Pub Date : 2026-01-01DOI: 10.1016/j.ejrad.2025.112632
Kecheng Yuan , Qingyun Liu , Weijie Zheng , Xiaoqiong Tang , Chuang Zhang , Jiantai Zhou , Penghui Luo , Fulang Qi , Lin Chen , Bensheng Qiu
Purpose
To investigate age-related compositional changes in the infrapatellar fat pad (IFP) using chemical shift-encoded MRI-derived proton density fat fraction (PDFF) and T2* at 1.5 T, and to evaluate their associations with clinical symptoms and structural abnormalities.
Methods
A cross-sectional study was performed in 100 adults (mean age: 44 ± 14 years; 54 men, 46 women) who underwent 1.5 T MRI using a six-echo spoiled gradient-echo sequence for quantitative assessment of PDFF and T2* in the IFP. Clinical symptoms were assessed using the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), and structural abnormalities were evaluated using the Whole-Organ Magnetic Resonance Imaging Score (WORMS). Spearman correlation and mediation analyses were performed.
Results
Age showed a strong negative correlation with IFP PDFF (r = − 0.46, p < 0.001) and a moderate inverse association with T2* (r = − 0.30, p = 0.003). Lower PDFF was associated with worse WOMAC pain (r = − 0.33, p = 0.001) and WORMS scores (r = − 0.41, p < 0.001). The mediation analysis revealed a significant indirect pathway linking age to both pain (ACME = 0.19, 95 % CI: 0.02–0.38) and WORMS scores (ACME = 0.24, 95 % CI: 0.06–0.45) through PDFF. In contrast, the indirect pathway through T2* was weaker and less precise (ACME = 0.12, 95 % CI: 0.01–0.28).
Conclusions
PDFF and T2* reflect age-related compositional and microstructural changes in the IFP, with PDFF demonstrating stronger associations with clinical symptoms and joint degeneration. These findings indicate PDFF as a superior imaging biomarker for early detection, monitoring, and potential therapeutic targeting in age-related musculoskeletal degeneration.
目的：利用化学位移编码mri衍生质子密度脂肪分数（PDFF）和T2*在1.5 T时研究年龄相关的髌下脂肪垫（IFP）组成变化，并评估其与临床症状和结构异常的关系。方法：对100名成年人（平均年龄：44±14岁；男性54名，女性46名）进行横断面研究，这些成年人接受1.5 T MRI检查，使用六回波破坏梯度回波序列定量评估IFP中的PDFF和T2*。临床症状采用西安大略和麦克马斯特大学骨关节炎指数（WOMAC）进行评估，结构异常采用全器官磁共振成像评分（WORMS）进行评估。进行Spearman相关分析和中介分析。结果：年龄与IFP PDFF呈显著负相关(r = - 0.46, p 2* (r = - 0.30, p = 0.003)。较低的PDFF与较差的WOMAC疼痛（r = - 0.33, p = 0.001）和WORMS评分(r = - 0.41， p 2*较弱且较不精确（ACME = 0.12, 95% CI: 0.01-0.28）相关。结论：PDFF和T2*反映了IFP中与年龄相关的组成和微观结构变化，其中PDFF与临床症状和关节退变有更强的相关性。这些发现表明PDFF是一种优越的成像生物标志物，可用于年龄相关性肌肉骨骼变性的早期检测、监测和潜在的治疗靶点。
{"title":"Evaluation of Age-Related compositional changes in the infrapatellar fat pad using MRI-Derived PDFF and T2* at 1.5 T","authors":"Kecheng Yuan , Qingyun Liu , Weijie Zheng , Xiaoqiong Tang , Chuang Zhang , Jiantai Zhou , Penghui Luo , Fulang Qi , Lin Chen , Bensheng Qiu","doi":"10.1016/j.ejrad.2025.112632","DOIUrl":"10.1016/j.ejrad.2025.112632","url":null,"abstract":"<div><h3>Purpose</h3><div>To investigate age-related compositional changes in the infrapatellar fat pad (IFP) using chemical shift-encoded MRI-derived proton density fat fraction (PDFF) and T<sub>2</sub>* at 1.5 T, and to evaluate their associations with clinical symptoms and structural abnormalities.