Pub Date : 2025-12-21DOI: 10.1016/j.clinimag.2025.110705
Teresa Chapman , Cory M. Pfeifer , Jennifer N. Kucera , Kristin A. Leland , David M. Biko , Paula N. Dickson , Desi M. Schiess , Sarah S. Milla
{"title":"Ensuring radiologist camaraderie and excellence in teaching with virtual pediatric radiology workflows","authors":"Teresa Chapman , Cory M. Pfeifer , Jennifer N. Kucera , Kristin A. Leland , David M. Biko , Paula N. Dickson , Desi M. Schiess , Sarah S. Milla","doi":"10.1016/j.clinimag.2025.110705","DOIUrl":"10.1016/j.clinimag.2025.110705","url":null,"abstract":"","PeriodicalId":50680,"journal":{"name":"Clinical Imaging","volume":"130 ","pages":"Article 110705"},"PeriodicalIF":1.5,"publicationDate":"2025-12-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145878982","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-21DOI: 10.1016/j.clinimag.2025.110706
Amir Hassankhani , Parya Valizadeh , Payam Jannatdoust , Melika Amoukhteh , Abbas Mohammadi , Cem Bilgin , Ali Gholamrezanezhad , Ali Haq
Purpose
To evaluate the diagnostic accuracy of Chest ultrasound (CUS) for pulmonary TB (PTB).
Methods
A systematic review and meta-analysis was conducted following established guidelines. PubMed, Scopus, and Embase were searched up to July 1st, 2025. Studies reporting CUS diagnostic accuracy for PTB were included. Data were extracted and analyzed using R software.
Results
Five studies with 548 participants (256 confirmed PTB cases) were included. Pooled sensitivity of CUS for detecting any abnormal lung finding was high at 88.4 % (95 % CI: 80.2–93.5 %), but specificity was limited at 42.0 % (95 % CI: 20.2–67.5 %). Irregular pleural lines showed 71.7 % sensitivity and 41.7 % specificity. Consolidation (any location) had 66.4 % sensitivity and 62.6 % specificity; apical consolidation had the highest specificity (89.0 %) but low sensitivity (43.6 %). B-lines and pleural effusion showed poor diagnostic accuracy. Likelihood ratios (LRs) for all features fell below thresholds for confident rule-in or rule-out (LR+ < 10, LR− > 0.1).
Conclusion
CUS is a sensitive adjunct for PTB detection but lacks sufficient specificity and likelihood ratio values to serve as a standalone diagnostic tool. Standardized protocols and improved implementation strategies are needed to enhance its diagnostic performance.
{"title":"A systematic review and meta-analysis on the diagnostic performance of chest ultrasound for pulmonary tuberculosis","authors":"Amir Hassankhani , Parya Valizadeh , Payam Jannatdoust , Melika Amoukhteh , Abbas Mohammadi , Cem Bilgin , Ali Gholamrezanezhad , Ali Haq","doi":"10.1016/j.clinimag.2025.110706","DOIUrl":"10.1016/j.clinimag.2025.110706","url":null,"abstract":"<div><h3>Purpose</h3><div>To evaluate the diagnostic accuracy of Chest ultrasound (CUS) for pulmonary TB (PTB).</div></div><div><h3>Methods</h3><div>A systematic review and meta-analysis was conducted following established guidelines. PubMed, Scopus, and Embase were searched up to July 1st, 2025. Studies reporting CUS diagnostic accuracy for PTB were included. Data were extracted and analyzed using R software.</div></div><div><h3>Results</h3><div>Five studies with 548 participants (256 confirmed PTB cases) were included. Pooled sensitivity of CUS for detecting any abnormal lung finding was high at 88.4 % (95 % CI: 80.2–93.5 %), but specificity was limited at 42.0 % (95 % CI: 20.2–67.5 %). Irregular pleural lines showed 71.7 % sensitivity and 41.7 % specificity. Consolidation (any location) had 66.4 % sensitivity and 62.6 % specificity; apical consolidation had the highest specificity (89.0 %) but low sensitivity (43.6 %). B-lines and pleural effusion showed poor diagnostic accuracy. Likelihood ratios (LRs) for all features fell below thresholds for confident rule-in or rule-out (LR<sup>+</sup> < 10, LR<sup>−</sup> > 0.1).