首页> 外文会议>International joint conference on computational intelligence >Automated Diagnostic Model Based on Isoline Map Analysis of Myocardial Tissue Structure
【24h】

Automated Diagnostic Model Based on Isoline Map Analysis of Myocardial Tissue Structure

机译:基于等值线图分析的心肌组织结构自动诊断模型

获取原文

摘要

Diagnostics of heart diseases by myocardial tissue structure analysis is very important since there exist several cardiac pathologies, indistinguishable by other symptoms. The most similar existing methods for automatic diagnostics by myocardium analysis in CT images are based only on intensity histogram features. In this work we describe the proposed method that uses isoline map-based features. We use real MSCT images with manually segmented LV myocardium to compare the existing algorithm and variations of the proposed algorithm, utilizing different strategies of isoline map construction (single- and double-level maps) for further computation of features, feature selection methods (Information Gain and Chi-squared test) and algorithms for binary classification of slice images into normal/abnormal classes (SVM and Random Forest). All considered types of isoline map-based diagnostic models demonstrate better results than histogram-based model. The best diagnostic models achieve 99.7% ROC AUC, 96.7% F-score and 0.7% false negative rate.
机译:通过心肌组织结构分析对心脏病的诊断非常重要,因为存在几种心脏病,其他症状无法区分。通过CT图像中的心肌分析进行自动诊断的现有方法中,最相似的方法仅基于强度直方图特征。在这项工作中,我们描述了使用基于等值线图的特征的建议方法。我们使用带有手动分段LV心肌的真实MSCT图像来比较现有算法和该算法的变体,利用等值线图构造(单层和双层图)的不同策略进一步计算特征,特征选择方法(信息增益)和卡方检验)和将切片图像按正常/异常类别(SVM和随机森林)进行二进制分类的算法。与基于直方图的模型相比,所有考虑的基于等值线图的诊断模型类型均显示出更好的结果。最好的诊断模型可实现99.7%的ROC AUC,96.7%的F评分和0.7%的假阴性率。

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