</div></div><div><h3>Methods</h3><div>A cross-sectional study was performed in 100 adults (mean age: 44 ± 14 years; 54 men, 46 women) who underwent 1.5 T MRI using a six-echo spoiled gradient-echo sequence for quantitative assessment of PDFF and T<sub>2</sub>* in the IFP. Clinical symptoms were assessed using the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), and structural abnormalities were evaluated using the Whole-Organ Magnetic Resonance Imaging Score (WORMS). Spearman correlation and mediation analyses were performed.</div></div><div><h3>Results</h3><div>Age showed a strong negative correlation with IFP PDFF (r = − 0.46, p < 0.001) and a moderate inverse association with T<sub>2</sub>* (r = − 0.30, p = 0.003). Lower PDFF was associated with worse WOMAC pain (r = − 0.33, p = 0.001) and WORMS scores (r = − 0.41, p < 0.001). The mediation analysis revealed a significant indirect pathway linking age to both pain (ACME = 0.19, 95 % CI: 0.02–0.38) and WORMS scores (ACME = 0.24, 95 % CI: 0.06–0.45) through PDFF. In contrast, the indirect pathway through T<sub>2</sub>* was weaker and less precise (ACME = 0.12, 95 % CI: 0.01–0.28).</div></div><div><h3>Conclusions</h3><div>PDFF and T<sub>2</sub>* reflect age-related compositional and microstructural changes in the IFP, with PDFF demonstrating stronger associations with clinical symptoms and joint degeneration. These findings indicate PDFF as a superior imaging biomarker for early detection, monitoring, and potential therapeutic targeting in age-related musculoskeletal degeneration.</div></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":"196 ","pages":"Article 112632"},"PeriodicalIF":3.3,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145959128","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.ejrad.2025.112652
Sara Marziali , Andrea Cozzi , Veronica Magni , Laura Menicagli , Adrienn Benedek , Andrea Cisarri , Alessandra Marrocco , Giuseppe Di Giulio , Francesco Sardanelli
Objectives
To evaluate the performance of contrast-enhanced mammography (CEM) for the assessment of screening recalls at ≥ 3-year follow-up.
Materials and methods
Women recalled after screening mammography at two Italian centers were prospectively enrolled to undergo CEM alongside standard assessment (SA) through additional views, tomosynthesis, and/or ultrasound. SA and CEM were independently evaluated, recommending biopsy or referral to a subsequent screening round. Per-breast diagnostic performance of CEM (low-energy plus recombined images) was calculated by taking histopathology and/or ≥ 3-year follow-up as the reference standard.
Results
Of the original 414 breasts (from 207 women), 393 (from 198 women) had available biopsy or follow-up data for the purposes of this analysis, being negative either at the original biopsy or at follow-up (316/393, 80.4%), or having a malignant finding at the original biopsy (74/393, 18.8%), or an interval cancer during follow-up (3/393, 0.8%, two at 1-year, one at 3-year). These interval cancers were 1 DCIS, two node-negative invasive (one 5-mm mixed no special type/lobular; one 6-mm mucinous). Per-breast analysis of the 393 breasts yielded a 96.1% sensitivity (74/77, 95% confidence interval 89.3–98.8%), 94.9% specificity (300/316, 92.0–96.9%), 95.2% accuracy (374/393, 92.6–97.1%), 82.2% positive predictive value (74/90, 73.2–88.8%), 99.0% negative predictive value (300/303, 97.2–99.7%).
Conclusions
In a cohort of women undergoing screening recall with CEM, when considering ≥ 3-year follow-up, per-breast sensitivity was about 96 %, specificity 95 %, negative predictive value 99 %.
