A preliminary exploration of a deep learning-based artificial intelligence model for automatic quantification of echocardiographic left ventricular ejection fraction
-
摘要:目的
利用超声心动图静态视图, 构建一种基于深度学习的人工智能模型, 以自动量化左心室射血分数(LVEF)。
方法将1 902例成人左心室收缩末期和舒张末期的多切面超声心动图视图数据纳入本研究。将收集的数据集分为开发集(1 610例, 其中1 252例用于模型训练, 358例用于参数调整)、内部测试集(177例, 用于内部验证)和外部测试集(115例, 用于外部验证和泛化性检测)。该模型通过精确识别左心室心内膜边界和关键点检查, 实现左心室分割和自动量化LVEF。采用Dice系数评估左心室分割模型的性能; 采用Pearson相关系数和组内相关系数评估自动测量的LVEF与参考标准的相关性和一致性。
结果左心室分割模型性能良好, 内部和外部独立测试集的Dice系数均≥ 0.90; 自动测量的LVEF与心脏专家人工测量的一致性中等, 内部测试集的Pearson相关系数为0.46~0.71, 组内相关分析一致性为0.39~0.57; 外部测试集的Pearson相关系数为0.26~0.54, 组内相关分析一致性为0.23~0.50。
结论本研究构建了一种性能较好的左心室分割和关键点检测模型, 但初步应用该模型自动定量LVEF的效能一般, 尚需进一步优化算法, 提高模型泛化性。
Abstract:ObjectiveTo construct a deep learning-based artificial intelligence model to automatically quantify left ventricular ejection fraction (LVEF) using static views of echocardiography.
MethodsThe study included data of 1, 902 adults with left ventricular multi-slice echocardiographic views at end-systole and end-diastole. The collected dataset was divided into development set (1, 610 cases, with 1, 252 cases for model training and 358 cases for parameter adjustment), internal test set (177 cases for internal validation), and external test set (115 cases for external validation and generalization testing). The model achieved left ventricular segmentation and automatic quantification of LVEF through precise identification of the left ventricular endocardial boundary and inspection of key points. The Dice coefficient was employed to evaluate the performance of the left ventricular segmentation model, while the Pearson correlation coefficient and the intraclass correlation coefficient were used to assess the correlation and consistency between the automatically measured LVEF and the reference standard.
ResultsThe left ventricular segmentation model performed well, with Dice coefficients ≥ 0.90 for both the internal and external independent test sets; the agreement between the automatically measured LVEF and the cardiologists' manual measurements was moderate, with Pearson correlation coefficients ranging from 0.46 to 0.71 and intragroup correlation analysis agreements from 0.39 to 0.57 for the internal test set; and Pearson correlation coefficients for the independent external test set were 0.26 to 0.54 and intra-group correlation analysis agreement of 0.23 to 0.50.
ConclusionIn this study, a left ventricular segmentation model with better performance is constructed, and initial application of the model for automatic quantification of LVEF for two-dimensional echocardiography has general performance, which requires further optimisation of the algorithm to improve the model generalisation.
-
-
表 1 入组人群临床基线资料特征比较(x±s)[n(%)]
基线资料 开发集(n=1 610) 内部测试集(n=177) 外部测试集(n=115) 男 838(52.05) 102(57.62) 43(37.39) 女 772(47.95) 75(42.37) 72(62.61) 年龄/岁 59.33±19.60 59.32±12.94 56.37±24.85 体表面积/m2 1.76±0.27 1.75±0.19 1.56±0.44 左心室舒张末期容积/mL 97.54±30.46 98.88±33.19 71.29±30.31 左心室收缩末期容积/mL 35.15±18.73 35.24±18.80 29.35±14.31 左心室射血分数/% 64.80±5.86 65.36±6.44 81.86±20.45 左心室舒张末期直径/mm 45.63±6.27* 46.76±5.94 41.36±7.79 与内部测试集比较, * P < 0.05。 
