European Journal of Radiology Open最新文献_第4页

Magnetic resonance imaging radiomics-based machine learning model for preoperative prediction of pathological grade in pancreatic cancer: A preliminary study 基于磁共振成像放射组学的机器学习模型用于胰腺癌病理分级的术前预测：初步研究

IF 2.9 Q3 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING

European Journal of Radiology Open

Pub Date : 2025-11-07 DOI: 10.1016/j.ejro.2025.100703

Jiayi Bao , Yuanqing Liu , Yang Yu , Ximing Wang , Chunhong Hu , Su Hu

Background

To explore the value of MRI radiomics-based machine learning models for predicting the pathological grade of pancreatic cancer preoperatively.

Methods

125 patients with pathologically confirmed pancreatic cancer who underwent preoperative MRI were retrospectively enrolled. The primary cohort was randomized in an 8:2 ratio into a training cohort (n = 100) and a validation cohort (n = 25). 1316 radiomics features were extracted from contrast-enhanced T₁WI arterial phase (AP) or portal venous phase (PVP) images, respectively. After feature reduction and filtering, the best features were selected to construct machine learning models (K-nearest neighbor, KNN; support vector machine, SVM; logistic regression, LR; random forest, RF). Finally, the performance of these models was evaluated using the receiver operating characteristic curve (ROC).

Results

There were no statistical differences in clinical characteristics between the low-grade and high-grade cohorts (P > 0.05). The best radiomics features selected from the AP, PVP and AP+PVP images were 6, 6 and 10, respectively. Among the four models, the LR machine learning model achieved the best predictive performance. The distribution of the Radscore values was clinically significant between the low-grade and high-grade groups both in the training cohort (median, 0.26 vs 0.99; P < 0.001) and validation cohort (median, 0.63 vs 1.48; P = 0.011). LR model of AP+PVP performed the best with AUC value of 0.81 (95 % CI: 0.72–0.91) for the training cohort and 0.82 (95 % CI: 0.62–1.00) for the validation cohort.

Conclusions

MRI radiomics-based machine learning model is a potential non-invasive method to predict the pathological grade of pancreatic cancer.

目的探讨基于MRI放射学的机器学习模型在胰腺癌术前病理分级预测中的应用价值。方法回顾性分析术前行MRI检查的经病理证实的胰腺癌患者125例。主要队列按8:2的比例随机分为训练队列（n = 100）和验证队列（n = 25）。分别从增强T1WI动脉期（AP）或门静脉期（PVP）图像中提取1316个放射组学特征。经过特征约简和滤波，选择最优特征构建机器学习模型（K-nearest neighbor， KNN; support vector machine， SVM; logistic regression， LR; random forest， RF）。最后，使用受试者工作特征曲线（ROC）评估这些模型的性能。结果低分级组与高分级组临床特征比较，差异无统计学意义（P >； 0.05）。AP、PVP和AP+PVP图像的放射组学特征值分别为6、6和10。在四种模型中，LR机器学习模型的预测性能最好。在训练队列（中位数，0.26 vs 0.99; P <； 0.001）和验证队列（中位数，0.63 vs 1.48; P = 0.011）中，低分级组和高分级组之间的Radscore值分布具有临床意义。AP+PVP的LR模型表现最好，训练组的AUC值为0.81(95 % CI: 0.72-0.91)，验证组的AUC值为0.82（95 % CI: 0.62-1.00）。结论基于smri放射组学的机器学习模型是一种潜在的无创预测胰腺癌病理分级的方法。

{"title":"Magnetic resonance imaging radiomics-based machine learning model for preoperative prediction of pathological grade in pancreatic cancer: A preliminary study","authors":"Jiayi Bao , Yuanqing Liu , Yang Yu , Ximing Wang , Chunhong Hu , Su Hu","doi":"10.1016/j.ejro.2025.100703","DOIUrl":"10.1016/j.ejro.2025.100703","url":null,"abstract":"<div><h3>Background</h3><div>To explore the value of MRI radiomics-based machine learning models for predicting the pathological grade of pancreatic cancer preoperatively.</div></div><div><h3>Methods</h3><div>125 patients with pathologically confirmed pancreatic cancer who underwent preoperative MRI were retrospectively enrolled. The primary cohort was randomized in an 8:2 ratio into a training cohort (n = 100) and a validation cohort (n = 25). 1316 radiomics features were extracted from contrast-enhanced T<sub>1</sub>WI arterial phase (AP) or portal venous phase (PVP) images, respectively. After feature reduction and filtering, the best features were selected to construct machine learning models (K-nearest neighbor, KNN; support vector machine, SVM; logistic regression, LR; random forest, RF). Finally, the performance of these models was evaluated using the receiver operating characteristic curve (ROC).</div></div><div><h3>Results</h3><div>There were no statistical differences in clinical characteristics between the low-grade and high-grade cohorts (<em>P</em> > 0.05). The best radiomics features selected from the AP, PVP and AP+PVP images were 6, 6 and 10, respectively. Among the four models, the LR machine learning model achieved the best predictive performance. The distribution of the Radscore values was clinically significant between the low-grade and high-grade groups both in the training cohort (median, 0.26 vs 0.99; <em>P</em> < 0.001) and validation cohort (median, 0.63 vs 1.48; <em>P</em> = 0.011). LR model of AP+PVP performed the best with AUC value of 0.81 (95 % CI: 0.72–0.91) for the training cohort and 0.82 (95 % CI: 0.62–1.00) for the validation cohort.</div></div><div><h3>Conclusions</h3><div>MRI radiomics-based machine learning model is a potential non-invasive method to predict the pathological grade of pancreatic cancer.</div></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":"15 ","pages":"Article 100703"},"PeriodicalIF":2.9,"publicationDate":"2025-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145464879","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Cardiac MRI-derived ventricular-pulmonary arterial coupling predicts outcomes in connective tissue disease-associated pulmonary arterial hypertension patients 心脏mri衍生的心室-肺动脉耦合预测结缔组织病相关肺动脉高压患者的预后

IF 2.9 Q3 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING

European Journal of Radiology Open

Pub Date : 2025-11-05 DOI: 10.1016/j.ejro.2025.100700

Yidan Zhu , Ke Xu , Han Jia , Wangyan Liu , Xiaoxuan Sun , Qiang Wang , Yi Xu , Yinsu Zhu

Objectives

To evaluate the prognostic value of cardiovascular magnetic resonance (CMR) parameters in patients with connective tissue disease-associated pulmonary arterial hypertension (CTD-PAH).

Materials and methods

This retrospective cohort study involved 135 patients with documented CTD-PAH. Cardiac functional parameters including right ventricle end-diastolic volume index (RVEDVI), end-systolic volume index (RVESVI), stroke volume index (RVSVI), ejection fraction (RVEF), and SV/ESV; the volumetric parameter; T1 mapping and late gadolinium enhancement (LGE) parameters; strain parameters like RV global longitudinal strain(GLS); and hemodynamics parameters like mean velocity at the pulmonary artery (mvPA) and PA relative area change (PA RAC) were calculated from CMR images. Survival analysis was conducted using Kaplan–Meier and Cox regression. The endpoint was the occurrence of clinical deterioration events.

Results

The median follow-up time was 22.27 months (interquartile range: 11.5–31.9 months).RVIP ECV, RVGLS, RVEDVI, RVESVI, RVMI, and RAA, as well as mvPA, SV/ESV, RVEF, and PA RAC, were associated with adverse outcomes. SV/ESV ≤ 0.55, RV GLS > -12.1 %, and PA RAC≤ 16.1 % were significant independent predictors of prognosis. The combined parameters provided incremental prognostic value over COMPERA 2.0 (area under the curve (AUC) from 0.771 to 0.899; P = 0.001).And time-dependent ROC curve confirmed the predictive efficiency of the combined parameters in year 3, with the AUC reaching 0.821(95 % CI: 0.698–0.945).

Conclusion

Multiparametric CMR provides a non-invasive, efficient method to assess prognosis in CTD-PAH patients. Key parameters including SV/ESV, RV GLS, and PA RAC significantly predicted survival and confer incremental prognostic utility over COMPERA 2.0, offering reliable prognostic markers for clinical interventions.

