论著摘要 |【AI-超声】基于2D超声心动图视频的广泛纹理特征自动评估二尖瓣反流严重程度(双语版)

2018-01-09 11:40:32 admin 0
标签:  人工智能 AI 机器学习 超声 二尖瓣反流 SV SVM 纹理分析

Automatic assessment of mitral regurgitation severity based on extensive textural features on 2D echocardiography videos.

发表日期: 2016.06.01   来源:Comput Biol Med. 2016 Jun 1;73:47-55.

作者:

Hanie Moghaddasia1, Saeed Nourian2.

作者介绍:

1. Department of Biomedical Engineering, Amirkabir University of Technology, Tehran, Iran. Electronic address: hn_moghadas@aut.ac.ir.

2. Department of Cardiology, Tehran University of Medical Science, Tehran, Iran.

摘要

心脏病是发达国家死亡的主要原因,也是导致残疾的主要原因。二尖瓣反流(MR)是一种常见的心脏病,直到最后阶段才会引起症状。因此,疾病的早期诊断在治疗过程中至关重要。超声心动图是MR严重程度的常见诊断方法。因此,基于超声心动图视频,图像处理技术和人工智能的方法可能对临床医生有帮助,特别是在边缘病例中。在本文中,我们介绍了检测超声心动图影像的微观特征,以确定MR的严重程度。广泛的本地二进制模式(ELBP)和广泛体量局部二进制模式(EVLBP)被呈现为图像描述符,其包括来自特征向量心脏不同观点的细节。支持向量机(SVM),线性判别分析(LDA)和模板匹配技术被用作分类器,以确定基于纹理描述符的MR的严重性。具有广泛均匀局部二进制模式(ELBPU)和广泛体积局部二进制模式(EVLBP)的SVM分类器分别具有99.52%,99.38%,99.31%和99.59%的最佳精度,用于检测正常,轻度MR,中度 超声心动图影像中的MR和严重MR受试者。所提出的方法对MR和正常受试者的严重程度的检测具有99.38%的敏感性和99.63%的特异性。

Abstact

Heart disease is the major cause of death as well as a leading cause of disability in the developed countries. Mitral Regurgitation (MR) is a common heart disease which does not cause symptoms until its end stage. Therefore, early diagnosis of the disease is of crucial importance in the treatment process. Echocardiography is a common method of diagnosis in the severity of MR. Hence, a method which is based on echocardiography videos, image processing techniques and artificial intelligence could be helpful for clinicians, especially in borderline cases. In this paper, we introduce novel features to detect micro-patterns of echocardiography images in order to determine the severity of MR. Extensive Local Binary Pattern (ELBP) and Extensive Volume Local Binary Pattern (EVLBP) are presented as image descriptors which include details from different viewpoints of the heart in feature vectors. Support Vector Machine (SVM), Linear Discriminant Analysis (LDA) and Template Matching techniques are used as classifiers to determine the severity of MR based on textural descriptors. The SVM classifier with Extensive Uniform Local Binary Pattern (ELBPU) and Extensive Volume Local Binary Pattern (EVLBP) have the best accuracy with 99.52%, 99.38%, 99.31% and 99.59%, respectively, for the detection of Normal, Mild MR, Moderate MR and Severe MR subjects among echocardiography videos. The proposed method achieves 99.38% sensitivity and 99.63% specificity for the detection of the severity of MR and normal subjects.

关键词:

2D超声心动图,机器学习,微格局,二尖瓣反流,纹理分析

Keywords:

2D echocardiography; Machine learning; Micro-patterns; Mitral regurgitation; Textural analysis

阅读原文:PMID: 27082766  DOI: 10.1016/j.compbiomed.2016.03.026


慧影医疗科技(北京)有限公司

地点:北京市海淀区中关村东升科技园B2-C103

电话:400-890-9020

邮箱:radcloud@huiyihuiying.com

关闭
图片
图片