</div></div><div><h3>Conclusion</h3><div>CUS is a sensitive adjunct for PTB detection but lacks sufficient specificity and likelihood ratio values to serve as a standalone diagnostic tool. Standardized protocols and improved implementation strategies are needed to enhance its diagnostic performance.</div></div>","PeriodicalId":50680,"journal":{"name":"Clinical Imaging","volume":"130 ","pages":"Article 110706"},"PeriodicalIF":1.5,"publicationDate":"2025-12-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145835248","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-19DOI: 10.1016/j.clinimag.2025.110701
S. Dhanya Dedeepya , Vaishali Goel , Nivedita Nikhil Desai
{"title":"Comment on “Diagnostic accuracy and limitations of intravoxel incoherent motion diffusion-weighted imaging for differentiating breast tumors: A systematic review and meta-analysis”","authors":"S. Dhanya Dedeepya , Vaishali Goel , Nivedita Nikhil Desai","doi":"10.1016/j.clinimag.2025.110701","DOIUrl":"10.1016/j.clinimag.2025.110701","url":null,"abstract":"","PeriodicalId":50680,"journal":{"name":"Clinical Imaging","volume":"130 ","pages":"Article 110701"},"PeriodicalIF":1.5,"publicationDate":"2025-12-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145821869","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-18DOI: 10.1016/j.clinimag.2025.110698
Jacob Jahn , Nathaniel W. Jenkins , Miriyam Ghali , Allison Boden , Amiethab Aiyer , Ty Subhawong
Background
The Lawrence and Botte (LB) classification of proximal fifth metatarsal fractures guides clinical management, yet its reliability has been sparsely evaluated. This study investigated intra- and inter-rater reliability of the LB system among trained observers.
Methods
A retrospective chart review identified patients with proximal fifth metatarsal fractures. Deidentified radiographs (AP, lateral, oblique) were reviewed by two orthopedic surgeons and one radiologist. Each observer classified fractures using the LB system in two surveys separated by at least two weeks. Identical radiographs were reordered between surveys to assess inter- and intra-rater reliability. Statistical analysis included Kruskal–Wallis, Friedman, weighted kappa, Cohen's kappa, and Fleiss' kappa.
Results
200 patients were screened, and 85 radiographs included. Inter-rater reliability in Survey A showed no significant differences by Kruskal–Wallis; Friedman testing suggested minimal but significant variability. In Survey B, Kruskal–Wallis indicated significant differences in rankings. Weighted kappa coefficients for inter-rater reliability ranged 0.58–0.69 (moderate–substantial agreement). Intra-rater reliability ranged 0.61–0.82, with observer 2 highest (0.82). Cohen's kappa values were 0.58–0.71. Agreement between surveys (Fleiss' kappa) was 0.56, indicating moderate reliability.
Conclusion
The LB classification system demonstrates moderate inter-rater and substantial intra-rater reliability, though variability persists between raters and time points. While reasonably consistent, the findings highlight subjectivity in interpretation and suggest potential benefit in simplified schemes to improve agreement.