目的评价心血管磁共振（CMR）参数对结缔组织病相关性肺动脉高压（CTD-PAH）患者的预后价值。材料和方法本回顾性队列研究纳入135例有记录的CTD-PAH患者。心功能参数包括右心室舒张末期容积指数（RVEDVI）、收缩末期容积指数（RVESVI）、卒中容积指数（RVSVI）、射血分数（RVEF）和SV/ESV；体积参数；T1成像和晚期钆增强（LGE）参数；RV整体纵向应变（GLS）等应变参数；根据CMR图像计算肺动脉平均流速（mvPA）、肺动脉相对面积变化（PA RAC）等血流动力学参数。生存率分析采用Kaplan-Meier和Cox回归。终点是临床恶化事件的发生。结果中位随访时间为22.27个月（四分位数差为11.5 ~ 31.9个月）。RVIP、ECV、RVGLS、RVEDVI、RVESVI、RVMI和RAA，以及mvPA、SV/ESV、RVEF和PA RAC与不良结局相关。SV/ESV≤ 0.55,RV GLS >； -12.1 %,PA RAC≤ 16.1 %是预后的显著独立预测因子。综合参数在COMPERA 2.0上提供了增量预后价值(曲线下面积（AUC）从0.771到0.899； = 0.001页)。随时间变化的ROC曲线证实了联合参数在第3年的预测效率，AUC达到0.821（95 % CI: 0.698-0.945）。结论多参数CMR是一种无创、有效的评估CTD-PAH患者预后的方法。关键参数包括SV/ESV、RV GLS和PA RAC显著预测生存，并赋予比COMPERA 2.0更大的预后效用，为临床干预提供可靠的预后指标。

{"title":"Cardiac MRI-derived ventricular-pulmonary arterial coupling predicts outcomes in connective tissue disease-associated pulmonary arterial hypertension patients","authors":"Yidan Zhu , Ke Xu , Han Jia , Wangyan Liu , Xiaoxuan Sun , Qiang Wang , Yi Xu , Yinsu Zhu","doi":"10.1016/j.ejro.2025.100700","DOIUrl":"10.1016/j.ejro.2025.100700","url":null,"abstract":"<div><h3>Objectives</h3><div>To evaluate the prognostic value of cardiovascular magnetic resonance (CMR) parameters in patients with connective tissue disease-associated pulmonary arterial hypertension (CTD-PAH).</div></div><div><h3>Materials and methods</h3><div>This retrospective cohort study involved 135 patients with documented CTD-PAH. Cardiac functional parameters including right ventricle end-diastolic volume index (RVEDVI), end-systolic volume index (RVESVI), stroke volume index (RVSVI), ejection fraction (RVEF), and SV/ESV; the volumetric parameter; T1 mapping and late gadolinium enhancement (LGE) parameters; strain parameters like RV global longitudinal strain(GLS); and hemodynamics parameters like mean velocity at the pulmonary artery (mvPA) and PA relative area change (PA RAC) were calculated from CMR images. Survival analysis was conducted using Kaplan–Meier and Cox regression. The endpoint was the occurrence of clinical deterioration events.</div></div><div><h3>Results</h3><div>The median follow-up time was 22.27 months (interquartile range: 11.5–31.9 months).RVIP ECV, RVGLS, RVEDVI, RVESVI, RVMI, and RAA, as well as mvPA, SV/ESV, RVEF, and PA RAC, were associated with adverse outcomes. SV/ESV ≤ 0.55, RV GLS > -12.1 %, and PA RAC≤ 16.1 % were significant independent predictors of prognosis. The combined parameters provided incremental prognostic value over COMPERA 2.0 (area under the curve (AUC) from 0.771 to 0.899; P = 0.001).And time-dependent ROC curve confirmed the predictive efficiency of the combined parameters in year 3, with the AUC reaching 0.821(95 % CI: 0.698–0.945).</div></div><div><h3>Conclusion</h3><div>Multiparametric CMR provides a non-invasive, efficient method to assess prognosis in CTD-PAH patients. Key parameters including SV/ESV, RV GLS, and PA RAC significantly predicted survival and confer incremental prognostic utility over COMPERA 2.0, offering reliable prognostic markers for clinical interventions.</div></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":"15 ","pages":"Article 100700"},"PeriodicalIF":2.9,"publicationDate":"2025-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145464877","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Diagnostic accuracy of AI in chest radiography for pneumonia and lung cancer: A meta-analysis 人工智能在肺炎和肺癌胸片诊断中的准确性：一项荟萃分析

IF 2.9 Q3 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING

European Journal of Radiology Open

Pub Date : 2025-11-03 DOI: 10.1016/j.ejro.2025.100701

Mohammed Elmujtba Adam Essa

Rationale and Objectives

Chest radiography (CXR) is the most common imaging test worldwide for evaluating pulmonary disease, yet its sensitivity for pneumonia and lung cancer is limited. Artificial intelligence (AI) based image analysis has shown promise to aid radiographic diagnosis. A meta-analysis was performed for AI algorithms evaluation for detecting pneumonia or lung nodules on CXR, and AI performance was compared to human readers.

Materials and Methods

Following PRISMA guidelines, searched PubMed/Medline, Embase, Cochrane, and IEEE Xplore (Jan 2017–July 2025) for diagnostic accuracy studies of AI on CXR. Eligible studies included any prospective or retrospective design reporting sensitivity and specificity for AI-based pneumonia or lung nodule detection, with an independent reference standard. Data were extracted using a standardized form and quality was assessed with QUADAS2.

Results

Fifteen studies (≈12,000 CXRs) met inclusion criteria; individual AI algorithms achieved sensitivities of ∼70–97 % and specificities of ∼85–95 %. Meta-analysis yielded a pooled sensitivity of 88 % and specificity of 90 % for AI pneumonia detection. For lung nodules, pooled AI sensitivity was ≈ 72 % and specificity ≈ 95 %. One representative deep-learning model for detecting nodules. AI tended to miss very small or central nodules but detected ∼90 % of larger nodules. Crucially, using AI as a second reader improved radiologist performance, increasing sensitivity by approximately 9–10 %age points.

Conclusion

AI algorithms demonstrate high diagnostic accuracy for pneumonia on CXR and can markedly increase the detection of occult lung nodules when used as a second reader. However, performance varies by lesion characteristics. Overall, AI has strong potential to enhance clinical chest radiograph interpretation.

理由和目的胸部x线摄影（CXR）是世界范围内评估肺部疾病最常用的影像学检查，但其对肺炎和肺癌的敏感性有限。基于人工智能（AI）的图像分析显示出帮助放射诊断的前景。对在CXR上检测肺炎或肺结节的人工智能算法评价进行了荟萃分析，并将人工智能的表现与人类读者进行了比较。材料和方法：按照PRISMA指南，检索PubMed/Medline、Embase、Cochrane和IEEE Xplore（2017年1月- 2025年7月），获取人工智能在CXR上的诊断准确性研究。符合条件的研究包括任何前瞻性或回顾性设计，报告基于人工智能的肺炎或肺结节检测的敏感性和特异性，并有独立的参考标准。使用标准化表格提取数据，并使用QUADAS2评估质量。结果15项研究（约12,000例cxr）符合纳入标准；单个AI算法的灵敏度为~ 70-97 %，特异性为~ 85-95 %。荟萃分析得出AI肺炎检测的总敏感性为88 %，特异性为90 %。对于肺结节，人工智能综合敏感性≈ 72 %，特异性≈ 95 %。一种典型的结节检测深度学习模型。AI往往会遗漏非常小或中心的结节，但能检测到~ 90% %的较大结节。至关重要的是，使用人工智能作为第二阅读器提高了放射科医生的表现，将灵敏度提高了大约9 - 10%。结论人工智能算法对CXR肺炎具有较高的诊断准确性，作为第二阅读器可显著提高肺隐性结节的检出率。然而，表现因病变特征而异。总的来说，人工智能在增强临床胸片解释方面具有很强的潜力。

