Pub Date : 2024-10-31DOI: 10.1016/j.ejro.2024.100609
Na Feng , Shanshan Zhao , Kai Wang , Peizhe Chen , Yunpeng Wang , Yuan Gao , Zhengping Wang , Yidan Lu , Chen Chen , Jincao Yao , Zhikai Lei , Dong Xu
Objective
To develop a ultrasound images based dual-channel deep learning model to achieve accurate early diagnosis of thyroid nodules less than 1 cm.
Methods
A dual-channel deep learning model called thyroid nodule transformer network (TNT-Net) was proposed. The model has two input channels for transverse and longitudinal ultrasound images of thyroid nodules, respectively. A total of 9649 nodules from 8455 patients across five hospitals were retrospectively collected. The data were divided into a training set (8453 nodules, 7369 patients), an internal test set (565 nodules, 512 patients), and an external test set (631 nodules, 574 patients).
Results
TNT-Net achieved an area under the curve (AUC) of 0.953 (95 % confidence interval (CI): 0.934, 0.969) on the internal test set and 0.941 (95 % CI: 0.921, 0.957) on the external test set, significantly outperforming traditional deep convolutional neural network models and single-channel swin transformer model, whose AUCs ranged from 0.800 (95 % CI: 0.759, 0.837) to 0.856 (95 % CI: 0.819, 0.881). Furthermore, feature heatmap visualization showed that TNT-Net could extract richer and more energetic malignant nodule patterns.
Conclusion
The proposed TNT-Net model significantly improved the recognition capability for thyroid nodules with size less than 1 cm. This model has the potential to reduce overdiagnosis and overtreatment of such nodules, providing essential support for precise management of thyroid nodules while complementing fine-needle aspiration biopsy.
{"title":"Deep learning model for diagnosis of thyroid nodules with size less than 1 cm: A multicenter, retrospective study","authors":"Na Feng , Shanshan Zhao , Kai Wang , Peizhe Chen , Yunpeng Wang , Yuan Gao , Zhengping Wang , Yidan Lu , Chen Chen , Jincao Yao , Zhikai Lei , Dong Xu","doi":"10.1016/j.ejro.2024.100609","DOIUrl":"10.1016/j.ejro.2024.100609","url":null,"abstract":"<div><h3>Objective</h3><div>To develop a ultrasound images based dual-channel deep learning model to achieve accurate early diagnosis of thyroid nodules less than 1 cm.</div></div><div><h3>Methods</h3><div>A dual-channel deep learning model called thyroid nodule transformer network (TNT-Net) was proposed. The model has two input channels for transverse and longitudinal ultrasound images of thyroid nodules, respectively. A total of 9649 nodules from 8455 patients across five hospitals were retrospectively collected. The data were divided into a training set (8453 nodules, 7369 patients), an internal test set (565 nodules, 512 patients), and an external test set (631 nodules, 574 patients).</div></div><div><h3>Results</h3><div>TNT-Net achieved an area under the curve (AUC) of 0.953 (95 % confidence interval (CI): 0.934, 0.969) on the internal test set and 0.941 (95 % CI: 0.921, 0.957) on the external test set, significantly outperforming traditional deep convolutional neural network models and single-channel swin transformer model, whose AUCs ranged from 0.800 (95 % CI: 0.759, 0.837) to 0.856 (95 % CI: 0.819, 0.881). Furthermore, feature heatmap visualization showed that TNT-Net could extract richer and more energetic malignant nodule patterns.</div></div><div><h3>Conclusion</h3><div>The proposed TNT-Net model significantly improved the recognition capability for thyroid nodules with size less than 1 cm. This model has the potential to reduce overdiagnosis and overtreatment of such nodules, providing essential support for precise management of thyroid nodules while complementing fine-needle aspiration biopsy.</div></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":null,"pages":null},"PeriodicalIF":1.8,"publicationDate":"2024-10-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142561079","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-29DOI: 10.1016/j.ejro.2024.100608
Ruiting Wang , Lianting Zhong , Pingyi Zhu , Xianpan Pan , Lei Chen , Jianjun Zhou , Yuqin Ding
Purpose
We aim to develop an MRI-based radiomics model to improve the accuracy of differentiating non-ccRCC from benign renal tumors preoperatively.
Methods
The retrospective study included 195 patients with pathologically confirmed renal tumors (134 non-ccRCCs and 61 benign renal tumors) who underwent preoperative renal mass protocol MRI examinations. The patients were divided into a training set (n = 136) and test set (n = 59). Simple t-test and the Least Absolute Shrink and Selection Operator (LASSO) were used to select the most valuable features and the rad-scores of them were calculated. The clinicoradiologic models, single-sequence radiomics models, multi-sequence radiomics models and combined models for differentiation were constructed with 2 classifiers (support vector machine (SVM), logistic regression (LR)) in the training set and used for differentiation in the test set. Ten-fold cross validation was applied to obtain the optimal hyperparameters of the models. The performances of the models were evaluated by the area under the receiver operating characteristic (ROC) curve (AUC). Delong’s test was performed to compare the performances of models.
Results
After univariate and multivariate logistic regression analysis, the independent risk factors to differentiate non-ccRCC from benign renal tumors were selected as follows: age, tumor region, hemorrhage, pseudocapsule and enhancement degree. Among the 14 machine learning classification models constructed, the combined model with LR has the highest efficiency in differentiating non-ccRCC from benign renal tumors. The AUC in the training set is 0.964, and the accuracy is 0.919. The AUC in the test set is 0.936, and the accuracy is 0.864.
Conclusion
The MRI-based radiomics machine learning is feasible to differentiate non-ccRCC from benign renal tumors, which could improve the accuracy of clinical diagnosis.
