Comparison of efficacy of lung ultrasound and chest X-ray in diagnosing pulmonary edema and pleural effusion in ICU patients: A single centre, prospective, observational study

Background and Aims While chest X-ray (CXR) has been a conventional tool in intensive care units (ICUs) to identify lung pathologies, computed tomography (CT) scan remains the gold standard. Use of lung ultrasound (LUS) in resource-rich ICUs is still under investigation. The present study compares the utility of LUS to that of CXR in identifying pulmonary edema and pleural effusion in ICU patients. In addition, consolidation and pneumothorax were analyzed as secondary outcome measures. Material and Methods This is a prospective, single centric, observational study. Patients admitted in ICU were examined for lung pathologies, using LUS by a trained intensivist;and CXR done within 4 hours of each other. The final diagnosis was ascertained by an independent senior radiologist, based on the complete medical chart including clinical findings and the results of thoracic CT, if available. The results were compared and analyzed. Results Sensitivity, specificity and diagnostic accuracy of LUS was 95%, 94.4%, 94.67% for pleural effusion;and 98.33%, 97.78%, 98.00% for pulmonary edema respectively. Corresponding values with CXR were 48.33%, 76.67%, 65.33% for pleural effusion;and 36.67%, 82.22% and 64.00% for pulmonary edema respectively. Sensitivity, specificity and diagnostic accuracy of LUS was 91.30%, 96.85%, 96.00% for consolidation;and 100.00%, 79.02%, 80.00% for pneumothorax respectively. Corresponding values with CXR were 60.87%, 81.10%, 78.00% for consolidation;and 71.3%, 97.20%, 96.00% for pneumothorax respectively. Conclusion LUS has better diagnostic accuracy in diagnosis of pleural effusion and pulmonary edema when compared with CXR and is thus recommended as an effective alternative for diagnosis of these conditions in acute care settings. Our study recommends that a thoracic CT scan can be avoided in most of such cases.

M^(3)Res-Transformer:新冠肺炎胸部X-ray图像识别模型

新冠肺炎(COVID-19)自爆发以来严重影响人类生命健康,近年来残差神经网络广泛应用于COVID-19识别任务中,辅助医生快速地诊断COVID-19患者,但是COVID-19图像病变区域形状复杂、大小不一,与周围组织的边界模糊,导致网络难以提取有效特征.本文针对上述问题,提出一种M^(3)Res-Transformer的新冠肺炎胸部X-ray图像识别模型,采用Res-Transformer作为模型的主干网络,结合ResNet和ViT,有效地整合局部病变特征和全局特征;设计混合残差注意力模块(mixed residual attention Module,mraM),同时考虑通道和空间位置的相互依赖性,增强网络的特征表达能力;为了增大感受野,提取多尺度特征,通过叠加具有不同扩张率的扩张卷积构造多尺度扩张残差模块(multiscale dilated residual Module,mdrM),根据不同层次特征尺度的差异,使用3个逐渐收缩尺度的mdrM进行多尺度特征提取;提出上下文交叉感知模块(contextual cross-awareness Module,ccaM),使用深层特征中的语义信息来引导浅层特征,然后将浅层特征中的空间信息嵌入深层特征中,采用交叉加权注意力机制高效聚合深层和浅层特征,获得更丰富的上下文信息.为了验证本文所提模型的有效性,在新冠肺炎胸部X-ray图像数据集上进行实验,与先进的CNN分类模型、融合不同注意力机制的ResNet50模型、基于Transformer的分类模型对比以及消融实验.结果表明,本文所提模型的Acc、Pre、Rec、F1-Score与Spe指标分别为96.33%、96.36%、96.33%、96.35%与96.26%,在COVID-19胸部X-ray图像识别任务中有效提升了识别精度,并通过可视化方法对其进行进一步验证,为COVID-19的辅助诊断提供重要的参考价值.