{"title":"CEM in women prospectively assessed for screening recalls: Per-breast diagnostic performance with 3-year or longer follow-up","authors":"Sara Marziali , Andrea Cozzi , Veronica Magni , Laura Menicagli , Adrienn Benedek , Andrea Cisarri , Alessandra Marrocco , Giuseppe Di Giulio , Francesco Sardanelli","doi":"10.1016/j.ejrad.2025.112652","DOIUrl":"10.1016/j.ejrad.2025.112652","url":null,"abstract":"<div><h3>Objectives</h3><div>To evaluate the performance of contrast-enhanced mammography (CEM) for the assessment of screening recalls at ≥ 3-year follow-up.</div></div><div><h3>Materials and methods</h3><div>Women recalled after screening mammography at two Italian centers were prospectively enrolled to undergo CEM alongside standard assessment (SA) through additional views, tomosynthesis, and/or ultrasound. SA and CEM were independently evaluated, recommending biopsy or referral to a subsequent screening round. Per-breast diagnostic performance of CEM (low-energy plus recombined images) was calculated by taking histopathology and/or ≥ 3-year follow-up as the reference standard.</div></div><div><h3>Results</h3><div>Of the original 414 breasts (from 207 women), 393 (from 198 women) had available biopsy or follow-up data for the purposes of this analysis, being negative either at the original biopsy or at follow-up (316/393, 80.4%), or having a malignant finding at the original biopsy (74/393, 18.8%), or an interval cancer during follow-up (3/393, 0.8%, two at 1-year, one at 3-year). These interval cancers were 1 DCIS, two node-negative invasive (one 5-mm mixed no special type/lobular; one 6-mm mucinous). Per-breast analysis of the 393 breasts yielded a 96.1% sensitivity (74/77, 95% confidence interval 89.3–98.8%)<strong>,</strong> 94.9% specificity (300/316, 92.0–96.9%)<strong>,</strong> 95.2% accuracy (374/393, 92.6–97.1%)<strong>,</strong> 82.2% positive predictive value (74/90, 73.2–88.8%)<strong>,</strong> 99.0% negative predictive value (300/303, 97.2–99.7%)<strong>.</strong></div></div><div><h3>Conclusions</h3><div>In a cohort of women undergoing screening recall with CEM, when considering ≥ 3-year follow-up, per-breast sensitivity was about 96 %, specificity 95 %, negative predictive value 99 %.</div></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":"195 ","pages":"Article 112652"},"PeriodicalIF":3.3,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145917375","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.ejrad.2025.112634
Yiqiao Yan , Xinyuan Zhang , Chuan Liu , Hang Fu , Ke Xu , Huayan Xu
Background
To investigate the diagnostic performance of CCTA-based artificial intelligence (AI) in detecting ≥ 50 % coronary stenosis of coronary artery disease (CAD) at both the patient and vessel levels.
Methods
A systematic search of PubMed, Embase, and Web of Science databases (from inception to March 2025) was conducted to identify diagnostic studies evaluating CCTA-based AI methods for detecting CAD. Studies of ≥ 50 % coronary stenosis were enrolled. Data for diagnostic performance was extracted and meta analysis was performed. Statistical analyses were performed using RevMan 5.4, Meta-Disc 1.4, and Stata 16.0. This study was designed and reported following the PRISMA-DTA statement.
Results
The systematic search yielded 2,211 potentially relevant records. Following multi-stage screening, 11 eligible studies (1506 patients; 2896 vessels) were included. At the patient level, AI based assistant tools demonstrated pooled sensitivity of 0.95 [95 % CI (0.93–0.97)], specificity of 0.73 [95 % CI (0.61–0.82)], and AUC of 0.96 [95 % CI (0.94–0.97)] for CAD with ≥ 50 % stenosis diagnosis. The positive likelihood ratio (+LR) was 3.5 [95 % CI (2.4–5.2)] and negative likelihood ratio (−LR) was 0.07 [95 % CI (0.05–0.10)], with a pooled DOR of 52 [31–86]. At the vessel level of diagnosing ≥ 50 % stenosis diagnosis, AI-based assistant diagnostic tool showed pooled sensitivity of 0.87 [95 % CI (0.83–0.90)], specificity of 0.89 [95 % CI (0.82–0.93)], +LR of 7.7 [95 % CI (4.8–12.5)], −LR of 0.15 [95 % CI (0.11–0.19)], DOR of 53 [35–79], and AUC of 0.93 [95 % CI (0.90–0.95)].
Conclusion
When pooled across diverse deep-learning systems, AI-assisted CCTA demonstrates high sensitivity and solid diagnostic performance for detecting ≥ 50 % coronary stenosis. However, this reflects aggregated results from heterogeneous models rather than the capability of any single AI tool, limiting direct generalizability to specific systems or vendors.