背景：第五跖骨近端骨折的Lawrence and Botte （LB）分类指导临床处理，但其可靠性尚未得到充分评估。本研究调查了LB系统在训练过的观察者中的内部和内部可靠性。方法：回顾性分析第5跖骨近端骨折患者。两名骨科医生和一名放射科医生检查了未识别的x线片（正位，侧位，斜位）。每个观察人员在两次调查中使用LB系统对裂缝进行分类，间隔至少两周。在两次调查之间重新排列相同的x线片，以评估评估组间和组内的可靠性。统计分析包括Kruskal-Wallis、Friedman、加权kappa、Cohen’s kappa和Fleiss’s kappa。结果：共筛选200例患者，纳入85张x线片。Kruskal-Wallis测验A的信度差异不显著；弗里德曼测试表明，变异很小，但很重要。在调查B中，Kruskal-Wallis指出了排名的显著差异。评级间信度的加权kappa系数范围为0.58-0.69（中等-实质性一致）。内部信度范围为0.61-0.82，观察者2最高（0.82）。Cohen’s kappa值为0.58 ~ 0.71。调查间的一致性（Fleiss’kappa）为0.56，信度中等。结论：尽管评分者和时间点之间存在差异，但LB分类系统在评分者之间和评分者内部表现出适度的可靠性。虽然结果相当一致，但研究结果强调了解释的主观性，并提出了简化方案以提高一致性的潜在好处。
{"title":"Validation of Lawrence and Botte classification of proximal fifth metatarsal fractures","authors":"Jacob Jahn , Nathaniel W. Jenkins , Miriyam Ghali , Allison Boden , Amiethab Aiyer , Ty Subhawong","doi":"10.1016/j.clinimag.2025.110698","DOIUrl":"10.1016/j.clinimag.2025.110698","url":null,"abstract":"<div><h3>Background</h3><div>The Lawrence and Botte (LB) classification of proximal fifth metatarsal fractures guides clinical management, yet its reliability has been sparsely evaluated. This study investigated intra- and inter-rater reliability of the LB system among trained observers.</div></div><div><h3>Methods</h3><div>A retrospective chart review identified patients with proximal fifth metatarsal fractures. Deidentified radiographs (AP, lateral, oblique) were reviewed by two orthopedic surgeons and one radiologist. Each observer classified fractures using the LB system in two surveys separated by at least two weeks. Identical radiographs were reordered between surveys to assess inter- and intra-rater reliability. Statistical analysis included Kruskal–Wallis, Friedman, weighted kappa, Cohen's kappa, and Fleiss' kappa.</div></div><div><h3>Results</h3><div>200 patients were screened, and 85 radiographs included. Inter-rater reliability in Survey A showed no significant differences by Kruskal–Wallis; Friedman testing suggested minimal but significant variability. In Survey B, Kruskal–Wallis indicated significant differences in rankings. Weighted kappa coefficients for inter-rater reliability ranged 0.58–0.69 (moderate–substantial agreement). Intra-rater reliability ranged 0.61–0.82, with observer 2 highest (0.82). Cohen's kappa values were 0.58–0.71. Agreement between surveys (Fleiss' kappa) was 0.56, indicating moderate reliability.</div></div><div><h3>Conclusion</h3><div>The LB classification system demonstrates moderate inter-rater and substantial intra-rater reliability, though variability persists between raters and time points. While reasonably consistent, the findings highlight subjectivity in interpretation and suggest potential benefit in simplified schemes to improve agreement.</div></div>","PeriodicalId":50680,"journal":{"name":"Clinical Imaging","volume":"130 ","pages":"Article 110698"},"PeriodicalIF":1.5,"publicationDate":"2025-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145866613","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-17DOI: 10.1016/j.clinimag.2025.110700
Habip Eser Akkaya , Erhan Kaya , Rasim Gökmen , Muhammed Semih Gedik , Dorukan İnanç Akpınar
Objective
This study examined global research trends in Radiology, Nuclear Medicine, and Medical Imaging by analyzing the 500 most-cited articles in the Web of Science (WoS) Core Collection.
Methods
A bibliometric search was conducted on June 15, 2025. Biblioshiny and VOSviewer 1.6.20 were used for network visualization, including institutional collaboration, co-authorship, keyword co-occurrence, and country-level contributions. Temporal patterns were analyzed with Python 3.13.3, and descriptive statistics summarized publication data.