{"title":"MRI-based radiomics machine learning model to differentiate non-clear cell renal cell carcinoma from benign renal tumors","authors":"Ruiting Wang , Lianting Zhong , Pingyi Zhu , Xianpan Pan , Lei Chen , Jianjun Zhou , Yuqin Ding","doi":"10.1016/j.ejro.2024.100608","DOIUrl":"10.1016/j.ejro.2024.100608","url":null,"abstract":"<div><h3>Purpose</h3><div>We aim to develop an MRI-based radiomics model to improve the accuracy of differentiating non-ccRCC from benign renal tumors preoperatively.</div></div><div><h3>Methods</h3><div>The retrospective study included 195 patients with pathologically confirmed renal tumors (134 non-ccRCCs and 61 benign renal tumors) who underwent preoperative renal mass protocol MRI examinations. The patients were divided into a training set (n = 136) and test set (n = 59). Simple t-test and the Least Absolute Shrink and Selection Operator (LASSO) were used to select the most valuable features and the rad-scores of them were calculated. The clinicoradiologic models, single-sequence radiomics models, multi-sequence radiomics models and combined models for differentiation were constructed with 2 classifiers (support vector machine (SVM), logistic regression (LR)) in the training set and used for differentiation in the test set. Ten-fold cross validation was applied to obtain the optimal hyperparameters of the models. The performances of the models were evaluated by the area under the receiver operating characteristic (ROC) curve (AUC). Delong’s test was performed to compare the performances of models.</div></div><div><h3>Results</h3><div>After univariate and multivariate logistic regression analysis, the independent risk factors to differentiate non-ccRCC from benign renal tumors were selected as follows: age, tumor region, hemorrhage, pseudocapsule and enhancement degree. Among the 14 machine learning classification models constructed, the combined model with LR has the highest efficiency in differentiating non-ccRCC from benign renal tumors. The AUC in the training set is 0.964, and the accuracy is 0.919. The AUC in the test set is 0.936, and the accuracy is 0.864.</div></div><div><h3>Conclusion</h3><div>The MRI-based radiomics machine learning is feasible to differentiate non-ccRCC from benign renal tumors, which could improve the accuracy of clinical diagnosis.</div></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":null,"pages":null},"PeriodicalIF":1.8,"publicationDate":"2024-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142552679","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Chest radiographs (CXRs) are widely used to screen for infectious diseases like tuberculosis and COVID-19 among migrants. At such high-volume settings, manual CXR reporting is challenging and integrating artificial intelligence (AI) algorithms into the workflow help to rule out normal findings in minutes, allowing radiologists to focus on abnormal cases.
Methods
In this post-deployment study, all the CXRs acquired during the visa screening process across 33 centers in United Arab Emirates from January 2021 to June 2022 (18 months) were included. The qXR v2.1 chest X-ray interpretation software was used to classify the scans into normal and abnormal, and its agreement against radiologist was evaluated. Additionally, a digital survey was conducted among 20 healthcare professionals with prior AI experience to understand real-world implementation challenges and impact.
Results
The analysis of 1309,443 CXRs from 1309,431 patients (median age: 35 years; IQR [29–42]; 1030,071 males [78.7 %]) in this study revealed a Negative Predictive Value (NPV) of 99.92 % (95 % CI: 99.92, 99.93), Positive Predictive Value (PPV) of 5.06 % (95 % CI: 4.99, 5.13) and overall percent agreement of the AI with radiologists of 72.90 % (95 % CI: 72.82, 72.98). In the survey, majority (88.2 %) of the radiologists agreed to turnaround time reduction after AI integration, while 82 % suggested that the AI improved their diagnostic accuracy.
Discussion
In contrast with the existing studies, this research uses a substantially large data. A high NPV and satisfactory agreement with human readers indicate that AI can reliably identify normal CXRs, making it suitable for routine applications.
{"title":"Post-deployment performance of a deep learning algorithm for normal and abnormal chest X-ray classification: A study at visa screening centers in the United Arab Emirates","authors":"Amina Abdelqadir Mohamed AlJasmi , Hatem Ghonim , Mohyi Eldin Fahmy , Aswathy Nair , Shamie Kumar , Dennis Robert , Afrah Abdikarim Mohamed , Hany Abdou , Anumeha Srivastava , Bhargava Reddy","doi":"10.1016/j.ejro.2024.100606","DOIUrl":"10.1016/j.ejro.2024.100606","url":null,"abstract":"<div><h3>Background</h3><div>Chest radiographs (CXRs) are widely used to screen for infectious diseases like tuberculosis and COVID-19 among migrants. At such high-volume settings, manual CXR reporting is challenging and integrating artificial intelligence (AI) algorithms into the workflow help to rule out normal findings in minutes, allowing radiologists to focus on abnormal cases.</div></div><div><h3>Methods</h3><div>In this post-deployment study, all the CXRs acquired during the visa screening process across 33 centers in United Arab Emirates from January 2021 to June 2022 (18 months) were included. The qXR v2.1 chest X-ray interpretation software was used to classify the scans into normal and abnormal, and its agreement against radiologist was evaluated. Additionally, a digital survey was conducted among 20 healthcare professionals with prior AI experience to understand real-world implementation challenges and impact.</div></div><div><h3>Results</h3><div>The analysis of 1309,443 CXRs from 1309,431 patients (median age: 35 years; IQR [29–42]; 1030,071 males [78.7 %]) in this study revealed a Negative Predictive Value (NPV) of 99.92 % (95 % CI: 99.92, 99.93), Positive Predictive Value (PPV) of 5.06 % (95 % CI: 4.99, 5.13) and overall percent agreement of the AI with radiologists of 72.90 % (95 % CI: 72.82, 72.98). In the survey, majority (88.2 %) of the radiologists agreed to turnaround time reduction after AI integration, while 82 % suggested that the AI improved their diagnostic accuracy.</div></div><div><h3>Discussion</h3><div>In contrast with the existing studies, this research uses a substantially large data. A high NPV and satisfactory agreement with human readers indicate that AI can reliably identify normal CXRs, making it suitable for routine applications.</div></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":null,"pages":null},"PeriodicalIF":1.8,"publicationDate":"2024-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142534921","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-21DOI: 10.1016/j.ejro.2024.100607
Wenjiang Wang , Jiaojiao Li , Zimeng Wang , Yanjun Liu , Fei Yang , Shujun Cui
Purpose
To develop a multi-modal model combining multi-sequence breast MRI fusion radiomics and deep learning for the classification of benign and malignant breast lesions, to assist clinicians in better selecting treatment plans.