研究基于ccta的人工智能（AI）在患者和血管水平检测冠心病（CAD）冠状动脉狭窄≥50%的诊断性能。方法系统检索PubMed、Embase和Web of Science数据库（从创建到2025年3月），以确定评估基于ccta的人工智能方法检测CAD的诊断研究。纳入了冠状动脉狭窄≥50%的研究。提取诊断表现数据并进行meta分析。采用RevMan 5.4、Meta-Disc 1.4和Stata 16.0进行统计分析。本研究的设计和报告遵循PRISMA-DTA声明。结果系统检索得到2211条可能相关的记录。经过多阶段筛选，纳入了11项符合条件的研究（1506例患者，2896条血管）。在患者水平上，基于人工智能的辅助工具对狭窄程度≥50%的CAD诊断的总敏感性为0.95 [95% CI(0.93-0.97)]，特异性为0.73 [95% CI (0.61-0.82)]， AUC为0.96 [95% CI(0.94-0.97)]。阳性似然比（+LR）为3.5 [95% CI(2.4-5.2)]，阴性似然比（- LR）为0.07 [95% CI(0.05-0.10)]，合并DOR为52[31-86]。在诊断≥50%狭窄的血管水平上，人工智能辅助诊断工具的总灵敏度为0.87 [95% CI(0.83-0.90)]，特异性为0.89 [95% CI (0.82-0.93)]， +LR为7.7 [95% CI (4.8% - 12.5)]， - LR为0.15 [95% CI (0.11-0.19)]， DOR为53 [35-79]，AUC为0.93 [95% CI(0.90-0.95)]。结论：人工智能辅助CCTA在多种深度学习系统中对冠状动脉狭窄的检测具有较高的灵敏度和可靠的诊断性能。然而，这反映了来自异构模型的聚合结果，而不是任何单个AI工具的能力，限制了对特定系统或供应商的直接推广。
{"title":"Ccta-based AI for Diagnosing ≥ 50 % coronary Stenosis: A patient- and Vessel-Level meta-analysis","authors":"Yiqiao Yan , Xinyuan Zhang , Chuan Liu , Hang Fu , Ke Xu , Huayan Xu","doi":"10.1016/j.ejrad.2025.112634","DOIUrl":"10.1016/j.ejrad.2025.112634","url":null,"abstract":"<div><h3>Background</h3><div>To investigate the diagnostic performance of CCTA-based artificial intelligence (AI) in detecting ≥ 50 % coronary stenosis of coronary artery disease (CAD) at both the patient and vessel levels.</div></div><div><h3>Methods</h3><div>A systematic search of PubMed, Embase, and Web of Science databases (from inception to March 2025) was conducted to identify diagnostic studies evaluating CCTA-based AI methods for detecting CAD. Studies of ≥ 50 % coronary stenosis were enrolled. Data for diagnostic performance was extracted and meta analysis was performed. Statistical analyses were performed using RevMan 5.4, Meta-Disc 1.4, and Stata 16.0. This study was designed and reported following the PRISMA-DTA statement.</div></div><div><h3>Results</h3><div>The systematic search yielded 2,211 potentially relevant records. Following multi-stage screening, 11 eligible studies (1506 patients; 2896 vessels) were included. At the patient level, AI based assistant tools demonstrated pooled sensitivity of 0.95 [95 % CI (0.93–0.97)], specificity of 0.73 [95 % CI (0.61–0.82)], and AUC of 0.96 [95 % CI (0.94–0.97)] for CAD with ≥ 50 % stenosis diagnosis. The positive likelihood ratio (+LR) was 3.5 [95 % CI (2.4–5.2)] and negative likelihood ratio (−LR) was 0.07 [95 % CI (0.05–0.10)], with a pooled DOR of 52 [31–86]. At the vessel level of diagnosing ≥ 50 % stenosis diagnosis, AI-based assistant diagnostic tool showed pooled sensitivity of 0.87 [95 % CI (0.83–0.90)], specificity of 0.89 [95 % CI (0.82–0.93)], +LR of 7.7 [95 % CI (4.8–12.5)], −LR of 0.15 [95 % CI (0.11–0.19)], DOR of 53 [35–79], and AUC of 0.93 [95 % CI (0.90–0.95)].</div></div><div><h3>Conclusion</h3><div>When pooled across diverse deep-learning systems, AI-assisted CCTA demonstrates high sensitivity and solid diagnostic performance for detecting ≥ 50 % coronary stenosis. However, this reflects aggregated results from heterogeneous models rather than the capability of any single AI tool, limiting direct generalizability to specific systems or vendors.</div></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":"195 ","pages":"Article 112634"},"PeriodicalIF":3.3,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145921116","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.ejrad.2025.112635
Yongsheng Liu , Fengyi Zhang , Guiwen Shao, Yuxiang Liu, Yongjian Liu, Peng Ge, Feng Wang
Objectives
To evaluate the diagnostic performance of Angio-based fractional flow (Angio-FF), a computational fluid dynamics technique, for identifying hemodynamically significant intracranial atherosclerotic stenosis (ICAS) compared with the degree of stenosis (DS).
Materials and methods
This retrospective study included 348 patients with unilateral ICAS. Angio-FF was calculated using AccuFFicas V1.0 software. Diagnostic accuracy, sensitivity, specificity, and area under the ROC curve (AUC) were compared between Angio-FF and DS, stratified by stenosis severity, vascular territory, and Angio-FF gray zone.
