用于超声心动图诊断后收缩期卷曲的半自动算法开发
研究提出两种半自动算法(MIRA与MAIBA)用于超声心动图诊断后收缩期卷曲(PSC),在100例患者中测试;MAIBA与手动测量相关系数0.80、准确率91%,诊断PSC准确率85%,最佳截断值69°,提示其可能提高PSC识别能力。
要点
- MAIBA半自动算法诊断PSC准确率85%,优于MIRA的76%。
- MAIBA最佳截断值为69°,经ROC曲线与约登指数确定。
- 两种算法与手动测量相关性好,观察者间信度ICC为0.98(MIRA)和0.97(MAIBA)。
核心结论
- MIRA与MAIBA角度是有前景的诊断工具,可能提高临床识别PSC的可靠性与准确性,为心律失常风险的诊断支持提供帮助。
研究设计与样本
- 研究提出基于心脏组织追踪与生物医学图像处理技术的半自动算法,在100例患者(44例女性,中位年龄58岁)中测试,按初始目测估计分为PSC阳性与PSC阴性组,所有受试者接受胸骨旁长轴切面经胸超声心动图检查。
诊断标准与关键数值
- MAIBA角度最佳截断值为69°,通过受试者工作特征曲线分析与约登指数确定。
- MIRA与手动测量的Pearson相关系数为0.70,百分比准确率为89%。
- MAIBA与手动测量的Pearson相关系数为0.80,百分比准确率为91%。
- 诊断PSC的准确率MIRA为76%,MAIBA为85%。
- 观察者间信度组内相关系数MIRA为0.98,MAIBA为0.97。
临床应用与阅片要点
- 作者认为MIRA与MAIBA角度是有前景的诊断工具,可能提高临床识别PSC的可靠性与准确性。
- MAIBA半自动算法诊断PSC准确率为85%,MIRA为76%。
- Bland-Altman分析证实半自动方法与手动方法之间具有一致性。
局限性
- 原文未报告局限性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
英文原文(摘要)· 点击展开
OBJECTIVE: Posterior systolic curling (PSC) is a morphological and functional abnormality of the posterior mitral valve annulus, known in the literature as a significant risk factor for the onset of malignant ventricular arrhythmias and sudden cardiac death. This study proposes a semi-automatic algorithm to acquire the Mitral valve annulus to Infero-basal wall Rotation Angle (MIRA) and an innovative semi-automatic echocardiographic parameter, called Mitral Annulus to Infero-Basal wall Angle (MAIBA) for diagnosing PSC.
METHODS: Both algorithms leverage cardiac tissue tracking and biomedical image-processing techniques. They were tested on 100 patients (44 female, median age 58 years) and classified into PSC-positive and PSC-negative groups based on an initial eyeball estimation. All subjects underwent echocardiographic exams using parasternal long-axis view transthoracic echocardiography.
RESULTS: The MIRA and MAIBA algorithms achieved Pearson correlation coefficients of 0.70 and 0.80, respectively, with respective percentage accuracies of 89% and 91%, respectively, supporting a strong correlation with manual measurements. Bland-Altman analyses confirmed agreement between semi-automatic and manual methods, and inter-observer reliability was excellent, with intra-class correlation coefficient values of 0.98 for MIRA and 0.97 for MAIBA. The optimal cutoff value for the MAIBA angle was set equal to 69° through receiver operator characteristic curve analysis and Youden's index. The diagnostic accuracy of PSC identification was 76% for MIRA and 85% for MAIBA, demonstrating robust potential for clinical application.
CONCLUSION: In conclusion, statistical analyses underscore MIRA and MAIBA angles as promising diagnostic tools, potentially enhancing clinicians' ability to identify PSC with high reliability and accuracy, paving the way for improved diagnostic support in cases of arrhythmogenic risk.
引用
- Bergo N, Piccolo A, Susin FM, et al. Development of Semi-automatic Algorithms for the Echocardiographic Diagnosis of Posterior Systolic Curling. Ultrasound Med Biol. 2025. PMID: 40713309. doi:10.1016/j.ultrasmedbio.2025.06.020
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。