下载: 导出CSV
表 2 分割任务Dice相似系数(x±s)
Dice相似系数 心尖2腔心切面 心尖3腔心切面 心尖4腔心切面 内部测试集 外部测试集 内部测试集 外部测试集 内部测试集 外部测试集 舒张末期帧 0.94±0.04 0.93±0.04 0.94±0.03 0.91±0.03 0.96±0.02 0.94±0.03 收缩末期帧 0.92±0.03 0.90±0.06 0.91±0.06 0.90±0.05 0.93±0.03 0.91±0.04 舒张末期帧+收缩末期帧 0.93±0.04 0.92±0.05 0.94±0.03 0.91±0.04 0.94±0.03 0.93±0.04
下载: 导出CSV
表 3 模型自动测量左心室与参考标准Pearson相关系数和ICC
相关系数 数据集 心尖2腔心+心尖4腔心切面 心尖2腔心+心尖3腔心切面 心尖3腔心+心尖4腔心切面 收缩末期帧 舒张末期帧 收缩末期帧 舒张末期帧 收缩末期帧 舒张末期帧 Pearson相关系数 内部测试集 0.77 0.58 0.68 0.51 0.64 0.53 外部测试集 0.39 0.19 0.38 0.30 0.36 0.23 内部测试集+外部测试集 0.54 0.25 0.49 0.30 0.42 0.25 组内相关系数 内部测试集 0.73(0.58, 0.82) 0.47(0.13, 0.67) 0.59(0.27, 0.76) 0.33(-0.04, 0.47) 0.56(0.70, 0.31) 0.41(0.08, 0.62) 外部测试集 0.32(0.13, 0.48) 0.15(-0.02, 0.32) 0.35(0.18, 0.51) 0.29(0.11, 0.45) 0.31(0.14, 0.47) 0.20(0.03, 0.37) 内部测试集+外部测试集 0.53(0.44, 0.61) 0.25(0.14, 0.36) 0.47(0.37, 0.56) 0.27(0.15, 0.39) 0.42(0.31, 0.51) 0.25(0.14, 0.36)
下载: 导出CSV
-
[1] SAVARESE G, BECHER P M, LUND L H, et al. Global burden of heart failure: a comprehensive and updated review of epidemiology[J]. Cardiovasc Res, 2023, 118(17): 3272-3287. doi: 10.1093/cvr/cvac013
[2] AGGARWAL R, YEH R W, JOYNT MADDOX K E, et al. Cardiovascular risk factor prevalence, treatment, and control in US adults aged 20 to 44 years, 2009 to March 2020[J]. JAMA, 2023, 329(11): 899-909. doi: 10.1001/jama.2023.2307
[3] SHAH K S, XU H L, MATSOUAKA R A, et al. Heart failure with preserved, borderline, and reduced ejection fraction: 5-year outcomes[J]. J Am Coll Cardiol, 2017, 70(20): 2476-2486. doi: 10.1016/j.jacc.2017.08.074
[4] KOH A S, TAY W T, TENG T H K, et al. A comprehensive population-based characterization of heart failure with mid-range ejection fraction[J]. Eur J Heart Fail, 2017, 19(12): 1624-1634. doi: 10.1002/ejhf.945
[5] 湛先发, 余小亚, 王洪军, 等. 3种机器学习算法评估脑梗死患者颈动脉斑块稳定性的效能比较[J]. 实用临床医药杂志, 2023, 27(22): 6-12. doi: 10.7619/jcmp.20232657 [6] 裴昌军, 孙雪丽, 王鑫, 等. 人工智能结合多层螺旋CT检查在机关体检人群肺结节筛查中的应用[J]. 实用临床医药杂志, 2023, 27(24): 89-92. doi: 10.7619/jcmp.20232282 [7] 曾研. 医学超声若干目标检测深度学习方法研究[D]. 北京: 北京工业大学, 2022. [8] 张浩, 常建东. 基于文献计量方法的人工智能在超声心动图中的应用进展研究[J]. 中国医疗设备, 2023, 38(1): 127-133. https://www.cnki.com.cn/Article/CJFDTOTAL-YLSX202301024.htm [9] 中华医学会超声医学分会超声心动图学组. 中国成年人超声心动图检查测量指南[J]. 中华超声影像学杂志, 2016, 25(8): 645-666. https://cdmd.cnki.com.cn/Article/CDMD-10632-1018178662.htm [10] LANG R M, BADANO L P, MOR-AVI V, et al. Recommendations for cardiac chamber quantification by echocardiography in adults: an update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging[J]. Eur Heart J Cardiovasc Imaging, 2015, 16(3): 233-270. doi: 10.1093/ehjci/jev014
[11] LINDENHEIM-LOCHER W, SWITONSKI A, KRZESZOWSKI T, et al. YOLOv5 drone detection using multimodal data registered by the vicon system[J]. Sensors, 2023, 23(14): 6396. doi: 10.3390/s23146396
[12] LIU X, FAN Y T, LI S, et al. Deep learning-based automated left ventricular ejection fraction assessment using 2-D echocardiography[J]. Am J Physiol Heart Circ Physiol, 2021, 321(2): H390-H399. doi: 10.1152/ajpheart.00416.2020
[13] XIE S N, GIRSHICK R, DOLLAR P, et al. Aggregated residual transformations for deep neural networks[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017: 1492-1500.