{"title":"Diagnostic accuracy of AI in chest radiography for pneumonia and lung cancer: A meta-analysis","authors":"Mohammed Elmujtba Adam Essa","doi":"10.1016/j.ejro.2025.100701","DOIUrl":"10.1016/j.ejro.2025.100701","url":null,"abstract":"<div><h3>Rationale and Objectives</h3><div>Chest radiography (CXR) is the most common imaging test worldwide for evaluating pulmonary disease, yet its sensitivity for pneumonia and lung cancer is limited. Artificial intelligence (AI) based image analysis has shown promise to aid radiographic diagnosis. A meta-analysis was performed for AI algorithms evaluation for detecting pneumonia or lung nodules on CXR, and AI performance was compared to human readers.</div></div><div><h3>Materials and Methods</h3><div>Following PRISMA guidelines, searched PubMed/Medline, Embase, Cochrane, and IEEE Xplore (Jan 2017–July 2025) for diagnostic accuracy studies of AI on CXR. Eligible studies included any prospective or retrospective design reporting sensitivity and specificity for AI-based pneumonia or lung nodule detection, with an independent reference standard. Data were extracted using a standardized form and quality was assessed with QUADAS2.</div></div><div><h3>Results</h3><div>Fifteen studies (≈12,000 CXRs) met inclusion criteria; individual AI algorithms achieved sensitivities of ∼70–97 % and specificities of ∼85–95 %. Meta-analysis yielded a pooled sensitivity of 88 % and specificity of 90 % for AI pneumonia detection. For lung nodules, pooled AI sensitivity was ≈ 72 % and specificity ≈ 95 %. One representative deep-learning model for detecting nodules. AI tended to miss very small or central nodules but detected ∼90 % of larger nodules. Crucially, using AI as a second reader improved radiologist performance, increasing sensitivity by approximately 9–10 %age points.</div></div><div><h3>Conclusion</h3><div>AI algorithms demonstrate high diagnostic accuracy for pneumonia on CXR and can markedly increase the detection of occult lung nodules when used as a second reader. However, performance varies by lesion characteristics. Overall, AI has strong potential to enhance clinical chest radiograph interpretation.</div></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":"15 ","pages":"Article 100701"},"PeriodicalIF":2.9,"publicationDate":"2025-11-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145464878","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Dual-metric Bayesian optimization of B-spline mesh size for 4DCT lung registration 4DCT肺配准b样条网格尺寸双度量贝叶斯优化

IF 2.9 Q3 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING

European Journal of Radiology Open

Pub Date : 2025-11-01 DOI: 10.1016/j.ejro.2025.100698

Liang Tan, Liyuan Chen, Huanli Luo, Xin Yang, Bin Feng, Fu Jin

Objectives

We aim to optimize the patient-specific mesh size (N) in the B-spline deformable image registration method, enhancing the computational efficiency of 4DCT lung image registration.

Methods

This study included 37 subjects (10 from the DIRLAB public dataset and 27 from a private 4DCT cohort), each consisting of 10 respiratory phases. A Bayesian optimization (BO) framework was proposed to determine patient-specific N within [2, 50]. Registration accuracy was evaluated using the Dice Similarity Coefficient (DSC) and Hausdorff Distance (HD). To further validate registration robustness, extreme-phase registrations were additionally tested, and inverse consistency error (ICE) was calculated to assess deformation invertibility. A global evaluation approach was also applied across the full respiratory cycle, and the computational cost of the traditional grid search (GS) was analyzed for comparison.

Results

BO efficiently determined patient-specific N, with optimal values ranging from 6 to 15 (overall mean = 10.4 ± 2.6), achieving DSC = 0.976 and HD = 0.814. In the extreme-phase tests, registration performance remained stable between forward and reverse directions, with DSC > 0.94, HD₉₅< 3 mm, and small ICE differences (ICE₉₅ = 0.467 ± 0.230 mm), indicating strong inverse consistency and deformation stability. Compared with GS, BO achieved 50.7 %–99.4 % time savings, while GS showed a power-law increase in runtime (exponent = 2.53).

Conclusions

The proposed BO framework efficiently optimized patient-specific mesh sizes, achieving high registration accuracy and significantly reduced computation time, thereby offering a promising tool to improve efficiency in adaptive radiotherapy and motion-compensated treatment planning.

目的优化b样条可变形图像配准方法中患者特异性网格尺寸(N)，提高4DCT肺部图像配准的计算效率。方法本研究包括37名受试者（10名来自DIRLAB公共数据集，27名来自私人4DCT队列），每个受试者由10个呼吸期组成。提出了一个贝叶斯优化（BO）框架来确定患者特异性N[2,50]。采用Dice Similarity Coefficient （DSC）和Hausdorff Distance （HD）评价配准精度。为了进一步验证配准的稳健性，我们对极端相位配准进行了额外的测试，并计算了逆一致性误差（ICE）来评估变形可逆性。采用全呼吸周期的全局评价方法，对比分析了传统网格搜索方法的计算成本。结果bo能有效测定患者特异性N，最优值为6 ~ 15（总平均值= 10.4 ± 2.6），DSC = 0.976,HD = 0.814。在极相试验中，正反方向配准性能保持稳定，DSC >； 0.94,HD95<； 3 mm， ICE差异较小（ICE95 = 0.467 ± 0.230 mm），具有较强的逆一致性和变形稳定性。与GS相比，BO节省了50.7 % -99.4 %的时间，而GS在运行时间上呈幂律增长（指数= 2.53）。结论提出的BO框架有效地优化了患者特异性网格尺寸，实现了高配准精度，显著减少了计算时间，从而为提高自适应放疗和运动补偿治疗计划的效率提供了一个有希望的工具。

{"title":"Dual-metric Bayesian optimization of B-spline mesh size for 4DCT lung registration","authors":"Liang Tan, Liyuan Chen, Huanli Luo, Xin Yang, Bin Feng, Fu Jin","doi":"10.1016/j.ejro.2025.100698","DOIUrl":"10.1016/j.ejro.2025.100698","url":null,"abstract":"<div><h3>Objectives</h3><div>We aim to optimize the patient-specific mesh size (N) in the B-spline deformable image registration method, enhancing the computational efficiency of 4DCT lung image registration.</div></div><div><h3>Methods</h3><div>This study included 37 subjects (10 from the DIRLAB public dataset and 27 from a private 4DCT cohort), each consisting of 10 respiratory phases. A Bayesian optimization (BO) framework was proposed to determine patient-specific N within [2, 50]. Registration accuracy was evaluated using the Dice Similarity Coefficient (DSC) and Hausdorff Distance (HD). To further validate registration robustness, extreme-phase registrations were additionally tested, and inverse consistency error (ICE) was calculated to assess deformation invertibility. A global evaluation approach was also applied across the full respiratory cycle, and the computational cost of the traditional grid search (GS) was analyzed for comparison.</div></div><div><h3>Results</h3><div>BO efficiently determined patient-specific N, with optimal values ranging from 6 to 15 (overall mean = 10.4 ± 2.6), achieving DSC = 0.976 and HD = 0.814. In the extreme-phase tests, registration performance remained stable between forward and reverse directions, with DSC > 0.94, HD<sub>95</sub>< 3 mm, and small ICE differences (ICE<sub>95</sub> = 0.467 ± 0.230 mm), indicating strong inverse consistency and deformation stability. Compared with GS, BO achieved 50.7 %–99.4 % time savings, while GS showed a power-law increase in runtime (exponent = 2.53).</div></div><div><h3>Conclusions</h3><div>The proposed BO framework efficiently optimized patient-specific mesh sizes, achieving high registration accuracy and significantly reduced computation time, thereby offering a promising tool to improve efficiency in adaptive radiotherapy and motion-compensated treatment planning.</div></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":"15 ","pages":"Article 100698"},"PeriodicalIF":2.9,"publicationDate":"2025-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145416105","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Improvement of machine learning models for predicting high-grade subtypes of lung adenocarcinoma based on delta radiomics: A multicenter cohort study 基于放射组学的预测肺腺癌高级别亚型的机器学习模型的改进：一项多中心队列研究

IF 2.9 Q3 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING

European Journal of Radiology Open

Pub Date : 2025-11-01 DOI: 10.1016/j.ejro.2025.100699

Feiyang Zhong , Ting Li , Wenping Li , Lijun Wu , Pengju Zhang , Pengxin Yu , Yuan Fang , Meiyan Liao , Shaohong Zhao

Objectives

To evaluate the effectiveness of delta radiomics in predicting high-grade components in lung adenocarcinoma and to develop a robust machine learning model for clinical application.