Methods
A total of 314 patients who underwent breast MRI examinations were included. They were randomly divided into training, validation, and test sets in a ratio of 7:1:2. Subsequently, features of T1-weighted images (T1WI), T2-weighted images (T2WI), and dynamic contrast-enhanced MRI (DCE-MRI) were extracted using the convolutional neural network ResNet50 for fusion, and then combined with radiomic features from the three sequences. The following models were established: T1 model, T2 model, DCE model, DCE_T1_T2 model, and DCE_T1_T2_rad model. The performance of the models was evaluated by the area under the receiver operating characteristic (ROC) curve (AUC), accuracy, sensitivity, specificity, positive predictive value, and negative predictive value. The differences between the DCE_T1_T2_rad model and the other four models were compared using the Delong test, with a P-value < 0.05 considered statistically significant.
Results
The five models established in this study performed well, with AUC values of 0.53 for the T1 model, 0.62 for the T2 model, 0.79 for the DCE model, 0.94 for the DCE_T1_T2 model, and 0.98 for the DCE_T1_T2_rad model. The DCE_T1_T2_rad model showed statistically significant differences (P < 0.05) compared to the other four models.
Conclusion
The use of a multi-modal model combining multi-sequence breast MRI fusion radiomics and deep learning can effectively improve the diagnostic performance of breast lesion classification.
{"title":"Study on the classification of benign and malignant breast lesions using a multi-sequence breast MRI fusion radiomics and deep learning model","authors":"Wenjiang Wang , Jiaojiao Li , Zimeng Wang , Yanjun Liu , Fei Yang , Shujun Cui","doi":"10.1016/j.ejro.2024.100607","DOIUrl":"10.1016/j.ejro.2024.100607","url":null,"abstract":"<div><h3>Purpose</h3><div>To develop a multi-modal model combining multi-sequence breast MRI fusion radiomics and deep learning for the classification of benign and malignant breast lesions, to assist clinicians in better selecting treatment plans.</div></div><div><h3>Methods</h3><div>A total of 314 patients who underwent breast MRI examinations were included. They were randomly divided into training, validation, and test sets in a ratio of 7:1:2. Subsequently, features of T1-weighted images (T1WI), T2-weighted images (T2WI), and dynamic contrast-enhanced MRI (DCE-MRI) were extracted using the convolutional neural network ResNet50 for fusion, and then combined with radiomic features from the three sequences. The following models were established: T1 model, T2 model, DCE model, DCE_T1_T2 model, and DCE_T1_T2_rad model. The performance of the models was evaluated by the area under the receiver operating characteristic (ROC) curve (AUC), accuracy, sensitivity, specificity, positive predictive value, and negative predictive value. The differences between the DCE_T1_T2_rad model and the other four models were compared using the Delong test, with a <em>P</em>-value < 0.05 considered statistically significant.</div></div><div><h3>Results</h3><div>The five models established in this study performed well, with AUC values of 0.53 for the T1 model, 0.62 for the T2 model, 0.79 for the DCE model, 0.94 for the DCE_T1_T2 model, and 0.98 for the DCE_T1_T2_rad model. The DCE_T1_T2_rad model showed statistically significant differences (<em>P</em> < 0.05) compared to the other four models.</div></div><div><h3>Conclusion</h3><div>The use of a multi-modal model combining multi-sequence breast MRI fusion radiomics and deep learning can effectively improve the diagnostic performance of breast lesion classification.</div></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":null,"pages":null},"PeriodicalIF":1.8,"publicationDate":"2024-10-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142534920","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-19DOI: 10.1016/j.ejro.2024.100605
Yi Xiang Tay , Marcus EH Ong , Shane J. Foley , Robert Chun Chen , Lai Peng Chan , Ronan Killeen , May San Mak , Jonathan P. McNulty , Kularatna Sanjeewa
Objectives
There is a lack of clear and consistent cost reporting for cost-effectiveness analysis in radiology. Estimates are often obtained using costing derived from hospital charge records. This study aims to evaluate the accuracy of hospital charge records compared to a Singapore hospital's true diagnostic imaging costs.
Methods
A seven-step process involving a bottom-up micro-costing approach was devised and followed to calculate the cost of imaging using actual data from a clinical setting. We retrieved electronic data from a random sample of 96 emergency department patients who had CT brain, CT and X-ray cervical spine, and X-ray lumbar spine performed to calculate the parameters required for cost estimation. We adjusted imaging duration and number of performing personnel to account for variations.
