|
| |
|
| 本文已被：浏览 1660次 下载 1106次 |
码上扫一扫！ |
| 深度学习在儿童心脏超声标准切面自动智能识别中的应用 |
| 刘贻曼1,2,3,韩啸翔4,张玉奇2,3,张志芳2,沈蓉2,陈丽君2,董斌3,袁加俊3,胡孟晗1,李庆利1,陈建刚1* |
|
|
(1. 华东师范大学通信与电子工程学院上海市多维度信息处理重点实验室, 上海 200241;
2. 上海交通大学医学院附属上海儿童医学中心心内科, 上海 200127;
3. 上海智慧儿科临床诊治技术工程技术研究中心, 上海 200127;
4. 上海理工大学健康科学与工程学院智能医学影像与计算医疗实验室, 上海 200093
*通信作者) |
|
| 摘要: |
| 目的 探讨深度学习在儿童心脏超声标准切面自动识别中的可行性和准确性。方法 在上海交通大学医学院附属上海儿童医学中心心脏超声诊断中心影像归档和通信系统数据库中，选取2022年9－10月行心脏超声检查儿童的4 035张心脏超声图像，按照6∶2∶2的比例将图像随机分为训练集（2 421张）、验证集（807张）、测试集（807张）。通过改进密集连接网络（DenseNet）开发了一个轻量、高效的深度学习模型，实现对15个儿童心脏超声标准切面的自动识别，并与DenseNet121、InceptionV3、MobileNetV3 3种常用的深度学习模型进行比较。以人工手动标注结果为金标准，采用准确度、精确率、特异度、召回率和F1指数评价深度学习模型的识别性能。使用参数量、模型大小和浮点运算数3个指标评估模型的识别效率。采用混淆矩阵展示模型的识别结果，并通过热力图反映模型对图像特征的关注度。结果 DenseNet121、InceptionV3、MobileNetV3模型和所提出的深度学习模型识别15个儿童心脏超声标准切面和非标准切面的平均F1指数分别为94.59%、95.13%、92.41%、94.73%，参数量分别为7.0×106、24.4×106、4.2×106、1.8×106，模型大小分别为13.941、48.777、8.445、3.588 MB，浮点运算数分别为11.16×109、12.89×109、0.86×109、3.05×109。从混淆矩阵和热力图可以看出，所提出的模型对15个儿童心脏超声标准切面和非标准切面的识别率高于DenseNet121、InceptionV3、MobileNetV3模型，且能够关注到超声切面中的关键特征区域。结论 所提出的深度学习模型可准确地识别儿童心脏超声标准切面，且模型的参数量较少，运行效率较高。 |
| 关键词: 心脏超声 儿童 深度学习 标准切面 智能识别 |
| DOI：10.16781/j.CN31-2187/R.20220936 |
| 投稿时间：2022-12-17修订日期：2023-03-07 |
| 基金项目:国家自然科学基金（61975056），上海市自然科学基金（19ZR1416000），上海市科学技术委员会资助项目（20440713100）. |
|
| Deep learning for automatic intelligent identification of pediatric echocardiography standard views |
| LIU Yiman1,2,3,HAN Xiaoxiang4,ZHANG Yuqi2,3,ZHANG Zhifang2,SHEN Rong2,CHEN Lijun2,DONG Bin3,YUAN Jiajun3,HU Menghan1,LI Qingli1,CHEN Jiangang1* |
(1. Shanghai Key Laboratory of Multidimensional Information Processing, School of Communication and Electronic Engineering, East China Normal University, Shanghai 200241, China;
2. Department of Pediatric Cardiology, Shanghai Children's Medical Center, Shanghai Jiao Tong University School of Medicine, Shanghai 200127, China;
3. Shanghai Engineering Research Center of Intelligence Pediatrics(SERCIP), Shanghai 200127, China;
4. Intelligent Medical Imaging and Computational Medicine Laboratory, School of Health Sciences and Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China
*Corresponding author) |
| Abstract: |
| Objective To explore the feasibility and accuracy of deep learning in automatic identification of standard views of pediatric echocardiography. Methods A total of 4 035 pediatric echocardiography images from the picture archiving and communication system database of Shanghai Children’s Medical Center, Shanghai Jiao Tong University School of Medicine, which were collected from Sep. to Oct. 2022, were selected and randomly divided into training set (2 421 images), validation set (807 images), and testing set (807 images) in a ratio of 6∶2∶2. A lightweight and efficient deep learning model was developed by improving DenseNet to achieve automatic identification of 15 standard views of pediatric echocardiography, and was compared with 3 commonly used deep learning models, including DenseNet121, InceptionV3, and MobileNetV3. With manual annotation results as the gold standard, the identification performance of deep learning models was evaluated using accuracy, precision, specificity, recall, and F1 score. The efficiency of the identification model was evaluated using 3 indicators: the number of parameters, model size, and floating-point operations. The identification results of the model were displayed using a confusion matrix, and the model’s concerns to image features were reflected using a heatmap. Results The average F1 scores of DenseNet121, InceptionV3, MobileNetV3, and the proposed model for identifying 15 standard and non-standard views of pediatric echocardiography were 94.59%, 95.13%, 92.41%, and 94.73%, the numbers of parameters were 7.0×106, 24.4×106, 4.2×106, and 1.8×106, the model sizes were 13.941, 48.777, 8.445, and 3.588 MB, and the floating-point operations were 11.16×109, 12.89×109, 0.86×109, and 3.05×109, respectively. The confusion matrix and heatmap showed that the proposed model had a higher recognition rate for 15 standard and non-standard views of pediatric echocardiography than DenseNet121, InceptionV3 and MobileNetV3, and was able to focus on the key feature areas in the ultrasonic views. Conclusion The deep learning model proposed in this study can accurately identify standard cardiac ultrasound views in children; moreover, the model has a small number of parameters and can be operated with high efficiency. |
| Key words: echocardiography child deep learning standard view intelligent recognition |
|
|
|
|
|
