An Interpretable CNN for the Segmentation of the Left Ventricle in Cardiac MRI by Real-Time Visualization

作     者：Jun Liu Geng Yuan Changdi Yang Houbing Song Liang Luo 

作者机构：Robotics InstituteSchool of Computer ScienceCarnegie Mellon UniversityPittsburghPA15217USA Department of Electrical&Computer EngineeringCollege of EngineeringNortheastern UniversityBostonMA02115USA Security and Optimization for Networked Globe Laboratory(SONG Lab)Embry-Riddle Aeronautical UniversityDaytona BeachFL32114USA School of Computer Science and EngineeringUniversity of Electronic Science and Technology of ChinaChengdu610054China 

出 版 物：《Computer Modeling in Engineering & Sciences》 (工程与科学中的计算机建模（英文）)

年 卷 期：2023年第135卷第5期

页      面：1571-1587页

核心收录：

学科分类：0831[工学-生物医学工程（可授工学、理学、医学学位）] 100207[医学-影像医学与核医学] 1002[医学-临床医学] 08[工学] 080203[工学-机械设计及理论] 1010[医学-医学技术（可授医学、理学学位）] 0802[工学-机械工程] 10[医学] 

基　　金：The National Natural Science Foundation of China (62176048)provided funding for this research 

主　　题：Interpretable graphics training visualization image segmentation left ventricle CNNs global average pooling 

摘      要：The interpretability of deep learning models has emerged as a compelling area in artificial intelligence *** safety criteria for medical imaging are highly stringent,and models are required for an ***,existing convolutional neural network solutions for left ventricular segmentation are viewed in terms of inputs and ***,the interpretability of CNNs has come into the *** medical imaging data are limited,many methods to fine-tune medical imaging models that are popular in transfer models have been built using massive public Image Net datasets by the transfer learning ***,this generates many unreliable parameters and makes it difficult to generate plausible explanations from these *** this study,we trained from scratch rather than relying on transfer learning,creating a novel interpretable approach for autonomously segmenting the left ventricle with a cardiac *** enhanced GPU training system implemented interpretable global average pooling for graphics using deep *** deep learning tasks were *** included data management,neural network architecture,and *** system monitored and analyzed the gradient changes of different layers with dynamic visualizations in real-time and selected the optimal deployment *** results demonstrated that the proposed method was feasible and efficient:the Dice coefficient reached 94.48%,and the accuracy reached 99.7%.It was found that no current transfer learning models could perform comparably to the ImageNet transfer learning *** model is lightweight and more convenient to deploy on mobile devices than transfer learning models.