Transfer Learning Approach to Classify the X-Ray Image that Corresponds to Corona Disease Using ResNet50 Pre-Trained by ChexNet

The COVID-19 pandemic has had a widespread negative impact globally. It shares symptoms with other respiratory illnesses such as pneumonia and influenza, making rapid and accurate diagnosis essential to treat individuals and halt further transmission. X-ray imaging of the lungs is one of the most reliable diagnostic tools. Utilizing deep learning, we can train models to recognize the signs of infection, thus aiding in the identification of COVID-19 cases. For our project, we developed a deep learning model utilizing the ResNet50 architecture, pre-trained with ImageNet and CheXNet datasets. We tackled the challenge of an imbalanced dataset, the CoronaHack Chest X-Ray dataset provided by Kaggle, through both binary and multi-class classification approaches. Additionally, we evaluated the performance impact of using Focal loss versus Cross-entropy loss in our model.

COVID-19 Detection from Chest X-Ray Images Using Convolutional Neural Network Approach

COVID-19 is a respiratory illness caused by the SARS-CoV-2 virus, first identified in 2019. The primary mode of transmission is through respiratory droplets when an infected person coughs or sneezes. Symptoms can range from mild to severe, and timely diagnosis is crucial for effective treatment. Chest X-Ray imaging is one diagnostic tool used for COVID-19, and a Convolutional Neural Network (CNN) is a popular technique for image classification. In this study, we proposed a CNN-based approach for detecting COVID-19 in chest X-Ray images. The model was trained on a dataset containing both COVID-19 positive and negative cases and evaluated on a separate test dataset to measure its accuracy. Our results indicated that the CNN approach could accurately detect COVID-19 in chest X-Ray images, with an overall accuracy of 97%. This approach could potentially serve as an early diagnostic tool to reduce the spread of the virus.

基于深度可变形卷积的胸部X-ray病灶检测与应用

深度学习算法的不断优化为临床诊治提供了理论支持,推动了我国医疗事业的发展。为了降低肺部疾病对个体的损害,研究提出了基于深度可变形卷积的胸部X-ray病灶检测模型,在该模型中,采用深度可变形卷积来优化传统卷积层,最后通过仿真实验和应用分析来验证其有效性。在结果中显示,仿真中的深度可变现卷积损失值降低至0.13,并且在检测模型的应用中显示其检测准确率显著高于卷积神经网络和BP神经网络。以上结果表明,采用深度可变形卷积来优化胸部X-ray病灶检测具有有效性,对我国临床诊治发展具有显著参考价值。

结合多标签共现关系的深度哈希胸部X光影像检索

为使哈希检索模型兼具强判别力和小量化误差,现有方法通常需要优化多个损失函数,导致模型训练存在困难.通过最大化连续哈希码与对应正交化二值码之间相似性,尽管只需优化单个损失函数就能实现这一目标,然而在多标签图像检索任务中,这类方法忽视了标签间的语义相关性,导致检索性能下降.本文提出一种基于标签共现的深度哈希检索算法.首先通过挖掘阳性疾病标签的共现信息,利用图卷积神经网络建模多标签共现关系,动态生成哈希目标.其次,通过引入标签平滑交叉熵损失函数,进一步增强图像哈希码与标签哈希目标的一致性.在胸片数据集上的实验结果表明,方法在关键性能指标上优于同类算法,验证了建模多标签共现关系对提升模型性能的重要性.