Methods

This retrospective multi-center cohort study included lung cancer patients from three hospitals who had pre-surgery CT follow-up scans. Training (n = 491) and validation (n = 210) were performed using cases from Center 1, and testing was conducted using cases from Centers 2 and 3 (n = 92). Radiomic features were extracted from baseline and follow-up CT images, and delta radiomic features were calculated. The LASSO algorithm was used for radiomic feature selection, and rad-score and delta rad-score were constructed. Significant clinical and radiomic features were combined to build the final machine learning model. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), DeLong test, decision curve analysis (DCA), and integrated discrimination improvement (IDI) analysis.

Results

In the external test cohort, the integrated machine learning model constructed based on clinical features (CTR, smoking status, maximum diameter of the solid component), rad-score, and delta rad-score showed that the random forest model performed the best, with an AUC of 0.91. The random forest model outperformed the clinical model (AUC = 0.80), rad-score (AUC = 0.79), and delta rad-score (AUC = 0.81). DCA and IDI indicated that the random forest model provides superior clinical benefit and improvement.

Conclusion

Delta radiomics significantly aids in identifying high-grade subtypes of lung adenocarcinoma. The integrated machine learning model offers an effective approach for prediction of high-grade components, with potential clinical implications.

Clinical Relevance Statement

This study presents a novel application of delta radiomics to predict high-grade lung adenocarcinoma, which may influence surgical management and improve patient outcomes.

目的评估δ放射组学在预测肺腺癌高级别成分中的有效性，并为临床应用开发一个强大的机器学习模型。方法回顾性多中心队列研究纳入三家医院的肺癌患者术前CT随访扫描。使用中心1的病例进行训练（n = 491）和验证（n = 210），使用中心2和3的病例进行测试（n = 92）。从基线和随访CT图像中提取放射学特征，并计算δ放射学特征。采用LASSO算法进行放射学特征选择，并构造rad-score和delta rad-score。将重要的临床和放射学特征结合起来构建最终的机器学习模型。采用受试者工作特征曲线下面积（AUC）、DeLong检验、决策曲线分析（DCA）和综合判别改进（IDI）分析对模型性能进行评价。结果在外部测试队列中，基于临床特征（CTR、吸烟状况、实体成分最大直径）、rad-score和delta rad-score构建的综合机器学习模型显示随机森林模型表现最好，AUC为0.91。随机森林模型优于临床模型（AUC = 0.80）、rad-score （AUC = 0.79）和delta rad-score （AUC = 0.81）。DCA和IDI表明随机森林模型具有较好的临床疗效和改善效果。结论delta放射组学在鉴别肺腺癌高级别亚型中具有重要的辅助作用。集成的机器学习模型为预测高级成分提供了有效的方法，具有潜在的临床意义。临床相关性声明本研究提出了delta放射组学预测高级别肺腺癌的新应用，这可能会影响手术治疗并改善患者预后。

{"title":"Improvement of machine learning models for predicting high-grade subtypes of lung adenocarcinoma based on delta radiomics: A multicenter cohort study","authors":"Feiyang Zhong , Ting Li , Wenping Li , Lijun Wu , Pengju Zhang , Pengxin Yu , Yuan Fang , Meiyan Liao , Shaohong Zhao","doi":"10.1016/j.ejro.2025.100699","DOIUrl":"10.1016/j.ejro.2025.100699","url":null,"abstract":"<div><h3>Objectives</h3><div>To evaluate the effectiveness of delta radiomics in predicting high-grade components in lung adenocarcinoma and to develop a robust machine learning model for clinical application.</div></div><div><h3>Methods</h3><div>This retrospective multi-center cohort study included lung cancer patients from three hospitals who had pre-surgery CT follow-up scans. Training (n = 491) and validation (n = 210) were performed using cases from Center 1, and testing was conducted using cases from Centers 2 and 3 (n = 92). Radiomic features were extracted from baseline and follow-up CT images, and delta radiomic features were calculated. The LASSO algorithm was used for radiomic feature selection, and rad-score and delta rad-score were constructed. Significant clinical and radiomic features were combined to build the final machine learning model. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), DeLong test, decision curve analysis (DCA), and integrated discrimination improvement (IDI) analysis.</div></div><div><h3>Results</h3><div>In the external test cohort, the integrated machine learning model constructed based on clinical features (CTR, smoking status, maximum diameter of the solid component), rad-score, and delta rad-score showed that the random forest model performed the best, with an AUC of 0.91. The random forest model outperformed the clinical model (AUC = 0.80), rad-score (AUC = 0.79), and delta rad-score (AUC = 0.81). DCA and IDI indicated that the random forest model provides superior clinical benefit and improvement.</div></div><div><h3>Conclusion</h3><div>Delta radiomics significantly aids in identifying high-grade subtypes of lung adenocarcinoma. The integrated machine learning model offers an effective approach for prediction of high-grade components, with potential clinical implications.</div></div><div><h3>Clinical Relevance Statement</h3><div>This study presents a novel application of delta radiomics to predict high-grade lung adenocarcinoma, which may influence surgical management and improve patient outcomes.</div></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":"15 ","pages":"Article 100699"},"PeriodicalIF":2.9,"publicationDate":"2025-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145416106","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Development of a hybrid 2.5D deep learning model for glioma survival prediction using T1-weighted MRI from the CGGA database 开发混合2.5D深度学习模型，利用CGGA数据库的t1加权MRI预测胶质瘤生存

IF 2.9 Q3 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING

European Journal of Radiology Open

Pub Date : 2025-10-26 DOI: 10.1016/j.ejro.2025.100697

Kai Jin, Caixing Sun, Liang Xia

Background

Current glioma survival prediction relies on invasive molecular profiling. To overcome this, a non-invasive deep learning framework using T1-weighted contrast-enhanced MRI (T1CE) was developed to predict overall survival. This framework addresses computational limitations associated with the volumetric analysis while preserving important spatial information.

Methods

We designed a hybrid 2.5D convolutional neural network to process multi-slice inputs, including the center slice and its adjacent slices, from 217 patients in the CGGA database. Transfer learning using ResNet and DenseNet architectures were employed to initialize the models. These models were subsequently fine-tuned with the Cox proportional hazards loss function. After the fine-tuning process was completed, the imaging signature was combined with clinical and molecular variables, including IDH and 1p19q status, to build an integrated model. Performance was evaluated via C-index, time-dependent AUC, and Kaplan-Meier analysis in independent training (70 %) and testing (30 %) cohorts.

Results

The Combined model achieved superior discrimination, with a training C-index of 0.819 (95 % CI: 0.758–0.880) and a testing C-index of 0.804 (95 % CI: 0.708–0.900). It significantly outperformed the isolated Radiomic, deep learning (2D and 2.5D), and Clinical models (all p < 0.05). Moreover, time-dependent ROC analysis demonstrated consistent model performance over 1–5 years, with AUC values ranging from 0.851 to 0.906. The stratified survival curves clearly revealed distinct prognostic groups (log-rank p < 0.001).

Conclusions

The 2.5D multi-source framework provides a clinically feasible, non-invasive tool for preoperative survival prediction, enabling personalized therapeutic strategies for glioma patients.

目前的胶质瘤生存预测依赖于侵入性分子谱分析。为了克服这一问题，研究人员开发了一种使用t1加权对比增强MRI （T1CE）的非侵入性深度学习框架来预测总生存率。该框架解决了与体积分析相关的计算限制，同时保留了重要的空间信息。方法设计一种混合2.5D卷积神经网络，对217例CGGA患者的中心切片及其相邻切片进行多片输入处理。使用ResNet和DenseNet架构进行迁移学习来初始化模型。这些模型随后用Cox比例风险损失函数进行微调。在微调过程完成后，结合临床和分子变量，包括IDH和1p19q状态，构建集成模型。在独立训练（70 %）和测试（30 %）队列中，通过c指数、时间相关AUC和Kaplan-Meier分析来评估表现。结果联合模型的训练c -指数为0.819(95 % CI: 0.758 ~ 0.880)，检验c -指数为0.804(95 % CI: 0.708 ~ 0.900)，具有较好的判别性。它明显优于孤立的Radiomic、深度学习（2D和2.5D）和临床模型（均p <； 0.05）。此外，时间相关的ROC分析显示，模型在1-5年内的表现一致，AUC值在0.851 ~ 0.906之间。分层生存曲线清楚地显示了不同的预后组（log-rank p <； 0.001）。结论2.5D多源框架为胶质瘤患者的术前生存预测提供了一种临床可行的、无创的工具，可为胶质瘤患者提供个性化的治疗策略。