Results
Symptomatic lesions exhibited lower Angio-FF (median 0.46 vs. 0.83, P < 0.001) and higher DS (69.5 % vs. 62.5 %, P < 0.001). Angio-FF showed superior AUC (0.884 vs. 0.664, P < 0.001) than DS. Angio-FF, using a cutoff value of 0.69, demonstrated significantly higher sensitivity (81.4 % vs. 50.0 %), specificity (88.9 % vs. 68.9 %), accuracy (83.3 % vs. 54.9 %), positive predictive value (PPV) (95.5 % vs. 82.2 %), and negative predictive value (NPV) (62.5 % vs. 32.5 %) compared to DS (≥ 70 % stenosis) for identifying culprit lesions (P < 0.001). Diagnostic performance was significantly better in anterior circulation lesions than in posterior ones, with higher sensitivity (91.8 % vs. 60.9 %), accuracy (89.5 % vs. 73.1 %), and NPV (73.1 % vs. 55.3 %) (all P < 0.05), while specificity and PPV remained high in both regions. Within the Angio-FF gray zone (0.64–0.74), diagnostic performance was lower than outside the zone, with reduced sensitivity (67.4 % vs. 84.4 %), accuracy (72.4 % vs. 85.5 %), and NPV (42.3 % vs. 67.6 %) (all P < 0.05), yet it still outperformed DS, with a higher AUC (0.795 vs. 0.524, P = 0.003).
Conclusions
Angio-FF outperformed conventional DS assessment in identifying hemodynamically significant ICAS, with particularly high accuracy in lesions with < 70 % stenosis. Despite reduced performance within the gray zone, it remained more reliable than DS.
Abbreviations: ICAS, intracranial atherosclerotic stenosis; TIA, transient ischemic attack; DS, degree of stenosis; FF, fractional flow; Angio-FF, Angio-based fractional flow; CFD, computational fluid dynamics; DSA, digital subtraction angiography; AUC, area under the curve; ROC, receiver operating characteristic; PPV, positive predictive value; NPV, negative predictive value; DCA, decision curve analysis; MCA, middle cerebral artery; BA, basilar artery; VA, vertebral artery; IQR, interquartile range; MRA, magnetic resonance angiography; Pa, proximal arterial pressure; Pd, distal arterial pressure; TIMI, Thrombolysis In Myocardial Infarction.
目的评价基于血管的分数血流（angiobased fractional flow，简称Angio-FF）作为一种计算流体动力学技术，对血流动力学意义显著的颅内动脉粥样硬化性狭窄（ICAS）和狭窄程度（DS）的诊断价值。材料与方法本研究纳入348例单侧ICAS患者。使用AccuFFicas V1.0软件计算血管内皮素水平。通过狭窄严重程度、血管范围和Angio-FF灰色区进行分层，比较了Angio-FF和DS的诊断准确性、敏感性、特异性和ROC曲线下面积（AUC）。结果有症状的病变表现为较低的Angio-FF（中位数0.46比0.83,P < 0.001）和较高的DS（中位数69.5%比62.5%,P < 0.001）。Angio-FF的AUC（0.884比0.664,P < 0.001）优于DS。与DS（狭窄度≥70%）相比，使用截断值0.69的Angio-FF在识别罪魁祸首病变方面表现出更高的敏感性（81.4%比50.0%）、特异性（88.9%比68.9%）、准确性（83.3%比54.9%）、阳性预测值（PPV）（95.5%比82.2%）和阴性预测值（NPV）（62.5%比32.5%）（P < 0.001）。前循环病变的诊断表现明显优于后循环病变，敏感性（91.8%比60.9%）、准确性（89.5%比73.1%）和NPV（73.1%比55.3%）均较高（P < 0.05），特异性和PPV在两个区域均保持较高水平。在Angio-FF灰色区域（0.64-0.74）内，诊断性能低于该区域外，敏感性（67.4% vs. 84.4%）、准确性（72.4% vs. 85.5%）和NPV (42.3% vs. 67.6%)（均P <； 0.05），但仍优于DS， AUC较高（0.795 vs. 0.524, P = 0.003）。结论sangio - ff在识别血流动力学意义显著的ICAS方面优于传统DS评估，在狭窄<； 70%的病变中准确率特别高。尽管在灰色区域内性能下降，但它仍然比DS更可靠。缩写：ICAS，颅内动脉粥样硬化性狭窄；TIA，短暂性脑缺血发作；DS：狭窄程度；FF，分流；Angio-FF，血管分流；计算流体力学；DSA，数字减影血管造影；AUC：曲线下面积；ROC，接收机工作特性；PPV，阳性预测值；NPV，负预测值；DCA，决策曲线分析；MCA，大脑中动脉；BA，基底动脉；VA，椎动脉；IQR，四分位间距；磁共振血管造影；Pa，近端动脉压；Pd，远端动脉压；TIMI，心肌梗死中的溶栓。
{"title":"Diagnostic performance of Angio-Based fractional flow for hemodynamic assessment in intracranial Atherosclerosis","authors":"Yongsheng Liu , Fengyi Zhang , Guiwen Shao, Yuxiang Liu, Yongjian Liu, Peng Ge, Feng Wang","doi":"10.1016/j.ejrad.2025.112635","DOIUrl":"10.1016/j.ejrad.2025.112635","url":null,"abstract":"<div><h3>Objectives</h3><div>To evaluate the diagnostic performance of Angio-based fractional flow (Angio-FF), a computational fluid dynamics technique, for identifying hemodynamically significant intracranial atherosclerotic stenosis (ICAS) compared with the degree of stenosis (DS).</div></div><div><h3>Materials and methods</h3><div>This retrospective study included 348 patients with unilateral ICAS. Angio-FF was calculated using AccuFFicas V1.0 software. Diagnostic accuracy, sensitivity, specificity, and area under the ROC curve (AUC) were compared between Angio-FF and DS, stratified by stenosis severity, vascular territory, and Angio-FF gray zone.