Diagnostic Value of the Thoracic Ultrasonography Compared to Conventional Chest X-Rays in Pneumonia for Children between 0 to 15 Years: Case Study in Two Hospitals in Yaoundé

Introduction: The diagnosis of pneumonia is usually made based on clinical manifestations and chest X-ray. The use of ultrasound in detecting pulmonary diseases in general, and especially consolidation syndrome has been demonstrated. The objective of this study was to determine the accuracy of thoracic ultrasound compared to chest X-ray in the diagnosis of infectious pneumonia in children. Methods: Children between 0 to 15 years were included in our study. The lung ultrasound results obtained were compared with those of the chest X-ray used as the reference. Our data were introduced into the EpiInfo 3.5.4 software and analyzed with the EpiInfo 3.5.4 and IBMSPSS Statistics version 20.0 softwares. Microsoft Office Excel 2016 was used to produce Charts. Continuous quantitative variables were presented. Cohen’s Kappa concordance test was applied with confidence interval of 95%. Results: 52 children were enrolled in the study. In imaging, the dominant sign was consolidation syndrome (75.0%) of cases by chest radiography, and in 78.8% of cases by lung ultrasound (p Conclusion: Our study demonstrated that lung echography is a non-ionizing and reliable tool in the diagnosis of childhood’s pneumonia.

改进残差网络的医学X射线影像分类与加密传输系统

随着X射线影像在医疗诊断领域的快速发展,在大量胸腔X射线影像产出的情况下,医生根据经验进行人为判断分析的方式已不能满足诊断效率与准确率的需求,高效率、高准确率处理批量X射线影像分类的问题亟待解决;通过改进残差网络对胸腔X射线影像进行分类,并设计一种加密传输系统,可有效解决上述问题;利用对X射线影像进行基于马尔可夫随机场的图像增强,再采用深层信息挖掘能力较强的ResNet50作为主干网络,增加自注意力机制并采用CELU激活函数优化;经Kaggle整合数据集实验测试结果表明,在保证分类准确性的前提下,分类的召回率从0.432提升到0.652;同时,系统采用基于Logistic混沌序列的图像加密算法,保证了远程医疗诊断的私密性,满足实际远程医疗场景的应用需求。

肺炎支原体肺炎患儿外周血miR-106a、miR-20a表达与胸部X线表现及炎症的相关性

目的:探讨肺炎支原体肺炎(MPP)患儿外周血微小RNA(miR)-106a、miR-20a表达与胸部X线(CXR)表现及炎症的相关性。方法:选取2020年1月至2021年12月本院收治的393例MPP患儿作为研究对象。采用全自动血细胞分析仪检测白细胞计数、红细胞计数、血小板计数、中性分叶核白细胞等,免疫透射比浊法检测C反应蛋白(CRP)水平。采用实时定量聚合酶链反应法检测外周血miR-106a和miR-20a表达。所有患儿均在入组后接受CXR检查。结果:MPP患儿外周血miR-106a和miR-20a的表达水平分别为1.00(0.61,2.11)、1.06(0.73,1.67)。根据中位值分组,与miR-106a和miR-20a低表达组相比,高表达组患儿CRP水平和肺实变患儿比例明显更高(P<0.05);miR-20a高表达组患儿胸腔积液患儿比例也显著高于低表达组(P<0.05)。MPP患儿外周血miR-106a和miR-20a与CRP呈显著正相关(r值分别为0.212、0.230,均P<0.001)。肺实变组外周血miR-106a和miR-20a表达水平均显著高于非肺实变组(P<0.05)。经ROC曲线分析,外周血miR-106a和miR-20a预测MPP肺实变的曲线下面积(AUC)为0.811(95%CI:0.767~0.855)、0.807(95%CI:0.762~0.852),灵敏度分别为75.3%和77.4%,特异度分别为76.5%和76.4%,对应的截断值为1.20和1.18。Logistic回归分析显示,外周血miR-106a和miR-20a高表达是MPP肺实变的独立影响因素(P<0.05)。结论:MPP患儿外周血miR-106a和miR-20a过表达与CXR严重表现和炎症加重有关,两者可作为指示MPP严重程度和炎症进展的标志物。