Results
Our approach determined the average cost for the following imaging procedures: CT brain (€154.00), CT and X-ray cervical spine (€177.14 and €68.22), and X-ray lumbar spine (€79.85). We found that the true cost of both conventional radiography procedures was marginally higher than the subsidized patient charge, and all costs were slightly lower than the private patient charge except for X-ray lumbar spine (€73.49 vs.€79.85). We identified larger differences in cost for both CT procedures and smaller differences in cost for conventional radiography procedures, depending on the patient's private or subsidized status. For private status, the differences were: CT brain (Min: €194.20; Max: €264.40), CT cervical spine (Min: €219.54; Max: €399.05), X-ray cervical spine (Min: €5.27; Max: €61.94), and X-ray lumbar spine (Min: €6.36; Max: €108.04), while for subsidized status, the differences were: CT brain (Min: €7.56; Max: €62.64), CT cervical spine (Min: €47.02; Max: €132.49), X-ray cervical spine (Min: €15.88; Max: €103.44), and X-ray lumbar spine (Min: €13.66; Max: €149.44). Considering examination duration and the number of personnel engaged in a procedure, there were significant variations in the minimum, average, and maximum imaging costs.
Conclusion
There is a modest gap between hospital charges and actual costs, and we must therefore exercise caution and recognize the limitations of utilizing hospital charge records as absolute metrics for cost-effectiveness analysis. Our detailed approach can potentially enable more accurate imaging cost determination for future studies.
目的放射学成本效益分析缺乏清晰一致的成本报告。估算通常使用医院收费记录中的成本计算。本研究旨在评估医院收费记录与新加坡医院真实影像诊断成本相比的准确性。方法我们设计了一个包含自下而上的微观成本计算方法的七步流程，并利用临床环境中的实际数据计算影像成本。我们从随机抽样的 96 名急诊科患者中获取了电子数据，这些患者分别接受了脑部 CT、颈椎 CT 和 X 光检查以及腰椎 X 光检查，从而计算出成本估算所需的参数。我们对成像持续时间和执行人员数量进行了调整，以考虑到差异。结果我们的方法确定了以下成像程序的平均成本：我们的方法确定了以下成像程序的平均成本：脑部 CT（154.00 欧元）、颈椎 CT 和 X 光（177.14 欧元和 68.22 欧元）以及腰椎 X 光（79.85 欧元）。我们发现，这两项常规放射检查的实际费用略高于受资助病人的费用，除腰椎 X 光检查（73.49 欧元对 79.85 欧元）外，其他费用均略低于私人病人的费用。我们发现，根据患者的私立或补贴身份，两种 CT 程序的成本差异较大，而传统放射程序的成本差异较小。就私人身份而言，差异如下脑部 CT（最少：194.20 欧元；最多：264.40 欧元）、颈椎 CT（最少：219.54 欧元；最多：399.05 欧元）、颈椎 X 光（最少：5.27 欧元；最多：61.94 欧元）和腰椎 X 光（最少：6.36 欧元；最多：108.04 欧元）；而对于受补贴的患者，差异则为：脑部 CT（最少：7.27 欧元；最多：61.94 欧元）和腰椎 X 光（最少：6.36 欧元；最多：108.04 欧元）：脑部 CT（最低：7.56 欧元；最高：62.64 欧元）、颈椎 CT（最低：47.02 欧元；最高：132.49 欧元）、颈椎 X 光（最低：15.88 欧元；最高：103.44 欧元）和腰椎 X 光（最低：13.66 欧元；最高：149.44 欧元）。结论医院收费与实际成本之间存在一定差距，因此我们必须谨慎行事，并认识到利用医院收费记录作为成本效益分析绝对指标的局限性。我们的详细方法有可能为未来的研究提供更准确的成像成本测定。
{"title":"True cost estimation of common imaging procedures for cost-effectiveness analysis - insights from a Singapore hospital emergency department","authors":"Yi Xiang Tay , Marcus EH Ong , Shane J. Foley , Robert Chun Chen , Lai Peng Chan , Ronan Killeen , May San Mak , Jonathan P. McNulty , Kularatna Sanjeewa","doi":"10.1016/j.ejro.2024.100605","DOIUrl":"10.1016/j.ejro.2024.100605","url":null,"abstract":"<div><h3>Objectives</h3><div>There is a lack of clear and consistent cost reporting for cost-effectiveness analysis in radiology. Estimates are often obtained using costing derived from hospital charge records. This study aims to evaluate the accuracy of hospital charge records compared to a Singapore hospital's true diagnostic imaging costs.</div></div><div><h3>Methods</h3><div>A seven-step process involving a bottom-up micro-costing approach was devised and followed to calculate the cost of imaging using actual data from a clinical setting. We retrieved electronic data from a random sample of 96 emergency department patients who had CT brain, CT and X-ray cervical spine, and X-ray lumbar spine performed to calculate the parameters required for cost estimation. We adjusted imaging duration and number of performing personnel to account for variations.</div></div><div><h3>Results</h3><div>Our approach determined the average cost for the following imaging procedures: CT brain (€154.00), CT and X-ray cervical spine (€177.14 and €68.22), and X-ray lumbar spine (€79.85). We found that the true cost of both conventional radiography procedures was marginally higher than the subsidized patient charge, and all costs were slightly lower than the private patient charge except for X-ray lumbar spine (€73.49 vs.€79.85). We identified larger differences in cost for both CT procedures and smaller differences in cost for conventional radiography procedures, depending on the patient's private or subsidized status. For private status, the differences were: CT brain (Min: €194.20; Max: €264.40), CT cervical spine (Min: €219.54; Max: €399.05), X-ray cervical spine (Min: €5.27; Max: €61.94), and X-ray lumbar spine (Min: €6.36; Max: €108.04), while for subsidized status, the differences were: CT brain (Min: €7.56; Max: €62.64), CT cervical spine (Min: €47.02; Max: €132.49), X-ray cervical spine (Min: €15.88; Max: €103.44), and X-ray lumbar spine (Min: €13.66; Max: €149.44). Considering examination duration and the number of personnel engaged in a procedure, there were significant variations in the minimum, average, and maximum imaging costs.</div></div><div><h3>Conclusion</h3><div>There is a modest gap between hospital charges and actual costs, and we must therefore exercise caution and recognize the limitations of utilizing hospital charge records as absolute metrics for cost-effectiveness analysis<em>.</em> Our detailed approach can potentially enable more accurate imaging cost determination for future studies.</div></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":null,"pages":null},"PeriodicalIF":1.8,"publicationDate":"2024-10-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142535420","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-17DOI: 10.1016/j.ejro.2024.100603
Chuanjun Xu , Qinmei Xu , Li Liu , Mu Zhou , Zijian Xing , Zhen Zhou , Danyang Ren , Changsheng Zhou , Longjiang Zhang , Xiao Li , Xianghao Zhan , Olivier Gevaert , Guangming Lu
Purpose
The novel coronavirus pneumonia (COVID-19) has continually spread and mutated, requiring a patient risk stratification system to optimize medical resources and improve pandemic response. We aimed to develop a conformal prediction-based tri-light warning system for stratifying COVID-19 patients, applicable to both original and emerging variants.