</div></div><div><h3>Results</h3><div>Symptomatic lesions exhibited lower Angio-FF (median 0.46 vs. 0.83,<!--> <!-->P < 0.001) and higher DS (69.5 % vs. 62.5 %,<!--> <!-->P < 0.001). Angio-FF showed superior AUC (0.884 vs. 0.664,<!--> <!-->P < 0.001) than DS. Angio-FF, using a cutoff value of 0.69, demonstrated significantly higher sensitivity (81.4 % vs. 50.0 %), specificity (88.9 % vs. 68.9 %), accuracy (83.3 % vs. 54.9 %), positive predictive value (PPV) (95.5 % vs. 82.2 %), and negative predictive value (NPV) (62.5 % vs. 32.5 %) compared to DS (≥ 70 % stenosis) for identifying culprit lesions (P < 0.001). Diagnostic performance was significantly better in anterior circulation lesions than in posterior ones, with higher sensitivity (91.8 % vs. 60.9 %), accuracy (89.5 % vs. 73.1 %), and NPV (73.1 % vs. 55.3 %) (all P < 0.05), while specificity and PPV remained high in both regions. Within the Angio-FF gray zone (0.64–0.74), diagnostic performance was lower than outside the zone, with reduced sensitivity (67.4 % vs. 84.4 %), accuracy (72.4 % vs. 85.5 %), and NPV (42.3 % vs. 67.6 %) (all P < 0.05), yet it still outperformed DS, with a higher AUC (0.795 vs. 0.524, P = 0.003).</div></div><div><h3>Conclusions</h3><div>Angio-FF outperformed conventional DS assessment in identifying hemodynamically significant ICAS, with particularly high accuracy in lesions with < 70 % stenosis. Despite reduced performance within the gray zone, it remained more reliable than DS.</div><div>Abbreviations: ICAS, intracranial atherosclerotic stenosis; TIA, transient ischemic attack; DS, degree of stenosis; FF, fractional flow; Angio-FF, Angio-based fractional flow; CFD, computational fluid dynamics; DSA, digital subtraction angiography; AUC, area under the curve; ROC, receiver operating characteristic; PPV, positive predictive value; NPV, negative predictive value; DCA, decision curve analysis; MCA, middle cerebral artery; BA, basilar artery; VA, vertebral artery; IQR, interquartile range; MRA, magnetic resonance angiography; Pa, proximal arterial pressure; Pd, distal arterial pressure; TIMI, Thrombolysis In Myocardial Infarction.</div></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":"195 ","pages":"Article 112635"},"PeriodicalIF":3.3,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145880227","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-31DOI: 10.1016/j.ejrad.2025.112638
Domagoj Javor , Markus Leyer , Bryan K. Ward , Barbara Bennani-Baiti , Elisabeth Ranharter , Michael Bauer , Margit Kirschbaum , Markus Brunner , Bela Büki
It has been shown that static magnetic fields from high-strength magnetic resonance imaging (MRI) machines induce nystagmus in all humans with intact inner ear function. This effect can be explained by the magneto-hydrodynamic Lorentz force, which arises from the interaction of endolymphatic ionic currents and the strong static magnetic field of an MRI machine. Prior experiments demonstrated that MRI-induced nystagmus and vertigo vary with head pitch relative to the magnetic field, being reduced when the head is pitched forward and increased when extended. In another study it has been suggested that signal void artefacts reflected Lorentz-force-induced endolymph movement caused by the interaction between ionic currents flowing through the utricular macula and the static magnetic field of the MRI scanner. Based on these findings the present authors proposed that if the hypointensities are flow voids caused by Lorentz forces, their visibility should also vary with head pitch. In this case, both nystagmus and vestibular hypointensities would share a common mechanism. Twenty healthy volunteers (8 males and 12 females) were recruited to undergo a non-contrast 3 Tesla (T) MRI scan in one of two head pitch positions: chin up (head extension, pitched backward) and chin down (head flexion, pitched forward). A statistically significant increase in hypointensities was observed between the pitched forward and pitched backward positions for both ears (p < 0.01), while no significant differences were detected between corresponding positions of the left and right ears. These findings not only support a Lorentz‑force origin of vestibular hypointensities but also have immediate clinical applicability, with direct implications for radiological interpretation and protocol design to reduce misinterpretation and patient vertigo.