Results
Harvard University led institutional contributions with 54 publications, followed by Massachusetts General Hospital (n = 49), University of Oxford (n = 35), Washington University (n = 29), and University of Texas (n = 26). The United States accounted for 53.4 % of all outputs, followed by the United Kingdom (21.6 %), Germany (12 %), Canada (9 %), and France (8 %). Among authors, Stephen M. Smith contributed most (19 publications), followed by Jenkinson, M (n = 14), and Friston, KJ (n = 13). The most frequent keywords were “MRI” (n = 65), “Brain” (n = 43), “fMRI” (n = 37), “Segmentation” (n = 25), and “PET” (n = 24). In addition to leading all journals in citation impact (citations per article), Neuroimage was also identified as the most productive journal overall. Regarding the average citation impact, the top-performing entities in their respective categories were: the University of Oxford (among organizations), Germany (among countries), Smith Stephen M (among authors), and the journal Neuroimage (among journals). Emerging terms included “deep learning” and “artificial intelligence.” The most-cited article was Ronneberger et al.'s U-Net (2015), cited 63,448 times.
Conclusion
High-impact radiology research is concentrated in North America and Western Europe, with neuroimaging and artificial intelligence representing key emerging domains. These insights provide a roadmap for research prioritization and collaboration strategies.
目的通过分析Web of Science （WoS）核心馆藏中被引用次数最多的500篇文章，研究放射学、核医学和医学影像学的全球研究趋势。方法于2025年6月15日进行文献计量学检索。使用Biblioshiny和VOSviewer 1.6.20进行网络可视化，包括机构合作、合著、关键词共现和国家层面的贡献。使用Python 3.13.3对时间模式进行分析，并对发表数据进行描述性统计。结果哈佛大学共发表54篇论文，是发表论文最多的大学，其次是马萨诸塞州总医院（49篇）、牛津大学（35篇）、华盛顿大学（29篇）和德克萨斯大学（26篇）。美国占所有产出的53.4%，其次是英国（21.6%）、德国（12%）、加拿大（9%）和法国（8%）。作者中，Stephen M. Smith贡献最多（19篇），其次是Jenkinson， M （n = 14）和Friston， KJ （n = 13）。最常见的关键词是“核磁共振”(n = 65),“大脑”(n = 43),“功能性磁共振成像”(n = 37), (n = 25)“分割”,和“宠物”(n = 24)。除了在引用影响（每篇文章的引用次数）上领先所有期刊外，Neuroimage还被认为是最具生产力的期刊。在平均引用影响方面，在各自类别中表现最好的实体是：牛津大学（在组织中）、德国（在国家中）、Smith Stephen M（在作者中）和Neuroimage杂志（在期刊中）。新兴词汇包括“深度学习”和“人工智能”。被引用最多的文章是Ronneberger等人的U-Net(2015)，被引用63448次。结论高影响放射学研究集中在北美和西欧，神经影像学和人工智能是重点新兴领域。这些见解为研究优先级和合作策略提供了路线图。
{"title":"Global trends and collaboration networks in radiology: A bibliometric analysis of the 500 most-cited articles in web of science","authors":"Habip Eser Akkaya , Erhan Kaya , Rasim Gökmen , Muhammed Semih Gedik , Dorukan İnanç Akpınar","doi":"10.1016/j.clinimag.2025.110700","DOIUrl":"10.1016/j.clinimag.2025.110700","url":null,"abstract":"<div><h3>Objective</h3><div>This study examined global research trends in Radiology, Nuclear Medicine, and Medical Imaging by analyzing the 500 most-cited articles in the Web of Science (WoS) Core Collection.</div></div><div><h3>Methods</h3><div>A bibliometric search was conducted on June 15, 2025. Biblioshiny and VOSviewer 1.6.20 were used for network visualization, including institutional collaboration, co-authorship, keyword co-occurrence, and country-level contributions. Temporal patterns were analyzed with Python 3.13.3, and descriptive statistics summarized publication data.