Methods
We retrospectively collected data from 3646 patients across multiple centers in China. The dataset was divided into a training set (n = 1451), a validation set (n = 662), an external test set from Huoshenshan Field Hospital (n = 1263), and a specific test set for Delta and Omicron variants (n = 544). The tri-light warning system extracts radiomic features from CT (computed tomography) and integrates clinical records to classify patients into high-risk (red), uncertain-risk (yellow), and low-risk (green) categories. Models were built to predict ICU (intensive care unit) admissions (adverse cases in training/validation/Huoshenshan/variant test sets: n = 39/21/262/11) and were evaluated using AUROC ((area under the receiver operating characteristic curve)) and AUPRC ((area under the precision-recall curve)) metrics.
Results
The dataset included 1830 men (50.2 %) and 1816 women (50.8 %), with a median age of 53.7 years (IQR [interquartile range]: 42–65 years). The system demonstrated strong performance under data distribution shifts, with AUROC of 0.89 and AUPRC of 0.42 for original strains, and AUROC of 0.77–0.85 and AUPRC of 0.51–0.60 for variants.
Conclusion
The tri-light warning system can enhance pandemic responses by effectively stratifying COVID-19 patients under varying conditions and data shifts.
{"title":"A tri-light warning system for hospitalized COVID-19 patients: Credibility-based risk stratification for future pandemic preparedness","authors":"Chuanjun Xu , Qinmei Xu , Li Liu , Mu Zhou , Zijian Xing , Zhen Zhou , Danyang Ren , Changsheng Zhou , Longjiang Zhang , Xiao Li , Xianghao Zhan , Olivier Gevaert , Guangming Lu","doi":"10.1016/j.ejro.2024.100603","DOIUrl":"10.1016/j.ejro.2024.100603","url":null,"abstract":"<div><h3>Purpose</h3><div>The novel coronavirus pneumonia (COVID-19) has continually spread and mutated, requiring a patient risk stratification system to optimize medical resources and improve pandemic response. We aimed to develop a conformal prediction-based tri-light warning system for stratifying COVID-19 patients, applicable to both original and emerging variants.</div></div><div><h3>Methods</h3><div>We retrospectively collected data from 3646 patients across multiple centers in China. The dataset was divided into a training set (n = 1451), a validation set (n = 662), an external test set from Huoshenshan Field Hospital (n = 1263), and a specific test set for Delta and Omicron variants (n = 544). The tri-light warning system extracts radiomic features from CT (computed tomography) and integrates clinical records to classify patients into high-risk (red), uncertain-risk (yellow), and low-risk (green) categories. Models were built to predict ICU (intensive care unit) admissions (adverse cases in training/validation/Huoshenshan/variant test sets: n = 39/21/262/11) and were evaluated using AUROC ((area under the receiver operating characteristic curve)) and AUPRC ((area under the precision-recall curve)) metrics.</div></div><div><h3>Results</h3><div>The dataset included 1830 men (50.2 %) and 1816 women (50.8 %), with a median age of 53.7 years (IQR [interquartile range]: 42–65 years). The system demonstrated strong performance under data distribution shifts, with AUROC of 0.89 and AUPRC of 0.42 for original strains, and AUROC of 0.77–0.85 and AUPRC of 0.51–0.60 for variants.</div></div><div><h3>Conclusion</h3><div>The tri-light warning system can enhance pandemic responses by effectively stratifying COVID-19 patients under varying conditions and data shifts.</div></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":null,"pages":null},"PeriodicalIF":1.8,"publicationDate":"2024-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142446893","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-10DOI: 10.1016/j.ejro.2024.100604
Roberto Francischello , Salvatore Claudio Fanni , Martina Chiellini , Maria Febi , Giorgio Pomara , Claudio Bandini , Lorenzo Faggioni , Riccardo Lencioni , Emanuele Neri , Dania Cioni
Purpose
To investigate the potential role of radiomics-based machine learning in differentiating small renal oncocytoma (RO) from clear cells carcinoma (ccRCC) on contrast-enhanced CT (CECT).
