End-to-End 2D Convolutional Neural Network Architecture for Lung Nodule Identification and Abnormal Detection in Cloud

作     者：Safdar Ali Saad Asad Zeeshan Asghar Atif Ali Dohyeun Kim 

作者机构：Department of Software Engineeringthe University of LahoreDefence Road CampusLahore55150Pakistan Department of Computer EngineeringJeju National UniversityAra CampusJeju City63243Korea 

出 版 物：《Computers, Materials & Continua》 (计算机、材料和连续体（英文）)

年 卷 期：2023年第75卷第4期

页      面：461-475页

核心收录：

学科分类：12[管理学] 08[工学] 0831[工学-生物医学工程（可授工学、理学、医学学位）] 0710[理学-生物学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 1002[医学-临床医学] 1001[医学-基础医学(可授医学、理学学位)] 081104[工学-模式识别与智能系统] 0805[工学-材料科学与工程（可授工学、理学学位）] 0835[工学-软件工程] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 0702[理学-物理学] 

基　　金：supported this research through the National Research Foundation of Korea (NRF)funded by the Ministry of Science,ICT (2019M3F2A1073387) this work was supported by the Institute for Information&communications Technology Promotion (IITP) (NO.2022-0-00980 Cooperative Intelligence Framework of Scene Perception for Autonomous IoT Device) 

主　　题：Convolutional neural networks medical image processing lung nodule identification data imbalance deep learning 

摘      要：The extent of the peril associated with cancer can be perceivedfrom the lack of treatment, ineffective early diagnosis techniques, and mostimportantly its fatality rate. Globally, cancer is the second leading cause ofdeath and among over a hundred types of cancer;lung cancer is the secondmost common type of cancer as well as the leading cause of cancer-relateddeaths. Anyhow, an accurate lung cancer diagnosis in a timely manner canelevate the likelihood of survival by a noticeable margin and medical imagingis a prevalent manner of cancer diagnosis since it is easily accessible to peoplearound the globe. Nonetheless, this is not eminently efficacious consideringhuman inspection of medical images can yield a high false positive rate. Ineffectiveand inefficient diagnosis is a crucial reason for such a high mortalityrate for this malady. However, the conspicuous advancements in deep learningand artificial intelligence have stimulated the development of exceedinglyprecise diagnosis systems. The development and performance of these systemsrely prominently on the data that is used to train these systems. A standardproblem witnessed in publicly available medical image datasets is the severeimbalance of data between different classes. This grave imbalance of data canmake a deep learning model biased towards the dominant class and unableto generalize. This study aims to present an end-to-end convolutional neuralnetwork that can accurately differentiate lung nodules from non-nodules andreduce the false positive rate to a bare minimum. To tackle the problem ofdata imbalance, we oversampled the data by transforming available images inthe minority class. The average false positive rate in the proposed method isa mere 1.5 percent. However, the average false negative rate is 31.76 *** proposed neural network has 68.66 percent sensitivity and 98.42 percentspecificity.