基于ConvNeXt模型的胸部X线图像的疾病分类与可视化

目的 胸部X线是临床实践中常见的胸部疾病筛查和诊断方式。由于放射科医生长时间阅片容易视觉疲劳以及医疗资源分配不均衡的问题,导致误诊和漏诊的情况时有发生。针对这一问题,本研究运用深度学习技术,提出了一个基于ConvNeXt模型的胸部X线图像的疾病检测方法,旨在提高胸部疾病诊断准确度、减轻误诊风险并提高医生工作效率。方法 利用大规模公开胸部X线图像数据集ChestX-ray14训练ConvNeXt模型,该模型在ResNet模型的基础上,融合了视觉Transformer结构的优势,可以有效提高模型的特征提取和识别能力,同时以AUC(ROC曲线下方的面积)作为模型性能的评价指标,与已有的分类模型CheXNet、ResNet及Swin Transformer进行了对比。此外,通过引入Grad-CAM可视化方法,利用卷积神经网络特征图的梯度信息生成胸部X线图像的类激活热力图,实现对病灶区域的定位,从而提高医生的诊断效率。结果 基于ConvNeXt模型的诊断方法在识别14种胸部疾病时平均AUC值可达0.842,特别在识别积液(AUC值为0.883)、水肿(AUC值为0.902)和疝气(AUC值为0.942)等疾病时表现较为令人满意。结论 本文提出的方法在胸部X线图像的疾病检测中具有较好的性能,是一种对胸部X线图像进行胸部疾病诊断进而协助医生提高工作效率的有益尝试。

Pertinence of Children’s Chest X-Ray Request Form and Practice at the Regional Hospital of Ngaoundere Cameroon

Background: Chest X-ray is frequently performed for evaluation of chest disease in both adults and children. Children are more exposed to the adverse effects of radiation as compared to adults. During our daily practice, we noticed that most of children’s chest X-ray results were normal. Purpose: This study aimed to evaluate the indications, the technic, the irradiation and the result of chest X-rays in children in order to know if the practice of these X-rays was relevant. Method: Cross-sectional and descriptive study conducted at the Imaging Regional Center of Ngaoundere from April to August 2017. A total number of 145 radiographs and 140 X-ray requests of 140 children were considered in this work. The conformity of the request were verified according to the recommendations of the National Agency for Accreditation and Health Evaluation in France (NAAHE), technical condition of realization and results were appreciated and the entrance surface dose (ESD) of the patients was estimated using a mathematical algorithm. Results: Children under 5 years (63.5%) were more represented in our study. The main indications were: cough (22.1%), suspicion of pneumonia (16.4%) and bronchitis (15.7%). No indication was mentioned on 69.3% of the request forms. After confrontation to the “Guide for proper use of medical imaging examinations” (GPU), we only had 24% conformity of indications. 82.7% of the examinations required immobilization assistance by the parents. Most of the children were imaged in a standing-up position (82.9%) and the anterior-posterior view (77.9%) was more practiced. After the analysis of the pictures, 62% of them presented an optimal contrast, while 42.1% of X-ray were performed without beam collimation. 25 X-rays were repeated: 12 (48%) because of patient’s motion and 13 (52%) of mispositionning. After interpretation, 87 (62.14%) chest X-ray were normal. Main lesion observed were pneumonia (17.14%) followed by bronchopeumopathy (5.71%) and bronchitis (5%). The obtained ESD values were 0.11, 0.15 and 0.17 mGy respectively for the 0 - 1 year, 1 - 5 year and 5 - 10 year age groups;0.2 and 0.57 respectively for postero-anterior (PA) and lateral (LAT) view for the age group 10 - 15 years, which were slightly greater than the values in internationally published studies. Conclusion: The request for children chest X-ray is not relevant in terms of indication, technical conditions of realization and irradiation.