{"title":"Head position matters: Position‑dependent vestibular flow void artifacts in inner ear MRI and their clinical implications","authors":"Domagoj Javor , Markus Leyer , Bryan K. Ward , Barbara Bennani-Baiti , Elisabeth Ranharter , Michael Bauer , Margit Kirschbaum , Markus Brunner , Bela Büki","doi":"10.1016/j.ejrad.2025.112638","DOIUrl":"10.1016/j.ejrad.2025.112638","url":null,"abstract":"<div><div>It has been shown that static magnetic fields from high-strength magnetic resonance imaging (MRI) machines induce nystagmus in all humans with intact inner ear function. This effect can be explained by the magneto-hydrodynamic Lorentz force, which arises from the interaction of endolymphatic ionic currents and the strong static magnetic field of an MRI machine. Prior experiments demonstrated that MRI-induced nystagmus and vertigo vary with head pitch relative to the magnetic field, being reduced when the head is pitched forward and increased when extended. In another study it has been suggested that signal void artefacts reflected Lorentz-force-induced endolymph movement caused by the interaction between ionic currents flowing through the utricular macula and the static magnetic field of the MRI scanner. Based on these findings the present authors proposed that if the hypointensities are flow voids caused by Lorentz forces, their visibility should also vary with head pitch. In this case, both nystagmus and vestibular hypointensities would share a common mechanism. Twenty healthy volunteers (8 males and 12 females) were recruited to undergo a non-contrast 3 Tesla (T) MRI scan in one of two head pitch positions: chin up (head extension, pitched backward) and chin down (head flexion, pitched forward). A statistically significant increase in hypointensities was observed between the pitched forward and pitched backward positions for both ears (p < 0.01), while no significant differences were detected between corresponding positions of the left and right ears. These findings not only support a Lorentz‑force origin of vestibular hypointensities but also have immediate clinical applicability, with direct implications for radiological interpretation and protocol design to reduce misinterpretation and patient vertigo.</div></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":"195 ","pages":"Article 112638"},"PeriodicalIF":3.3,"publicationDate":"2025-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145921120","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-31DOI: 10.1016/j.ejrad.2025.112642
Christian Roller , Till Ittermann , Annika Syperek , Matthias Mühler , Mark O Wielpütz , Jens-Peter Kühn , Sophia FU Blum , Felix Schön , Patrick Winter , Susanne Schnell , Marie-Luise Kromrey
Purpose
This study aims to evaluate whether quantitative imaging features analyzed by an artificial intelligence (AI) tool are associated with success rate, histopathological results, and complication risks of CT-guided lung biopsies.
Methods
A retrospective study was conducted on 120 CT-guided biopsies of suspicious pulmonary lesions with pathology reports. Associations between technical success, histopathology, occurrence of peri-interventional complications, as well as intervention-related factors such as lesion diameter and biopsy pathway and the AI-derived parameters lesion volume, malignancy probability and emphysema ratio were assessed using t-test for continuous data and Chi-square test for categorical data. Adjusted multivariate logistic regression models and predictive performance of AI parameters were calculated.