[14] LAM C S P, SOLOMON S D. Classification of HeartFailure according to ejection fraction: JACC review topic of the week[J]. J Am Coll Cardiol, 2021, 77(25): 3217-3225. doi: 10.1016/j.jacc.2021.04.070
[15] ZAMZMI G, HSU L Y, LI W, et al. Harnessing machine intelligence in automatic echocardiogram analysis: current status, limitations, and future directions[J]. IEEE Rev Biomed Eng, 2021, 14: 181-203. doi: 10.1109/RBME.2020.2988295
[16] OSTVIK A, SMISTAD E, AASE S A, et al. Real-time standard view classification in transthoracic echocardiography using convolutional neural networks[J]. Ultrasound Med Biol, 2019, 45(2): 374-384. doi: 10.1016/j.ultrasmedbio.2018.07.024
[17] MORADI S, OGHLI M G, ALIZADEHASL A, et al. MFP-Unet: a novel deep learning based approach for left ventricle segmentation in echocardiography[J]. Phys Med, 2019, 67: 58-69. doi: 10.1016/j.ejmp.2019.10.001
[18] OUYANG D, HE B, GHORBANI A, et al. Video-based AI for beat-to-beat assessment of cardiac function[J]. Nature, 2020, 580(7802): 252-256. doi: 10.1038/s41586-020-2145-8
[19] REYNAUD H, VLONTZOS A, HOU B, et al. Ultrasound video transformers for cardiac ejection fraction estimation[C]//Medical Image Computing and Computer Assisted Intervention-MICCAI 2021: 24th International Conference, Strasbourg, France, September 27-October 1, 2021, Proceedings, Part Ⅵ. ACM, 2021: 495-505.
[20] PRADA G, FRITZ A V, RESTREPO-HOLGUÍN M, et al. Focused cardiac ultrasonography for left ventricular systolic function[J]. N Engl J Med, 2019, 381(21): e36.
[21] ASCH F M, MOR-AVI V, RUBENSON D, et al. Deep learning-based automated echocardiographic quantification of left ventricular ejection fraction: a point-of-care solution[J]. Circ Cardiovasc Imaging, 2021, 14(6): e012293. doi: 10.1161/CIRCIMAGING.120.012293
-
期刊类型引用(9)
1. 马雪萍，王晓丽，阿里木江·司马义，徐桂萍. 腹股沟上髂筋膜阻滞复合全身麻醉对高龄髋关节置换患者术后谵妄发生的影响. 新疆医学. 2022(02): 131-134 . 
百度学术
2. 张宏，李淑萍. 老年患者髋关节置换术后谵妄的发生现状及其相关影响因素分析. 长春中医药大学学报. 2022(10): 1155-1159 . 百度学术
3. 刘贵政，郑婷婷，杜斌. 老年髋部骨折患者术后谵妄发生现况及危险因素研究. 贵州医药. 2022(09): 1405-1406 . 百度学术
4. 王秀环，鲍乐乐，马漪洁，陈宁宁. 不同麻醉方法对老年髋关节置换患者术后谵妄发生的影响. 广州医科大学学报. 2021(02): 40-44 . 百度学术
5. 刘丹，杨万翔. 人文关怀护理对人工髋关节置换术后谵妄患者临床症状的影响. 中国当代医药. 2021(25): 270-272+276 . 百度学术
6. 陈立红，徐芙蓉，叶洁玉，许华亮. 高龄骨科髋关节置换术后患者发生谵妄的危险因素分析. 现代医学与健康研究电子杂志. 2021(23): 115-118 . 百度学术
7. 毛俊岚. 高龄患者髋关节置换术后谵妄1例的护理. 基层医学论坛. 2020(03): 420-421 . 百度学术
8. 欧玉琼，吴建颖，周嫦娥. 以临床路径为指导的谵妄管理对老年股骨头置换术后患者的影响. 护理实践与研究. 2020(02): 86-88 . 百度学术
9. 姜红卫. 骨科老年患者髋关节术后谵妄发生原因及疼痛干预护理进展. 系统医学. 2020(13): 196-198 . 百度学术
其他类型引用(1)
计量
- 文章访问数: 221
- HTML全文浏览量: 67
- PDF下载量: 32
- 被引次数: 10
苏公网安备 32100302010246号