Material and methods
Fifty-two patients with small renal masses who underwent CECT before surgery between January 2016 and December 2020 were retrospectively included in the study. At pathology examination 39 ccRCC and 13 RO were identified. All lesions were manually delineated unenhanced (B), arterial (A) and venous (V) phases. Radiomics features were extracted using three different fixed bin widths (bw) of 25 HU, 10 HU, and 5 HU from each phase (B, A, V), and with different combinations (B+A, B+V, B+A+V, A+V), leading to 21 different datasets. Montecarlo Cross Validation technique was used to quantify the estimator performance. The final model built using the hyperparameter selected with Optuna was trained again on the training set and the final performance evaluation was made on the test set.
Results
The A+V bw 10 achieved the greater median (IQR) balanced accuracy considering all the models of 0.70 (0.64–0.75), while A bw 10 considering only the monophasic ones. The A bw 10 model achieved a median (IQR) sensitivity of 0.60 (0.40–0.60), specificity of 0.80 (0.73–0.87), AUC-ROC of 0.77 (0.66–0.84), accuracy of 0.75 (0.70–0.80), and a Phi Coefficient of 0.38 (0.20–0.47). None of the nine models with the lowest mean balanced accuracy values implemented features from A.
Conclusion
The A bw 10 model was identified as the most efficient mono-phasic model in differentiating small RO from ccRCC.
{"title":"Radiomics-based machine learning role in differential diagnosis between small renal oncocytoma and clear cells carcinoma on contrast-enhanced CT: A pilot study","authors":"Roberto Francischello , Salvatore Claudio Fanni , Martina Chiellini , Maria Febi , Giorgio Pomara , Claudio Bandini , Lorenzo Faggioni , Riccardo Lencioni , Emanuele Neri , Dania Cioni","doi":"10.1016/j.ejro.2024.100604","DOIUrl":"10.1016/j.ejro.2024.100604","url":null,"abstract":"<div><h3>Purpose</h3><div>To investigate the potential role of radiomics-based machine learning in differentiating small renal oncocytoma (RO) from clear cells carcinoma (ccRCC) on contrast-enhanced CT (CECT).</div></div><div><h3>Material and methods</h3><div>Fifty-two patients with small renal masses who underwent CECT before surgery between January 2016 and December 2020 were retrospectively included in the study. At pathology examination 39 ccRCC and 13 RO were identified. All lesions were manually delineated unenhanced (B), arterial (A) and venous (V) phases. Radiomics features were extracted using three different fixed bin widths (bw) of 25 HU, 10 HU, and 5 HU from each phase (B, A, V), and with different combinations (B+A, B+V, B+A+V, A+V), leading to 21 different datasets. Montecarlo Cross Validation technique was used to quantify the estimator performance. The final model built using the hyperparameter selected with Optuna was trained again on the training set and the final performance evaluation was made on the test set.</div></div><div><h3>Results</h3><div>The A+V bw 10 achieved the greater median (IQR) balanced accuracy considering all the models of 0.70 (0.64–0.75), while A bw 10 considering only the monophasic ones. The A bw 10 model achieved a median (IQR) sensitivity of 0.60 (0.40–0.60), specificity of 0.80 (0.73–0.87), AUC-ROC of 0.77 (0.66–0.84), accuracy of 0.75 (0.70–0.80), and a Phi Coefficient of 0.38 (0.20–0.47). None of the nine models with the lowest mean balanced accuracy values implemented features from A.</div></div><div><h3>Conclusion</h3><div>The A bw 10 model was identified as the most efficient mono-phasic model in differentiating small RO from ccRCC.</div></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":null,"pages":null},"PeriodicalIF":1.8,"publicationDate":"2024-10-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142420067","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This study aimed to assess the diagnostic performance of dynamic chest radiography (DCR) and investigate its added value to chest radiography (CR) in detecting pulmonary embolism (PE).
Methods
Of 775 patients who underwent CR and DCR in our hospital between June 2020 and August 2022, individuals who also underwent contrast-enhanced CT (CECT) of the chest within 72 h were included in this study. PE or non-PE diagnosis was confirmed by CECT and the subsequent clinical course. The enrolled patients were randomized into two groups. Six observers, including two thoracic radiologists, two cardiologists, and two radiology residents, interpreted each chest radiograph with and without DCR using a crossover design with a washout period. Diagnostic performance was compared between CR with and without DCR in the standing and supine positions.
Results
Sixty patients (15 PE, 45 non-PE) were retrospectively enrolled. The addition of DCR to CR significantly improved the sensitivity, specificity, accuracy, and area under the curve (AUC) in the standing (35.6–70.0 % [P < 0.0001], 84.8–93.3 % [P = 0.0010], 72.5–87.5 % [P < 0.0001], and 0.66–0.85 [P < 0.0001], respectively) and supine (33.3–65.6 % [P < 0.0001], 78.5–92.2 % [P < 0.0001], 67.2–85.6 % [P < 0.0001], and 0.62–0.80 [P = 0.0002], respectively) positions for PE detection. No significant differences were found between the AUC values of DCR with CR in the standing and supine positions (P = 0.11) or among radiologists, cardiologists, and radiology residents (P = 0.14–0.68).
Conclusions
Incorporating DCR with CR demonstrated moderate sensitivity, high specificity, and high accuracy in detecting PE, all of which were significantly higher than those achieved with CR alone, regardless of scan position, observer expertise, or experience.