</div></div><div><h3>Results</h3><div>Harvard University led institutional contributions with 54 publications, followed by Massachusetts General Hospital (<em>n</em> = 49), University of Oxford (<em>n</em> = 35), Washington University (<em>n</em> = 29), and University of Texas (<em>n</em> = 26). The United States accounted for 53.4 % of all outputs, followed by the United Kingdom (21.6 %), Germany (12 %), Canada (9 %), and France (8 %). Among authors, Stephen M. Smith contributed most (19 publications), followed by Jenkinson, M (<em>n</em> = 14), and Friston, KJ (<em>n</em> = 13). The most frequent keywords were “MRI” (<em>n</em> = 65), “Brain” (<em>n</em> = 43), “fMRI” (<em>n</em> = 37), “Segmentation” (<em>n</em> = 25), and “PET” (<em>n</em> = 24). In addition to leading all journals in citation impact (citations per article), Neuroimage was also identified as the most productive journal overall. Regarding the average citation impact, the top-performing entities in their respective categories were: the University of Oxford (among organizations), Germany (among countries), Smith Stephen M (among authors), and the journal Neuroimage (among journals). Emerging terms included “deep learning” and “artificial intelligence.” The most-cited article was Ronneberger et al.'s U-Net (2015), cited 63,448 times.</div></div><div><h3>Conclusion</h3><div>High-impact radiology research is concentrated in North America and Western Europe, with neuroimaging and artificial intelligence representing key emerging domains. These insights provide a roadmap for research prioritization and collaboration strategies.</div></div>","PeriodicalId":50680,"journal":{"name":"Clinical Imaging","volume":"130 ","pages":"Article 110700"},"PeriodicalIF":1.5,"publicationDate":"2025-12-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145840388","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-13DOI: 10.1016/j.clinimag.2025.110697
Hyungin Park , Ji-Yeon Han , Jinwook Baek
Purpose
To evaluate the measurement variability and reproducibility of prone imaging with quantitative computed tomography (QCT) compared with supine imaging with QCT in patients with interstitial lung disease (ILD).
Methods
Data from patients who underwent paired supine and prone computed tomography (CT) for ILD assessment between January 2021 and December 2023 were retrospectively analyzed. Lung abnormalities were quantified using deep learning-based software. Measurement variability and reproducibility were assessed using Bland–Altman analysis and intraclass correlation coefficients (ICC). Correlations of CT measurements with pulmonary function test (PFT) parameters were evaluated using Pearson's correlation and Steiger's Z-test. Linear and logistic regression analyses were conducted to identify factors associated with interpositional variability.
Results
In 277 patients, QCT measurements showed high reproducibility between supine and prone CT across ILD subtypes (ICC > 0.75), with variability of approximately 2 % and 4 % in fibrosis extent and total ILD extent, respectively. Ground-glass opacity (GGO) showed the highest variability, particularly in extensive disease. In the prone position, honeycombing demonstrated stronger correlation with PFT parameters (diffusing capacity of the lungs for carbon monoxide: r = −0.456 vs. −0.391). Total lung volume (β = −0.955), total ILD extent (β = 0.604), and GGO extent (β = 0.475) were the strongest predictors of inter-positional variability. The hypersensitivity pneumonitis pattern was independently associated with greater variability in the total ILD extent (odds ratio, 11.121).
Conclusion
Prone CT measurements demonstrated high reproducibility and comparable correlation with PFT parameters, relative to supine CT measurements, with fibrosis extent variability of approximately 2 %. Prone CT may be reliable for single-time-point assessment; however, caution is warranted for longitudinal comparisons in advanced or GGO-dominant ILD.