{"title":"Development of a hybrid 2.5D deep learning model for glioma survival prediction using T1-weighted MRI from the CGGA database","authors":"Kai Jin, Caixing Sun, Liang Xia","doi":"10.1016/j.ejro.2025.100697","DOIUrl":"10.1016/j.ejro.2025.100697","url":null,"abstract":"<div><h3>Background</h3><div>Current glioma survival prediction relies on invasive molecular profiling. To overcome this, a non-invasive deep learning framework using T1-weighted contrast-enhanced MRI (T1CE) was developed to predict overall survival. This framework addresses computational limitations associated with the volumetric analysis while preserving important spatial information.</div></div><div><h3>Methods</h3><div>We designed a hybrid 2.5D convolutional neural network to process multi-slice inputs, including the center slice and its adjacent slices, from 217 patients in the CGGA database. Transfer learning using ResNet and DenseNet architectures were employed to initialize the models. These models were subsequently fine-tuned with the Cox proportional hazards loss function. After the fine-tuning process was completed, the imaging signature was combined with clinical and molecular variables, including IDH and 1p19q status, to build an integrated model. Performance was evaluated via C-index, time-dependent AUC, and Kaplan-Meier analysis in independent training (70 %) and testing (30 %) cohorts.</div></div><div><h3>Results</h3><div>The Combined model achieved superior discrimination, with a training C-index of 0.819 (95 % CI: 0.758–0.880) and a testing C-index of 0.804 (95 % CI: 0.708–0.900). It significantly outperformed the isolated Radiomic, deep learning (2D and 2.5D), and Clinical models (all p < 0.05). Moreover, time-dependent ROC analysis demonstrated consistent model performance over 1–5 years, with AUC values ranging from 0.851 to 0.906. The stratified survival curves clearly revealed distinct prognostic groups (log-rank p < 0.001).</div></div><div><h3>Conclusions</h3><div>The 2.5D multi-source framework provides a clinically feasible, non-invasive tool for preoperative survival prediction, enabling personalized therapeutic strategies for glioma patients.</div></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":"15 ","pages":"Article 100697"},"PeriodicalIF":2.9,"publicationDate":"2025-10-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145415984","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Enhancing accuracy of detecting left atrial dilatation on CT pulmonary angiography 提高CT肺血管造影检测左房扩张的准确性

IF 2.9 Q3 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING

European Journal of Radiology Open

Pub Date : 2025-10-15 DOI: 10.1016/j.ejro.2025.100696

Louis Tapper , Samer Alabed , Ahmed Maiter , Andrzej Lejawka , Mahan Salehi , Krit Dwivedi , Pankaj Garg , David G. Kiely , Peter Metherall , Rob J. van der Geest , Kavita Karunasaagarar , Michael Sharkey , Andrew J. Swift

Introduction

Left atrial (LA) dilatation predicts several cardiovascular disorders. Identifying LA dilatation on computed tomography pulmonary angiography (CTPA) could aid diagnosis of cardiovascular disease. This study assessed an artificial intelligence (AI) segmentation model’s performance at detecting LA dilatation on CTPA.

Methods

Patients with suspected pulmonary hypertension (PH) who underwent CTPA and cardiac MRI (CMR) were retrospectively identified from a single centre registry. The LA was segmented by an AI tool for CTPA and a validated AI tool for CMR. LA volume measurements were categorised for LA dilatation based on existing threshold values. The expert radiologist's reports of the CTPA studies were also categorised for LA dilatation. Automated CTPA LA volumes and corresponding radiologist reports were compared against the reference standard of CMR.

Results

451 patients were included (mean age 64 ± 13 years, 62.5 % female, 85.8 % white). Automated LA volume measurements on CTPA showed strong positive correlation with those on CMR (ρ = 0.92, p < 0.001) with minimal bias on Bland-Altman analysis (-4 mL, 95 %CI −39 to +31 mL). Automated LA measurements on CTPA showed higher agreement with those on CMR (κ = 0.80) than the radiologist reports (κ = 0.62). Automated LA measurements on CTPA showed higher accuracy metrics (sensitivity 81.0 %, specificity 96.8 %, positive predictive value (PPV) 88.5 %, negative predictive value (NPV) 94.4 %) than the radiologist reports (sensitivity 66.7 %, specificity 93.1 %, PPV 74.5 %, NPV 90.2 %).

Conclusion

Deep learning increases the accuracy of LA volume measurements on non-ECG gated CTPA, improving radiologist performance in detecting LA dilatation.

左心房（LA）扩张可预测几种心血管疾病。通过ct肺血管造影（CTPA）识别LA扩张有助于心血管疾病的诊断。本研究评估了人工智能（AI）分割模型在检测CTPA上LA扩张方面的性能。方法回顾性分析单中心登记的疑似肺动脉高压（PH）患者行CTPA和心脏MRI （CMR）检查。LA通过CTPA的人工智能工具和CMR的经过验证的人工智能工具进行分割。根据现有阈值对LA扩张的容积测量进行分类。放射科专家的CTPA研究报告也被归类为LA扩张。自动CTPA LA体积和相应的放射科医生报告与CMR参考标准进行比较。结果纳入451例患者（平均年龄64岁 ± 13岁，女性62.5% %，白人85.8% %）。CTPA的自动LA体积测量与CMR的体积测量显示出很强的正相关（ρ = 0.92, p <； 0.001），Bland-Altman分析的偏差最小（-4 mL， 95% %CI - 39至+31 mL）。CTPA的自动LA测量值与CMR的一致性（κ = 0.80）高于放射科医生报告的一致性（κ = 0.62）。自动LA测量CTPA的准确性指标（敏感性81.0 %，特异性96.8 %，阳性预测值（PPV） 88.5 %，阴性预测值（NPV） 94.4 %）高于放射科医生报告的准确性指标（敏感性66.7 %，特异性93.1 %,PPV 74.5 %,NPV 90.2 %）。结论深度学习提高了非ecg门控CTPA左室容积测量的准确性，提高了放射科医生检测左室扩张的能力。

{"title":"Enhancing accuracy of detecting left atrial dilatation on CT pulmonary angiography","authors":"Louis Tapper , Samer Alabed , Ahmed Maiter , Andrzej Lejawka , Mahan Salehi , Krit Dwivedi , Pankaj Garg , David G. Kiely , Peter Metherall , Rob J. van der Geest , Kavita Karunasaagarar , Michael Sharkey , Andrew J. Swift","doi":"10.1016/j.ejro.2025.100696","DOIUrl":"10.1016/j.ejro.2025.100696","url":null,"abstract":"<div><h3>Introduction</h3><div>Left atrial (LA) dilatation predicts several cardiovascular disorders. Identifying LA dilatation on computed tomography pulmonary angiography (CTPA) could aid diagnosis of cardiovascular disease. This study assessed an artificial intelligence (AI) segmentation model’s performance at detecting LA dilatation on CTPA.</div></div><div><h3>Methods</h3><div>Patients with suspected pulmonary hypertension (PH) who underwent CTPA and cardiac MRI (CMR) were retrospectively identified from a single centre registry. The LA was segmented by an AI tool for CTPA and a validated AI tool for CMR. LA volume measurements were categorised for LA dilatation based on existing threshold values. The expert radiologist's reports of the CTPA studies were also categorised for LA dilatation. Automated CTPA LA volumes and corresponding radiologist reports were compared against the reference standard of CMR.</div></div><div><h3>Results</h3><div>451 patients were included (mean age 64 ± 13 years, 62.5 % female, 85.8 % white). Automated LA volume measurements on CTPA showed strong positive correlation with those on CMR (ρ = 0.92, p < 0.001) with minimal bias on Bland-Altman analysis (-4 mL, 95 %CI −39 to +31 mL). Automated LA measurements on CTPA showed higher agreement with those on CMR (κ = 0.80) than the radiologist reports (κ = 0.62). Automated LA measurements on CTPA showed higher accuracy metrics (sensitivity 81.0 %, specificity 96.8 %, positive predictive value (PPV) 88.5 %, negative predictive value (NPV) 94.4 %) than the radiologist reports (sensitivity 66.7 %, specificity 93.1 %, PPV 74.5 %, NPV 90.2 %).</div></div><div><h3>Conclusion</h3><div>Deep learning increases the accuracy of LA volume measurements on non-ECG gated CTPA, improving radiologist performance in detecting LA dilatation.</div></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":"15 ","pages":"Article 100696"},"PeriodicalIF":2.9,"publicationDate":"2025-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145319582","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A systematic review and meta-analysis of imaging characteristics and upgrade rates in noninvasive lobular neoplasia of the breast 对乳腺非侵袭性小叶瘤的影像学特征和升级率的系统回顾和荟萃分析