Usefulness of Chest Computed Tomography for Diagnosis of Idiopathic Pneumomediastinum with Negative Findings on Plain X-Ray

Idiopathic pneumomediastinum is rare in children. Few cases of patients with pneumomediastinum show negative findings on X-ray examination. Chest computed tomography (CT) was very useful for the diagnosis and evaluation of the extent of pneumomediastinum. We report here a case of idiopathic pneumomediastinum in a 15-year-old boy who exhibited no significant chest X-ray finding. The patient was referred to our institute for the further evaluation of pre-cordial pain and breathing difficulty. Precordial pain suddenly developed, when he was carrying a portable shrine on his shoulder (day of onset). He was admitted to another institute 3 days after onset because of worsening precordial pain. On admission, he presented with 98% saturation of hemoglobin in the peripheral blood under room air. Plain chest X-ray also revealed no abnormal findings. A half-dissolved gastrographin swallow showed no leakage of gastrographin from the pharynx and esophagus to the mediastinum, and no diverticulum within the esophagus. Plain chest CT revealed extensive emphysema around the trachea from the neck to the portion inferior to the carina of trachea. The patient was diagnosed with idiopathic pneumomediastinum because the cause was unclear. We decided to admit him to our institute under fasting conditions and rest. His symptoms improved 3 days after onset. The lesion had disap-peared 8 days after onset on chest CT. When young people experience precordial pain which increases on inspiration, we must consider pneumomediastinum in a differential diagnosis, and it is important to perform chest CT.

胸部X线与多层螺旋CT单独及联合检查对中央型肺癌的诊断价值

目的 探讨胸部X线与多层螺旋CT单独及联合检查对中央型肺癌的诊断价值。方法 135例疑似中央型肺癌患者均行胸部X线及多层螺旋CT检查,以病理检查结果为金标准,比较胸部X线和多层螺旋CT单独及联合检查诊断中央型肺癌的结果与病理检查结果的一致性以及对中央型肺癌的诊断价值,比较胸部X线和多层螺旋CT检查诊断中央型肺癌TNM分期结果与病理分期的符合率,比较不同病理类型中央型肺癌患者多层螺旋CT检查的图像特征。结果 病理检查结果显示,135例疑似中央型肺癌患者中,阳性96例,阴性39例。胸部X线和多层螺旋CT联合检查(Kappa=0.563)诊断中央型肺癌的结果与病理检查结果的一致性高于二者单独检查(Kappa=0.491、0.558)。胸部X线和多层螺旋CT联合检查诊断中央型肺癌的准确度、灵敏度、特异度、阳性预测值及阴性预测值均高于二者单独检查。以病理检查结果为金标准,多层螺旋CT和胸部X线检查诊断中央型肺癌Ⅰ、Ⅱ、Ⅲ期的符合率比较,差异均无统计学意义(P﹥0.05)。结论 胸部X线和多层螺旋CT联合检查对中央型肺癌患者的诊断价值较高。多层螺旋CT检查能区分和鉴别不同病理类型的中央型肺癌。

Chest X-rays in detecting injuries caused by blunt trauma

BACKGROUND:The appropriate sequence of different imagings and indications of thoracic computed tomography(TCT)in evaluating chest trauma have not yet been clarified at present.The current study was undertaken to determine the value of chest X-ray(CXR)in detecting chest injuries in patients with blunt trauma.METHODS:A total of 447 patients with blunt thoracic trauma who had been admitted to the emergency department(ED)in the period of 2009–2013 were retrospectively reviewed.The patients met inclusion criteria(age>8 years,blunt injury to the chest,hemodynamically stable,and neurologically intact)and underwent both TCT and upright CXR in the ED.Radiological imagings were re-interpreted after they were collected from the hospital database by two skilled radiologists.RESULTS:Of the 447 patients,309(69.1%)were male.The mean age of the 447 patients was 39.5±19.2(range 9 and 87 years).158(35.3%)patients were injured in motor vehicle accidents(MVA).CXR showed the highest sensitivity in detecting clavicle fractures[95%CI 78.3(63.6–89)]but the lowest in pneuomediastinum[95%CI 11.8(1.5–36.4)].The specificity of CXR was close to 100%in detecting a wide array of entities.CONCLUSION:CXR remains to be the first choice in hemodynamically unstable patients with blunt chest trauma.Moreover,stable patients with normal CXR are candidates who should undergo TCT if significant injury has not been ruled out.