{"title":"Diagnostic accuracy and added value of dynamic chest radiography in detecting pulmonary embolism: A retrospective study","authors":"Yuzo Yamasaki , Kazuya Hosokawa , Takeshi Kamitani , Kohtaro Abe , Koji Sagiyama , Takuya Hino , Megumi Ikeda , Shunsuke Nishimura , Hiroyuki Toyoda , Shohei Moriyama , Masateru Kawakubo , Noritsugu Matsutani , Hidetake Yabuuchi , Kousei Ishigami","doi":"10.1016/j.ejro.2024.100602","DOIUrl":"10.1016/j.ejro.2024.100602","url":null,"abstract":"<div><h3>Purpose</h3><div>This study aimed to assess the diagnostic performance of dynamic chest radiography (DCR) and investigate its added value to chest radiography (CR) in detecting pulmonary embolism (PE).</div></div><div><h3>Methods</h3><div>Of 775 patients who underwent CR and DCR in our hospital between June 2020 and August 2022, individuals who also underwent contrast-enhanced CT (CECT) of the chest within 72 h were included in this study. PE or non-PE diagnosis was confirmed by CECT and the subsequent clinical course. The enrolled patients were randomized into two groups. Six observers, including two thoracic radiologists, two cardiologists, and two radiology residents, interpreted each chest radiograph with and without DCR using a crossover design with a washout period. Diagnostic performance was compared between CR with and without DCR in the standing and supine positions.</div></div><div><h3>Results</h3><div>Sixty patients (15 PE, 45 non-PE) were retrospectively enrolled. The addition of DCR to CR significantly improved the sensitivity, specificity, accuracy, and area under the curve (AUC) in the standing (35.6–70.0 % [<em>P</em> < 0.0001], 84.8–93.3 % [<em>P</em> = 0.0010], 72.5–87.5 % [<em>P</em> < 0.0001], and 0.66–0.85 [<em>P</em> < 0.0001], respectively) and supine (33.3–65.6 % [<em>P</em> < 0.0001], 78.5–92.2 % [<em>P</em> < 0.0001], 67.2–85.6 % [<em>P</em> < 0.0001], and 0.62–0.80 [<em>P</em> = 0.0002], respectively) positions for PE detection. No significant differences were found between the AUC values of DCR with CR in the standing and supine positions (P = 0.11) or among radiologists, cardiologists, and radiology residents (P = 0.14–0.68).</div></div><div><h3>Conclusions</h3><div>Incorporating DCR with CR demonstrated moderate sensitivity, high specificity, and high accuracy in detecting PE, all of which were significantly higher than those achieved with CR alone, regardless of scan position, observer expertise, or experience.</div></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":null,"pages":null},"PeriodicalIF":1.8,"publicationDate":"2024-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142420066","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-09-21DOI: 10.1016/j.ejro.2024.100601
Anass Benfares , Abdelali yahya Mourabiti , Badreddine Alami , Sara Boukansa , Nizar El Bouardi , Moulay Youssef Alaoui Lamrani , Hind El Fatimi , Bouchra Amara , Mounia Serraj , Smahi Mohammed , Cherkaoui Abdeljabbar , El affar Anass , Mamoun Qjidaa , Mustapha Maaroufi , Ouazzani Jamil Mohammed , Qjidaa Hassan
Purpose
To propose an intelligent, non-invasive, highly precise, and rapid method to predict the mutation status of the Epidermal Growth Factor Receptor (EGFR) to accelerate treatment with Tyrosine Kinase Inhibitor (TKI) for patients with untreated adenocarcinoma Non-Small Cell Lung Cancer.
Materials and methods
Real-world data from 521 patients with adenocarcinoma NSCLC who performed a CT scan and underwent surgery or pathological biopsy to determine EGFR gene mutation between January 2021 and July 2022, is collected. Solutions to the problems that prevent the model from achieving very high precision, namely: human errors made during the annotation of the database and the low precision of the output decision of the model, are proposed. Thus, among the 521 analyzed cases, only 40 were selected as patients with EGFR gene mutation and 98 cases with wild-type EGFR.
Results
The proposed model is trained, validated, and tested on 12,040 2D images extracted from the 138 CT scans images where patients were randomly partitioned into training (80 %) and test (20 %) sets. The performance obtained for EGFR gene mutation prediction was 95.22 % for accuracy, 960.2 for F1_score, 95.89 % for precision, 96.92 % for sensitivity, 94.01 % for Cohen kappa, and 98 % for AUC.
Conclusion
An EGFR gene mutation status prediction method, with high-performance thanks to an intelligent prediction model entrained by highly accurate annotated data is proposed. The outcome of this project will facilitate rapid decision-making when applying a TKI as an initial treatment.