IF 2.9 Q3 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING

European Journal of Radiology Open

Pub Date : 2025-10-15 DOI: 10.1016/j.ejro.2025.100691

Fatemeh Shakki Katouli , Negin Salehi , Faezeh Soveyzi , Mina Abedi , Parya Valizadeh , Hamed Ghorani , Seyedeh Melika Hashemi , Jayran Zebardast , Madjid Shakiba , Sadaf Alipour

Background

Non-invasive lobular neoplasia (NLN) encompasses a range of lobular breast lesions that may precede invasive breast cancer. Microcalcifications detected through mammography play a crucial role in evaluating breast lesions and are often associated with NLN. This study focuses on the prevalence and significance of microcalcifications in NLN, noting that they can be the sole radiographic finding in many cases. While mammography is highly sensitive for detecting microcalcifications, it has limitations in diagnosing NLN, as some lesions may not show up on scans. Advanced imaging techniques like magnetic resonance imaging (MRI) offer improved diagnostic accuracy, particularly in dense breast tissue, but more research is needed for their routine use. Additionally, the risk of NLN progressing to malignant lesions highlights the importance of vigilant monitoring and management. This study aims to analyze the relationship between microcalcifications and NLN, addressing progression risks and implications for clinical practice.

Method

This systematic review and meta-analysis (CRD42022346891) involved a comprehensive search of databases such as PubMed and Scopus from 2000 to 2023, using keywords related to lobular carcinoma and mammography. Eligible English-language studies included those reporting mammographic findings of pure NLN lesions confirmed by histopathologic evaluation and surgical excision. Exclusion criteria involved studies without surgical results or definitive imaging findings. Two independent reviewers assessed titles and abstracts, resolving discrepancies as needed. Data were systematically extracted using a standardized form, with the selection process depicted in a PRISMA flow diagram.

Result

Meta-analysis of included studies revealed that the pooled proportion of any mammographic microcalcification among all NLN lesions was 0.70 (95 % CI: 0.63–0.76). Pure microcalcification (without an associated mass or distortion) was the most common presentation, with a pooled proportion of 0.67 (95 % CI: 0.60–0.74) among all lesions and 0.99 (95 % CI: 0.96–1.00) among lesions presenting with any microcalcification. The overall pooled upgrade rate to malignancy was 0.18 (95 % CI: 0.11–0.25), with a significantly higher rate for lobular carcinoma in situ (LCIS) at 0.22 (95 % CI: 0.15–0.30) compared to atypical lobular hyperplasia (ALH) at 0.06 (95 % CI: 0.01–0.14). Microcalcifications were present in the majority of upgraded lesions (pooled proportion: 0.78, 95 % CI: 0.69–0.87). A small but significant proportion of lesions (0.08, 95 % CI: 0.03–0.17) had no mammographic findings. All pooled estimates showed high heterogeneity. Sensitivity analysis confirmed the robustness of the results, while Egger's test indicated potential publication bias.

Conclusion

In conclusion this study highlights the significant prevalence of microcalcifications i

背景：非浸润性乳腺小叶瘤（NLN）包括一系列可能发生于浸润性乳腺癌的乳腺小叶病变。通过乳房x线摄影检测到的微钙化在评估乳腺病变中起着至关重要的作用，通常与NLN相关。本研究的重点是NLN中微钙化的患病率和意义，注意到它们在许多病例中可能是唯一的影像学发现。虽然乳房x光检查对检测微钙化非常敏感，但它在诊断NLN方面有局限性，因为一些病变可能无法在扫描中显示出来。像磁共振成像（MRI）这样的先进成像技术提高了诊断的准确性，特别是在致密的乳腺组织中，但需要更多的研究才能将其常规使用。此外，NLN发展为恶性病变的风险突出了警惕监测和管理的重要性。本研究旨在分析微钙化与NLN之间的关系，解决进展风险及其对临床实践的影响。方法本系统综述和荟萃分析（CRD42022346891）对2000 - 2023年PubMed、Scopus等数据库进行综合检索，检索关键词为小叶癌和乳腺x线摄影。符合条件的英语研究包括那些经组织病理学评估和手术切除证实的纯NLN病变的乳房x光检查结果。排除标准包括没有手术结果或明确影像学发现的研究。两名独立审稿人评估标题和摘要，根据需要解决差异。使用标准化表格系统地提取数据，并在PRISMA流程图中描述选择过程。结果纳入研究的荟萃分析显示，所有NLN病变中任何乳房x线摄影微钙化的总比例为0.70（95 % CI: 0.63-0.76）。纯微钙化（无相关肿块或扭曲）是最常见的表现，在所有病变中合并比例为0.67(95 % CI: 0.60-0.74)，在任何微钙化病变中合并比例为0.99（95 % CI: 0.96-1.00）。总体合并恶性升级率为0.18(95 % CI: 0.11-0.25)，小叶原位癌（LCIS）的发生率为0.22(95 % CI: 0.15-0.30)，而非典型小叶增生（ALH）的发生率为0.06（95 % CI: 0.01-0.14）。大多数升级病变存在微钙化（合并比例：0.78,95 % CI: 0.69-0.87）。一小部分但有意义的病变（0.08,95 % CI: 0.03-0.17）没有乳房x线检查结果。所有汇总估计均显示高度异质性。敏感性分析证实了结果的稳健性，而Egger的检验表明了潜在的发表偏倚。总之，本研究强调了NLN病例中微钙化的显著患病率及其作为诊断特征的关键作用。尽管它们与疾病进展有关，但微钙化并不是恶性肿瘤升级的可靠预测因子。进一步的评估是必要的，以了解其临床意义，并改善对NLN患者的管理策略。

{"title":"A systematic review and meta-analysis of imaging characteristics and upgrade rates in noninvasive lobular neoplasia of the breast","authors":"Fatemeh Shakki Katouli , Negin Salehi , Faezeh Soveyzi , Mina Abedi , Parya Valizadeh , Hamed Ghorani , Seyedeh Melika Hashemi , Jayran Zebardast , Madjid Shakiba , Sadaf Alipour","doi":"10.1016/j.ejro.2025.100691","DOIUrl":"10.1016/j.ejro.2025.100691","url":null,"abstract":"<div><h3>Background</h3><div>Non-invasive lobular neoplasia (NLN) encompasses a range of lobular breast lesions that may precede invasive breast cancer. Microcalcifications detected through mammography play a crucial role in evaluating breast lesions and are often associated with NLN. This study focuses on the prevalence and significance of microcalcifications in NLN, noting that they can be the sole radiographic finding in many cases. While mammography is highly sensitive for detecting microcalcifications, it has limitations in diagnosing NLN, as some lesions may not show up on scans. Advanced imaging techniques like magnetic resonance imaging (MRI) offer improved diagnostic accuracy, particularly in dense breast tissue, but more research is needed for their routine use. Additionally, the risk of NLN progressing to malignant lesions highlights the importance of vigilant monitoring and management. This study aims to analyze the relationship between microcalcifications and NLN, addressing progression risks and implications for clinical practice.</div></div><div><h3>Method</h3><div>This systematic review and meta-analysis (CRD42022346891) involved a comprehensive search of databases such as PubMed and Scopus from 2000 to 2023, using keywords related to lobular carcinoma and mammography. Eligible English-language studies included those reporting mammographic findings of pure NLN lesions confirmed by histopathologic evaluation and surgical excision. Exclusion criteria involved studies without surgical results or definitive imaging findings. Two independent reviewers assessed titles and abstracts, resolving discrepancies as needed. Data were systematically extracted using a standardized form, with the selection process depicted in a PRISMA flow diagram.</div></div><div><h3>Result</h3><div>Meta-analysis of included studies revealed that the pooled proportion of any mammographic microcalcification among all NLN lesions was 0.70 (95 % CI: 0.63–0.76). Pure microcalcification (without an associated mass or distortion) was the most common presentation, with a pooled proportion of 0.67 (95 % CI: 0.60–0.74) among all lesions and 0.99 (95 % CI: 0.96–1.00) among lesions presenting with any microcalcification. The overall pooled upgrade rate to malignancy was 0.18 (95 % CI: 0.11–0.25), with a significantly higher rate for lobular carcinoma in situ (LCIS) at 0.22 (95 % CI: 0.15–0.30) compared to atypical lobular hyperplasia (ALH) at 0.06 (95 % CI: 0.01–0.14). Microcalcifications were present in the majority of upgraded lesions (pooled proportion: 0.78, 95 % CI: 0.69–0.87). A small but significant proportion of lesions (0.08, 95 % CI: 0.03–0.17) had no mammographic findings. All pooled estimates showed high heterogeneity. Sensitivity analysis confirmed the robustness of the results, while Egger's test indicated potential publication bias.</div></div><div><h3>Conclusion</h3><div>In conclusion this study highlights the significant prevalence of microcalcifications i","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":"15 ","pages":"Article 100691"},"PeriodicalIF":2.9,"publicationDate":"2025-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145319583","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Photon counting CT improves coronary stent imaging and fat attenuation index assessment across reconstruction modes 光子计数CT改善冠状动脉支架成像和脂肪衰减指数评估跨重建模式