{"title":"Quantification differences between supine and prone CT in interstitial lung disease","authors":"Hyungin Park , Ji-Yeon Han , Jinwook Baek","doi":"10.1016/j.clinimag.2025.110697","DOIUrl":"10.1016/j.clinimag.2025.110697","url":null,"abstract":"<div><h3>Purpose</h3><div>To evaluate the measurement variability and reproducibility of prone imaging with quantitative computed tomography (QCT) compared with supine imaging with QCT in patients with interstitial lung disease (ILD).</div></div><div><h3>Methods</h3><div>Data from patients who underwent paired supine and prone computed tomography (CT) for ILD assessment between January 2021 and December 2023 were retrospectively analyzed. Lung abnormalities were quantified using deep learning-based software. Measurement variability and reproducibility were assessed using Bland–Altman analysis and intraclass correlation coefficients (ICC). Correlations of CT measurements with pulmonary function test (PFT) parameters were evaluated using Pearson's correlation and Steiger's <em>Z</em>-test. Linear and logistic regression analyses were conducted to identify factors associated with interpositional variability.</div></div><div><h3>Results</h3><div>In 277 patients, QCT measurements showed high reproducibility between supine and prone CT across ILD subtypes (ICC > 0.75), with variability of approximately 2 % and 4 % in fibrosis extent and total ILD extent, respectively. Ground-glass opacity (GGO) showed the highest variability, particularly in extensive disease. In the prone position, honeycombing demonstrated stronger correlation with PFT parameters (diffusing capacity of the lungs for carbon monoxide: <em>r</em> = −0.456 vs. −0.391). Total lung volume (β = −0.955), total ILD extent (β = 0.604), and GGO extent (β = 0.475) were the strongest predictors of inter-positional variability. The hypersensitivity pneumonitis pattern was independently associated with greater variability in the total ILD extent (odds ratio, 11.121).</div></div><div><h3>Conclusion</h3><div>Prone CT measurements demonstrated high reproducibility and comparable correlation with PFT parameters, relative to supine CT measurements, with fibrosis extent variability of approximately 2 %. Prone CT may be reliable for single-time-point assessment; however, caution is warranted for longitudinal comparisons in advanced or GGO-dominant ILD.</div></div>","PeriodicalId":50680,"journal":{"name":"Clinical Imaging","volume":"130 ","pages":"Article 110697"},"PeriodicalIF":1.5,"publicationDate":"2025-12-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145795673","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-09DOI: 10.1016/j.clinimag.2025.110696
Romi Noy Achiron
{"title":"Reply: Comment on “Limited performance of ChatGPT-4v and ChatGPT-4o in image-based core radiology cases”","authors":"Romi Noy Achiron","doi":"10.1016/j.clinimag.2025.110696","DOIUrl":"10.1016/j.clinimag.2025.110696","url":null,"abstract":"","PeriodicalId":50680,"journal":{"name":"Clinical Imaging","volume":"130 ","pages":"Article 110696"},"PeriodicalIF":1.5,"publicationDate":"2025-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145851347","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-07DOI: 10.1016/j.clinimag.2025.110694
Hao-Yu Huang , Yu-Han Huang , Cheng-Hsun Lin , Wan-Ting Tao , Wei-Chen Liao , Shuning Yu , Huei-Chi Mo , Wenyen Feng , Yu-Ting Hsu , Jian-Chiao Wang , Kai-Hsiung Ko
Purpose
To evaluate the impact of an artificial intelligence (AI)-assisted computer-aided detection (CAD) system on the diagnostic accuracy and confidence in chest radiograph interpretation among nonthoracic radiologists and radiology residents with varying levels of experience.
Methods
In this retrospective multiple-reader, multiple-case (MRMC) study, 400 chest radiographs (100 each for pulmonary nodules, pleural effusion, pneumothorax, and controls) were independently interpreted by 12 readers (two nonthoracic radiologists, four senior residents, and six junior residents). Readings were conducted under CAD-assisted and unassisted conditions, with a 30-day washout period. Readers assigned confidence scores (0–100) to their diagnosis. Diagnostic performance was evaluated using the area under the curve (AUC), sensitivity, and specificity, while reader confidence was assessed by the proportion of high-confidence ratings among correctly interpreted cases.