{"title":"Non-invasive, fast, and high-performance EGFR gene mutation prediction method based on deep transfer learning and model stacking for patients with Non-Small Cell Lung Cancer","authors":"Anass Benfares , Abdelali yahya Mourabiti , Badreddine Alami , Sara Boukansa , Nizar El Bouardi , Moulay Youssef Alaoui Lamrani , Hind El Fatimi , Bouchra Amara , Mounia Serraj , Smahi Mohammed , Cherkaoui Abdeljabbar , El affar Anass , Mamoun Qjidaa , Mustapha Maaroufi , Ouazzani Jamil Mohammed , Qjidaa Hassan","doi":"10.1016/j.ejro.2024.100601","DOIUrl":"10.1016/j.ejro.2024.100601","url":null,"abstract":"<div><h3>Purpose</h3><p>To propose an intelligent, non-invasive, highly precise, and rapid method to predict the mutation status of the Epidermal Growth Factor Receptor (EGFR) to accelerate treatment with Tyrosine Kinase Inhibitor (TKI) for patients with untreated adenocarcinoma Non-Small Cell Lung Cancer.</p></div><div><h3>Materials and methods</h3><p>Real-world data from 521 patients with adenocarcinoma NSCLC who performed a CT scan and underwent surgery or pathological biopsy to determine EGFR gene mutation between January 2021 and July 2022, is collected. Solutions to the problems that prevent the model from achieving very high precision, namely: human errors made during the annotation of the database and the low precision of the output decision of the model, are proposed. Thus, among the 521 analyzed cases, only 40 were selected as patients with EGFR gene mutation and 98 cases with wild-type EGFR.</p></div><div><h3>Results</h3><p>The proposed model is trained, validated, and tested on 12,040 2D images extracted from the 138 CT scans images where patients were randomly partitioned into training (80 %) and test (20 %) sets. The performance obtained for EGFR gene mutation prediction was 95.22 % for accuracy, 960.2 for F1_score, 95.89 % for precision, 96.92 % for sensitivity, 94.01 % for Cohen kappa, and 98 % for AUC.</p></div><div><h3>Conclusion</h3><p>An EGFR gene mutation status prediction method, with high-performance thanks to an intelligent prediction model entrained by highly accurate annotated data is proposed. The outcome of this project will facilitate rapid decision-making when applying a TKI as an initial treatment.</p></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":null,"pages":null},"PeriodicalIF":1.8,"publicationDate":"2024-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S235204772400056X/pdfft?md5=6b569d6b0991ebec79c5235f88184fd5&pid=1-s2.0-S235204772400056X-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142271651","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-09-21DOI: 10.1016/j.ejro.2024.100600
Jinxin Chen, Xinyi Zeng, Feng Li, Jidong Peng
Objective
This study aims to analyze the quantitative parameters and morphological indices of three-dimensional (3D) visualization to differentiate lepidic predominant adenocarcinoma (LPA) from non-LPA subtypes, which include acinar predominant adenocarcinoma (APA), papillary predominant adenocarcinoma (PPA), micropapillary predominant adenocarcinoma (MPA), and solid predominant adenocarcinoma (SPA).
Methods
A group of 178 individuals diagnosed with lung adenocarcinoma were chosen and categorized into two groups: the LPA group and the non-LPA group, according to their pathological results. Quantitative parameters and morphological indexes such as 3D volume, solid proportion, and vascular cluster sign were obtained using 3D visualization and reconstruction techniques.
Results
Significant differences were observed in the vascular cluster sign, spiculation, shape, air bronchogram, bubble-like lucency, margin, pleural indentation, lobulation, maximum tumor diameter, 3D mean CT value, 3D volume, 3D mass, 3D density, and solid proportion between two groups (P<0.05). The optimal cut-off values for diagnosing non-LPA were a 3D mean CT value of −445.45 HU, a 3D density of 0.56 mg·mm−3, and a solid proportion reaching 27.95 %. Multivariate logistic regression analysis revealed that 3D mean CT value, lobulation, and margin characteristics independently predicted stageⅠinvasive lung adenocarcinoma. The combination of three indicators significantly improved prediction accuracy (AUC=0.881).
Conclusion
The utilization of 3D visualization technology in a systematic approach enables the acquisition of 3D quantitative parameters and morphological indicators of thin-slice CT lesions. These efforts significantly contribute to the identification of histopathological subtypes for stageⅠinvasive lung adenocarcinoma. When integrated with pertinent clinical data, this offers essential guidance for developing various surgical techniques and treatment plans.
{"title":"The value of non-enhanced CT 3D visualization in differentiating stage Ⅰ invasive lung adenocarcinoma between LPA and non-LPA","authors":"Jinxin Chen, Xinyi Zeng, Feng Li, Jidong Peng","doi":"10.1016/j.ejro.2024.100600","DOIUrl":"10.1016/j.ejro.2024.100600","url":null,"abstract":"<div><h3>Objective</h3><p>This study aims to analyze the quantitative parameters and morphological indices of three-dimensional (3D) visualization to differentiate lepidic predominant adenocarcinoma (LPA) from non-LPA subtypes, which include acinar predominant adenocarcinoma (APA), papillary predominant adenocarcinoma (PPA), micropapillary predominant adenocarcinoma (MPA), and solid predominant adenocarcinoma (SPA).</p></div><div><h3>Methods</h3><p>A group of 178 individuals diagnosed with lung adenocarcinoma were chosen and categorized into two groups: the LPA group and the non-LPA group, according to their pathological results. Quantitative parameters and morphological indexes such as 3D volume, solid proportion, and vascular cluster sign were obtained using 3D visualization and reconstruction techniques.</p></div><div><h3>Results</h3><p>Significant differences were observed in the vascular cluster sign, spiculation, shape, air bronchogram, bubble-like lucency, margin, pleural indentation, lobulation, maximum tumor diameter, 3D mean CT value, 3D volume, 3D mass, 3D density, and solid proportion between two groups (P<0.05). The optimal cut-off values for diagnosing non-LPA were a 3D mean CT value of −445.45 HU, a 3D density of 0.56 mg·mm<sup>−3</sup>, and a solid proportion reaching 27.95 %. Multivariate logistic regression analysis revealed that 3D mean CT value, lobulation, and margin characteristics independently predicted stageⅠinvasive lung adenocarcinoma. The combination of three indicators significantly improved prediction accuracy (AUC=0.881).</p></div><div><h3>Conclusion</h3><p>The utilization of 3D visualization technology in a systematic approach enables the acquisition of 3D quantitative parameters and morphological indicators of thin-slice CT lesions. These efforts significantly contribute to the identification of histopathological subtypes for stageⅠinvasive lung adenocarcinoma. When integrated with pertinent clinical data, this offers essential guidance for developing various surgical techniques and treatment plans.</p></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":null,"pages":null},"PeriodicalIF":1.8,"publicationDate":"2024-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2352047724000558/pdfft?md5=6ba180c2567cfc76febca8a162b97b4b&pid=1-s2.0-S2352047724000558-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142271652","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号