IF 2.9 Q3 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING

European Journal of Radiology Open

Pub Date : 2025-10-13 DOI: 10.1016/j.ejro.2025.100695

Yujie Gao , Yaru Yang , Qiuju Hu , Yane Zhao , Yong Yuan , Jiliang Chen , Bangjun Guo , Dongsheng Jin , song luo , Guangming Lu

Objectives

To investigate the effect of photon counting CT (PCCT) ultra–high resolution (UHR) mode on coronary stent visualization and fat attenuation index (FAI) assessment in individuals with percutaneous coronary intervention (PCI).

Methods

Patients who underwent PCI with stent placement and following coronary CT angiography (CCTA) by using a PCCT system were enrolled during January to August 2024. Simulated energy integrating detector CT (EID-CT) images (0.8 mm) were reconstructed with kernel Bv48, while UHR images (0.2 mm) were reconstructed using kernel Bv60 and Bv72. Objective and subjective image quality, stent-specific FAI and peri-stent FAI was evaluated.

Results

A total of 41 patients with 51 isolated stents (69.76 ± 8.75 years; 34 men) were included. Compared to simulated EID-CT, 0.2 mm Bv72 images showed larger in-stent diameters and reduced blooming artifacts (all p < 0.001). Subjective image quality scores for 0.2 mm UHR images were superior to those for simulated EID-CT (all p < 0.01). The stent-specific FAI and peri-stent FAI of 51 isolated stents was lower in the 0.2 mm UHR images than in simulated EID-CT (all p < 0.05). The reconstruction mode of 0.2 mm Bv72 showed the ability of stent-specific FAI and peri-stent FAI to distinguish stents with in-stent restenosis (ISR) < 50 % in diameter from stents without ISR, with cut-off value of −98HU and −99.5HU, respectively.

Conclusions

PCCT UHR mode improved the image quality of coronary stents, reduced the FAI values and provided cut-off values based on stent-specific FAI and peri-stent FAI.

目的探讨光子计数CT （PCCT）超高分辨率（UHR）模式对经皮冠状动脉介入治疗（PCI）患者冠脉支架显像及脂肪衰减指数（FAI）评估的影响。方法于2024年1月至8月，采用PCCT系统行PCI支架置入术及冠状动脉CT血管造影（CCTA）的患者入组。模拟能量积分检测器CT （EID-CT）图像（0.8 mm）用内核Bv48重建，UHR图像（0.2 mm）用内核Bv60和Bv72重建。评估客观和主观图像质量、支架特异性FAI和支架周围FAI。结果共纳入51例孤立支架患者41例（69.76 ± 8.75岁，男性34例）。与模拟EID-CT相比，0.2 mm Bv72图像显示支架内直径更大，假影减少（p均 <； 0.001）。0.2 mm UHR图像的主观图像质量评分优于模拟EID-CT （p均 <； 0.01）。51个离体支架在0.2 mm UHR图像上的支架特异性FAI和支架周围FAI均低于模拟reid - ct （p均 <； 0.05）。重建模式为0.2 mm Bv72，表明支架特异性FAI和支架周围FAI能够区分支架内再狭窄（ISR） <； 50 %直径的支架与无ISR的支架，截断值分别为- 98HU和- 99.5HU。结论spcct UHR模式提高了冠状动脉支架的图像质量，降低了FAI值，并提供了基于支架特异性FAI和支架周围FAI的截断值。

{"title":"Photon counting CT improves coronary stent imaging and fat attenuation index assessment across reconstruction modes","authors":"Yujie Gao , Yaru Yang , Qiuju Hu , Yane Zhao , Yong Yuan , Jiliang Chen , Bangjun Guo , Dongsheng Jin , song luo , Guangming Lu","doi":"10.1016/j.ejro.2025.100695","DOIUrl":"10.1016/j.ejro.2025.100695","url":null,"abstract":"<div><h3>Objectives</h3><div>To investigate the effect of photon counting CT (PCCT) ultra–high resolution (UHR) mode on coronary stent visualization and fat attenuation index (FAI) assessment in individuals with percutaneous coronary intervention (PCI).</div></div><div><h3>Methods</h3><div>Patients who underwent PCI with stent placement and following coronary CT angiography (CCTA) by using a PCCT system were enrolled during January to August 2024. Simulated energy integrating detector CT (EID-CT) images (0.8 mm) were reconstructed with kernel Bv48, while UHR images (0.2 mm) were reconstructed using kernel Bv60 and Bv72. Objective and subjective image quality, stent-specific FAI and peri-stent FAI was evaluated.</div></div><div><h3>Results</h3><div>A total of 41 patients with 51 isolated stents (69.76 ± 8.75 years; 34 men) were included. Compared to simulated EID-CT, 0.2 mm Bv72 images showed larger in-stent diameters and reduced blooming artifacts (all p < 0.001). Subjective image quality scores for 0.2 mm UHR images were superior to those for simulated EID-CT (all p < 0.01). The stent-specific FAI and peri-stent FAI of 51 isolated stents was lower in the 0.2 mm UHR images than in simulated EID-CT (all p < 0.05). The reconstruction mode of 0.2 mm Bv72 showed the ability of stent-specific FAI and peri-stent FAI to distinguish stents with in-stent restenosis (ISR) < 50 % in diameter from stents without ISR, with cut-off value of −98HU and −99.5HU, respectively.</div></div><div><h3>Conclusions</h3><div>PCCT UHR mode improved the image quality of coronary stents, reduced the FAI values and provided cut-off values based on stent-specific FAI and peri-stent FAI.</div></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":"15 ","pages":"Article 100695"},"PeriodicalIF":2.9,"publicationDate":"2025-10-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145319584","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Deep learning for Three‐Class Classification of ground-glass nodules on non-enhanced chest CT: A multicenter comparative study of CNN architectures 深度学习在非增强胸部CT磨玻璃结节三级分类中的应用：CNN架构的多中心对比研究

IF 2.9 Q3 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING

European Journal of Radiology Open

Pub Date : 2025-10-08 DOI: 10.1016/j.ejro.2025.100690

Meihua Shao , Jian Wang , Lin Zhu , Jianfei Tu , Guohua Cheng , Linyang He , Hengfeng Shi , Cui Zhang , Hong Yu

Objective

To develop, validate, and compare four three-dimensional (3D) convolutional neural network (CNN) models for differentiating ground-glass nodules (GGNs) on non-contrast chest computed tomography (CT) scans, specifically classifying them as adenomatous hyperplasia (AAH)/adenocarcinoma in situ (AIS), minimally invasive adenocarcinoma (MIA), or invasive adenocarcinoma (IA).