Results
The AI-assisted CAD system improved diagnostic performance across all abnormalities, with significant gains for pulmonary nodules (AUC: 0.781 → 0.854; P < 0.001) and pleural effusion (0.896 → 0.948; P < 0.001). The sensitivity increased by 7.2% for effusion, while the specificity for nodules improved markedly by 15.7%. Among all the readers, junior residents showed the greatest gains, especially for nodules, where the CAD closed their baseline AUC gap (originally −7.3%, P = 0.006) relative to nonthoracic radiologists. Reader confidence also increased significantly with the CAD, particularly for nodules (+15.2 %; P < 0.001).
Conclusion
The AI-assisted CAD system significantly enhanced diagnostic accuracy and reader confidence in chest radiograph interpretation, especially for junior radiology residents. This approach may bridge experience-related diagnostic gaps and support clinical decision-making, particularly in institutions lacking thoracic radiologists.
{"title":"AI-assisted chest radiograph interpretation enhances diagnostic confidence and standardizes diagnostic accuracy across radiologists: A multi-reader study","authors":"Hao-Yu Huang , Yu-Han Huang , Cheng-Hsun Lin , Wan-Ting Tao , Wei-Chen Liao , Shuning Yu , Huei-Chi Mo , Wenyen Feng , Yu-Ting Hsu , Jian-Chiao Wang , Kai-Hsiung Ko","doi":"10.1016/j.clinimag.2025.110694","DOIUrl":"10.1016/j.clinimag.2025.110694","url":null,"abstract":"<div><h3>Purpose</h3><div>To evaluate the impact of an artificial intelligence (AI)-assisted computer-aided detection (CAD) system on the diagnostic accuracy and confidence in chest radiograph interpretation among nonthoracic radiologists and radiology residents with varying levels of experience.</div></div><div><h3>Methods</h3><div>In this retrospective multiple-reader, multiple-case (MRMC) study, 400 chest radiographs (100 each for pulmonary nodules, pleural effusion, pneumothorax, and controls) were independently interpreted by 12 readers (two nonthoracic radiologists, four senior residents, and six junior residents). Readings were conducted under CAD-assisted and unassisted conditions, with a 30-day washout period. Readers assigned confidence scores (0–100) to their diagnosis. Diagnostic performance was evaluated using the area under the curve (AUC), sensitivity, and specificity, while reader confidence was assessed by the proportion of high-confidence ratings among correctly interpreted cases.</div></div><div><h3>Results</h3><div>The AI-assisted CAD system improved diagnostic performance across all abnormalities, with significant gains for pulmonary nodules (AUC: 0.781 → 0.854; P < 0.001) and pleural effusion (0.896 → 0.948; P < 0.001). The sensitivity increased by 7.2% for effusion, while the specificity for nodules improved markedly by 15.7%. Among all the readers, junior residents showed the greatest gains, especially for nodules, where the CAD closed their baseline AUC gap (originally −7.3%, P = 0.006) relative to nonthoracic radiologists. Reader confidence also increased significantly with the CAD, particularly for nodules (+15.2 %; P < 0.001).</div></div><div><h3>Conclusion</h3><div>The AI-assisted CAD system significantly enhanced diagnostic accuracy and reader confidence in chest radiograph interpretation, especially for junior radiology residents. This approach may bridge experience-related diagnostic gaps and support clinical decision-making, particularly in institutions lacking thoracic radiologists.</div></div>","PeriodicalId":50680,"journal":{"name":"Clinical Imaging","volume":"130 ","pages":"Article 110694"},"PeriodicalIF":1.5,"publicationDate":"2025-12-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145737227","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","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