在使用基于深度学习的注意力转换器和心电图的年轻患者中识别先天性膜声
IEEE journal of biomedical and health informatics
|January 23, 2024
概括
一个新的深度学习模型使用心脏声音记录 (心声学) 来自动检测先天性心脏病 (CHD). 这种具有成本效益的工具有助于年轻患者的早期诊断,克服了传统心声回声学的局限性.
科学领域:
- 生物医学工程 生物医学工程
- 人工智能在医学中的应用
- 心脏病学 心脏病学
背景情况:
- 先天性心脏病 (CHD) 影响25%的新生儿,需要早期诊断.
- 目前的诊断方法,如心声回声学是专家依赖,昂贵,耗时的.
- 低收入和中等收入国家在及时诊断心血管疾病方面面临重大障碍.
研究的目的:
- 开发和验证一个深度学习模型,用于自动检测心脏声,表明心脏病.
- 利用心电图 (PCG) 作为一个具有成本效益和可访问的诊断工具.
- 改善儿童群体中心脏异常的早期检测率.
主要方法:
- 开发了一个基于深度学习的注意力转换器模型.
- 分析了来自四个听诊地点的942名年轻患者的心电图 (PCG) 记录.
- 波形特征在深度学习推断之前被用于减小维度.
- 该模型使用10倍交叉验证进行了验证.
主要成果:
- 该模型在检测声方面实现了90.23%的平均准确度和72.41%的灵敏度.
- 在未见的数据上,声的缺席和存在之间的歧视达到了76.10%的准确性.
- 预测声存在的准确率为70% (婴儿),88% (儿童) 和86% (青少年).
- 模型解释强调了特定的听觉位置 (AV,MV,TV) 对预测的重要性.
结论:
- 应用到PCG记录的深度学习为年轻人早期发现心血管疾病提供了强大,具有成本效益的工具.
- 该模型可以作为一线诊断辅助,减少对昂贵设备和专家口译的依赖.
- 这种方法有可能显著提高CHD诊断的可访问性,特别是在资源有限的环境中.
相关概念视频
Cardiovascular System Abnormal Findings II: Auscultation
150
Auscultation, an essential part of a heart examination, is done using a stethoscope. It provides crucial information about heart function and possible heart problems. Due to heart problems, abnormal sounds can be heard during systole or diastole. These sounds include S3 and S4 gallops, opening snaps, systolic clicks, and murmurs.
Abnormal Heart Sounds
Gallops:
Abnormal Heart Sounds
Gallops:
150
Heart Sounds
1.9K
Heart sounds are generated by the turbulence in blood flow due to the closing of heart valves. These sounds are best perceived slightly away from the valves, where the blood flow disseminates the sound.
Auscultation is the process of listening to these internal body sounds using a stethoscope. The heart produces four types of sounds, but only two—S1 and S2—can usually be heard with a stethoscope.
S1, also known as the "lub" sound, is caused by the closure of atrioventricular (A-V)...
Auscultation is the process of listening to these internal body sounds using a stethoscope. The heart produces four types of sounds, but only two—S1 and S2—can usually be heard with a stethoscope.
S1, also known as the "lub" sound, is caused by the closure of atrioventricular (A-V)...
1.9K


