出，每种方法都有其自身的优点和缺点。 为了进一步推进这一研究领域，我们分析了用于磁共振图像的自动心脏分割方法。 分析基于深度学习方法，这些方法以逐切片的方式处理完整的MR扫描，以预测心脏区域的期望掩码。 我们设计了两个编码器-解码器类型的全卷积神经网络模型