Materials and methods

This multi-center study retrospectively enrolled 4284 consecutive patients with surgically resected and pathologically confirmed AAH/AIS, MIA, or IA from four hospitals between January 2015 and December 2023. GGNs were randomly partitioned into a training set (n = 3083, 72 %) and a validation set (n = 1277, 28 %). Four 3D deep learning models (Res2Net 3D, DenseNet3D, ResNet50 3D, Vision Transformer 3D) were implemented for GGN segmentation and three-class classification. Additionally, variants of the Res2Net 3D model were developed by incorporating clinical and CT features: Res2Net 3D_w2 (sex, age), Res2Net 3D_w6 (adding lesion size, location, and smoking history), and Res2Net 3D_w10 (sex, age, location, the mean, maximum, and standard deviation of CT attenuation, nodule volume, volume ratio, volume ratio within the left/right lung, and the maximum CT value of the entire lung). Model performance was evaluated using accuracy, recall, precision, F1-score, and area under the receiver operating characteristic curve (AUC).

Results

Res2Net 3D outperformed others, achieving AUCs of 0.91 (AAH/AIS), 0.88 (MIA), and 0.92 (IA). Its F1-scores were 0.416, 0.500, and 0.929, respectively. All Res2Net variants achieved accuracies between 0.83–0.84.

Conclusion

The Res2Net 3D model accurately differentiates GGN subtypes using non-contrast CT, showing high performance, especially for invasive adenocarcinoma.

目的建立、验证并比较4种三维（3D）卷积神经网络（CNN）模型在非对比胸部计算机断层扫描（CT）上鉴别毛玻璃结节（ggn），并将其分类为腺瘤性增生(AAH)/原位腺癌（AIS）、微创腺癌（MIA）和浸润性腺癌（IA）。材料和方法本多中心研究回顾性纳入了2015年1月至2023年12月间4家医院4284例手术切除并病理证实的AAH/AIS、MIA或IA患者。将ggn随机划分为训练集（n = 3083,72 %）和验证集（n = 1277,28 %）。采用Res2Net 3D、DenseNet3D、ResNet50 3D、Vision Transformer 3D四个三维深度学习模型对GGN进行分割和三类分类。此外，通过结合临床和CT特征，开发了Res2Net 3D模型的不同版本：Res2Net 3D_w2（性别、年龄），Res2Net 3D_w6（添加病变大小、位置和吸烟史），Res2Net 3D_w10（性别、年龄、位置、CT衰减平均值、最大值和标准差、结节体积、体积比、左右肺体积比、全肺最大CT值）。通过准确性、召回率、精密度、f1评分和受试者工作特征曲线（AUC）下面积来评估模型的性能。结果res2net 3D的auc分别为0.91 （AAH/AIS）、0.88 （MIA）和0.92 （IA）。其f1得分分别为0.416、0.500和0.929。所有Res2Net变体的准确率都在0.83-0.84之间。结论Res2Net三维模型在非对比CT上能准确鉴别GGN亚型，对浸润性腺癌具有较高的鉴别价值。

{"title":"Deep learning for Three‐Class Classification of ground-glass nodules on non-enhanced chest CT: A multicenter comparative study of CNN architectures","authors":"Meihua Shao , Jian Wang , Lin Zhu , Jianfei Tu , Guohua Cheng , Linyang He , Hengfeng Shi , Cui Zhang , Hong Yu","doi":"10.1016/j.ejro.2025.100690","DOIUrl":"10.1016/j.ejro.2025.100690","url":null,"abstract":"<div><h3>Objective</h3><div>To develop, validate, and compare four three-dimensional (3D) convolutional neural network (CNN) models for differentiating ground-glass nodules (GGNs) on non-contrast chest computed tomography (CT) scans, specifically classifying them as adenomatous hyperplasia (AAH)/adenocarcinoma in situ (AIS), minimally invasive adenocarcinoma (MIA), or invasive adenocarcinoma (IA).</div></div><div><h3>Materials and methods</h3><div>This multi-center study retrospectively enrolled 4284 consecutive patients with surgically resected and pathologically confirmed AAH/AIS, MIA, or IA from four hospitals between January 2015 and December 2023. GGNs were randomly partitioned into a training set (n = 3083, 72 %) and a validation set (n = 1277, 28 %). Four 3D deep learning models (Res2Net 3D, DenseNet3D, ResNet50 3D, Vision Transformer 3D) were implemented for GGN segmentation and three-class classification. Additionally, variants of the Res2Net 3D model were developed by incorporating clinical and CT features: Res2Net 3D_w2 (sex, age), Res2Net 3D_w6 (adding lesion size, location, and smoking history), and Res2Net 3D_w10 (sex, age, location, the mean, maximum, and standard deviation of CT attenuation, nodule volume, volume ratio, volume ratio within the left/right lung, and the maximum CT value of the entire lung). Model performance was evaluated using accuracy, recall, precision, F1-score, and area under the receiver operating characteristic curve (AUC).</div></div><div><h3>Results</h3><div>Res2Net 3D outperformed others, achieving AUCs of 0.91 (AAH/AIS), 0.88 (MIA), and 0.92 (IA). Its F1-scores were 0.416, 0.500, and 0.929, respectively. All Res2Net variants achieved accuracies between 0.83–0.84.</div></div><div><h3>Conclusion</h3><div>The Res2Net 3D model accurately differentiates GGN subtypes using non-contrast CT, showing high performance, especially for invasive adenocarcinoma.</div></div>","PeriodicalId":38076,"journal":{"name":"European Journal of Radiology Open","volume":"15 ","pages":"Article 100690"},"PeriodicalIF":2.9,"publicationDate":"2025-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145264884","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

首页上一页

1
2
3
4
5
6
7
...
10

下一页尾页

类型

全部化学•材料生命科学医学物理工程技术环境•农林材料科学地球科学法学管理学化学环境科学与生态学计算机科学教育学经济学农林科学人文科学生物学数学物理与天体物理心理学综合性期刊其他工业工程理学历史学农学文学信息工程

数据库

全部 ACS Publications Elsevier ieeexplore Springer The Royal Society of Chemistry Wiley

期刊

European Journal of Radiology Open

全部 Acc. Chem. Res. ACS Applied Bio Materials ACS Appl. Electron. Mater. ACS Appl. Energy Mater. ACS Appl. Mater. Interfaces ACS Appl. Nano Mater. ACS Appl. Polym. Mater. ACS BIOMATER-SCI ENG ACS Catal. ACS Cent. Sci. ACS Chem. Biol. ACS Chemical Health & Safety ACS Chem. Neurosci. ACS Comb. Sci. ACS Earth Space Chem. ACS Energy Lett. ACS Infect. Dis. ACS Macro Lett. ACS Mater. Lett. ACS Med. Chem. Lett. ACS Nano ACS Omega ACS Photonics ACS Sens. ACS Sustainable Chem. Eng. ACS Synth. Biol. Anal. Chem. BIOCHEMISTRY-US Bioconjugate Chem. BIOMACROMOLECULES Chem. Res. Toxicol. Chem. Rev. Chem. Mater. CRYST GROWTH DES ENERG FUEL Environ. Sci. Technol. Environ. Sci. Technol. Lett. Eur. J. Inorg. Chem. IND ENG CHEM RES Inorg. Chem. J. Agric. Food. Chem. J. Chem. Eng. Data J. Chem. Educ. J. Chem. Inf. Model. J. Chem. Theory Comput. J. Med. Chem. J. Nat. Prod. J PROTEOME RES J. Am. Chem. Soc. LANGMUIR MACROMOLECULES Mol. Pharmaceutics Nano Lett. Org. Lett. ORG PROCESS RES DEV ORGANOMETALLICS J. Org. Chem. J. Phys. Chem. J. Phys. Chem. A J. Phys. Chem. B J. Phys. Chem. C J. Phys. Chem. Lett. Analyst Anal. Methods Biomater. Sci. Catal. Sci. Technol. Chem. Commun. Chem. Soc. Rev. CHEM EDUC RES PRACT CRYSTENGCOMM Dalton Trans. Energy Environ. Sci. ENVIRON SCI-NANO ENVIRON SCI-PROC IMP ENVIRON SCI-WAT RES Faraday Discuss. Food Funct. Green Chem. Inorg. Chem. Front. Integr. Biol. J. Anal. At. Spectrom. J. Mater. Chem. A J. Mater. Chem. B J. Mater. Chem. C Lab Chip Mater. Chem. Front. Mater. Horiz. MEDCHEMCOMM Metallomics Mol. Biosyst. Mol. Syst. Des. Eng. Nanoscale Nanoscale Horiz. Nat. Prod. Rep. New J. Chem. Org. Biomol. Chem. Org. Chem. Front. PHOTOCH PHOTOBIO SCI PCCP Polym. Chem.

﹀