Chest Radiography: General Practitioners’ Compliance with Recommendations

Introduction: Chest radiography is the most frequently prescribed imaging test in general practice in France. We aimed to assess the extent to which general practitioners follow the recommendations of the French National Authority for Health in prescribing chest radiography. Methodology: We conducted a retrospective analysis study, in two radiology centers belonging to the same group in Saint-Omer and Aire-sur-la-Lys, of requests for chest radiography sent by general practitioners over the winter period between December 22, 2013, and March 21, 2014, for patients aged over 18 years. Results: One hundred and seventy-seven requests for chest X-rays were analyzed, 71.75% of which complied with recommendations. The most frequent reason was the search for bronchopulmonary infection, accounting for 70.08% of prescriptions, followed by 11.2% for requests to rule out pulmonary neoplasia, whereas the latter reason did not comply with recommendations. Chest X-rays contributed to a positive diagnosis in 28.81% of cases. The positive diagnosis was given by 36.22% of the recommended chest X-rays, versus 10% for those not recommended. Conclusion: In most cases, general practitioners follow HAS recommendations for prescribing chest X-rays. Non-recommended chest X-rays do not appear to make a major contribution to diagnosis or patient management, confirming the value of following the recommendations of the French National Authority for Health.

The Role of Chest X-Ray in Monitoring Lung Changes among COVID-19 Patients in Gaza Strip

Objective: To investigate the time course and findings severity of COVID-19 infection at chest radiography based on a 6-point radiological severity score, and correlates these with patients’ age and gender. Methods: This is a retrospective study of COVID-19 patients who were admitted at European Gaza Hospital and evaluated between October 6, 2020, and November 30, 2020. Baseline and serial chest radiographs, up to 4 images per patient, were reviewed and assessed for predominant pattern, side, and location of lung opacity. Utilized a 6-point scoring system, which divides the chest X-ray into 6 zones, to assess chest X-ray changes and correlate them with the severity of infection, age, and gender of patients. Results: The study included 136 COVID-19 patients: (51/136, 37%) were males and (85/136, 62.5%) were females, while age ranged from 7 months to 90 years with a mean age of 41.7 ± (19.5) years. Negative Chest x-rays were more observed than positive images. Ground-glass opacity was the most frequent pattern with a decreasing trend from 1st to 4th chest X-ray (from 33.8% to 3.7%), followed by consolidation (from 16.2% to 2.9%). Also, the commonest pattern of opacity was seen in peripheral areas (27/136, 19.9%), lower zone location (23/136, 16.9%), and bilateral opacity involvement (43/136;31.6%). No significant correlation was noticed between the patient’s gender, age, and severity score (P > 0.05). Conclusions: The 6-point chest X-ray severity score as a predictive tool in assessing the severity due to provide an assessment of the progression or regression pathway.

A Novel Method for Automated Lung Region Segmentation in Chest X-Ray Images

Detecting and segmenting the lung regions in chest X-ray images is an important part in artificial intelligence-based computer-aided diagnosis/detection (AI-CAD) systems for chest radiography. However, if the chest X-ray images themselves are used as training data for the AI-CAD system, the system might learn the irrelevant image-based information resulting in the decrease of system’s performance. In this study, we propose a lung region segmentation method that can automatically remove the shoulder and scapula regions, mediastinum, and diaphragm regions in advance from various chest X-ray images to be used as learning data. The proposed method consists of three main steps. First, employ the simple linear iterative clustering algorithm, the lazy snapping technique and local entropy filter to generate an entropy map. Second, apply morphological operations to the entropy map to obtain a lung mask. Third, perform automated segmentation of the lung field using the obtained mask. A total of 30 images were used for the experiments. In order to verify the effectiveness of the proposed method, two other texture maps, namely, the maps created from the standard deviation filtering and the range filtering, were used for comparison. As a result, the proposed method using the entropy map was able to appropriately remove the unnecessary regions. In addition, this method was able to remove the markers present in the image, but the other two methods could not. The experimental results have revealed that our proposed method is a highly generalizable and useful algorithm. We believe that this method might act an important role to enhance the performance of AI-CAD systems for chest X-ray images.

基于卷积神经网络的胸部X射线影像肺炎疾病分类研究

目的:自动检测肺炎以及区分2019冠状病毒病(COVID-19)和非COVID-19肺炎,旨在提高整体分类准确性.方法:数据集来自Kaggle存储库.使用编程环境MATLAB 2021a对所提出的模型进行开发和训练.该模型使用2913张胸部X线片图像(其中正常1005张,COVID-19900张,病毒性肺炎1 008张)进行训练,还使用从数据集中随机选择的一些未使用的胸部图像进行评估,并与现有深度学习方法相比较.结果:该模型在训练集上的平均准确率、召回率和精准率分别为0.989,0.983和0.984.此外,平均假阳性率和假阴性率分别为0.009和0.017.在验证集上,平均准确率、召回率和精准率分别为0.978、0.967和0.967.准确预测了未用于训练也未用于验证的图像60例(每类20例)中的58例.结论:利用卷积神经网络对胸部X线图像进行分类可以辅助放射科医师且能够减少他们之间可能由经验引起的图像解释的差异性.

融合改进变分自编码器与影像组学的X光片肺部疾病筛查算法

计算机辅助技术在肺部疾病筛查方面已经取得显著成效,然而现有研究大多面向已知类型的疾病进行建模,对未知类型疾病极易带来误诊及漏诊风险,且主要以追求高准确率为目标,对误诊及漏诊未加以约束,导致其难以应用于实际临床场景。针对以上问题,该文提出更适用于临床的计算机辅助肺部疾病筛查目标,即保证零漏诊率的同时降低误诊率。为完成上述肺部疾病筛查目标,该文基于单类别分类思想提出改进变分自编码网络对肺部疾病初筛,并提取X光片图像的深度编码特征,接着,融合基于医生经验的影像组学特征以及深度学习特征之间的互补优势,构建一个集成学习模型,最终完成肺部疾病的筛查。在仅有正常X光片图像参与训练的情况下,提升了所构建模型的分类效果,降低了模型的漏诊率。实验结果AUC值为0.9848±0.0023,漏诊率为0时,误诊率降低至0.1498±0.0057,证明该方法可以有效达到该文的肺部疾病筛查目标。与此同时,对比了所构建的集成模型以及单独的深度学习模型的筛查效果,发现集成模型明显优于深度学习模型,进一步凸显了融合医生经验的有效性。

WCF-MobileNetV3:轻量型新冠肺炎CXR图像识别网络

为了对新型冠状病毒引发的肺炎胸部X光(chest X-Ray,CXR)图像进行准确且快速的识别,提出了一种基于加权通道筛选(weighted channel filter,WCF)的轻量级模型WCF-MobileNetV3。将轻量级的MobileNetV3-small作为主干网络,并针对CXR图像样本类间差异小、难以提取区分性特征的问题,提出了WCF模块。提取输入特征图的高维与低维通道特征权重;采取加权随机抽样的方式生成高维与低维特征通道掩膜,将高维、低维的权重融合,并利用掩膜对融合后的权重进行通道筛选;将权重赋给输入特征图,实现通道特征增强。在Chest-X-Ray Image与COVID-19 Chest X-Ray Image Repository数据集上进行了实验,结果表明:WCF-MobileNetV3对新冠肺炎CXR图像识别的准确率、精确率、灵敏度分别为97.93%、98.64%、97.19%。与其他新冠肺炎识别算法相比,WCF-MobileNetV3能够准确且高效地识别新冠肺炎CXR图像,具有更好的识别性能。