Results
Ninety-eight of 120 biopsies (81.7 %) were technically successful. Peri-interventional pneumothorax occurred in 65 % of cases, 26.7 % needed drainage. Alveolar hemorrhage was documented in 53.3 %, high-grade hemorrhage in 30.8 %. Adjusted regression models showed significant association of AI-derived lesion volume with technical success (OR = 1.30, CI 1.00; 1.69), AI malignancy chance with histopathologically confirmed malignancy (OR = 1.17, CI 1.08; 1.28) and AI emphysema ratio with increased risk of pneumothorax requiring chest tube insertion (OR = 1.29, CI 1.12; 1.48). For alveolar hemorrhage, only AI lesion volume (p = 0.011) showed a significant inverse correlation in unadjusted models, while adjusted models identified emphysema ratio as the relevant AI feature.
Conclusions
AI derived imaging features show significant association with complication risks in CT-guided lung biopsies, like pneumothorax and alveolar hemorrhage. This may allow stricter patient selection and better execution of biopsies, which may lead to improved outcomes and patient safety. Additionally, software’s high association with histopathological malignancy supports reconsideration of its role in guiding the indication for lung biopsy in the assessment of pulmonary lesions.
Clinical Relevance Statement
With the advancing integration of AI tools into radiological workflows, this study highlights AI’s potential to provide pre-interventional risk stratification and outcome predictions for CT-guided lung biopsies, which are the gold standard for sampling peripheral lung lesions.
目的本研究旨在评估人工智能（AI）工具分析的定量影像学特征是否与ct引导下肺活检的成功率、组织病理学结果和并发症风险相关。方法对120例ct引导下的可疑肺病变活检病例进行回顾性分析。连续资料采用t检验，分类资料采用卡方检验，评估技术成功、组织病理学、介入周并发症发生、病变直径、活检路径等干预相关因素与人工智能衍生参数病变体积、恶性肿瘤概率、肺气肿率之间的关系。计算调整后的多元逻辑回归模型和人工智能参数的预测性能。结果120例活检中98例（81.7%）技术成功。65%的病例发生介入期气胸，26.7%需要引流。53.3%的患者有肺泡出血，30.8%的患者有重度出血。调整后的回归模型显示，人工智能病变体积与技术成功相关（OR = 1.30, CI 1.00; 1.69），人工智能恶性肿瘤发生率与组织病理学证实的恶性肿瘤相关（OR = 1.17, CI 1.08; 1.28），人工智能肺气肿发生率与需要插入胸管的气胸风险增加相关（OR = 1.29, CI 1.12; 1.48）。对于肺泡出血，在未调整的模型中，只有AI病变体积（p = 0.011）与肺气肿呈显著负相关，而调整后的模型将肺气肿比例确定为相关的AI特征。结论ct引导下肺活检的影像学特征与气胸、肺泡出血等并发症风险有显著相关性。这可能允许更严格的患者选择和更好的活检执行，这可能导致改善的结果和患者的安全。此外，软件与组织病理学恶性肿瘤的高度关联支持重新考虑其在肺病变评估中指导肺活检指征的作用。随着人工智能工具与放射工作流程的不断整合，本研究强调了人工智能在为ct引导的肺活检提供介入前风险分层和结果预测方面的潜力，而ct引导的肺活检是周围肺病变采样的金标准。
{"title":"Hitting the Bull’s AI: Artificial Intelligence-derived Imaging Features and their Association with Outcomes in CT-guided Lung Biopsy, a Retrospective Study","authors":"Christian Roller , Till Ittermann , Annika Syperek , Matthias Mühler , Mark O Wielpütz , Jens-Peter Kühn , Sophia FU Blum , Felix Schön , Patrick Winter , Susanne Schnell , Marie-Luise Kromrey","doi":"10.1016/j.ejrad.2025.112642","DOIUrl":"10.1016/j.ejrad.2025.112642","url":null,"abstract":"<div><h3>Purpose</h3><div>This study aims to evaluate whether quantitative imaging features analyzed by an artificial intelligence (AI) tool are associated with success rate, histopathological results, and complication risks of CT-guided lung biopsies.</div></div><div><h3>Methods</h3><div>A retrospective study was conducted on 120 CT-guided biopsies of suspicious pulmonary lesions with pathology reports. Associations between technical success, histopathology, occurrence of peri-interventional complications, as well as intervention-related factors such as lesion diameter and biopsy pathway and the AI-derived parameters lesion volume, malignancy probability and emphysema ratio were assessed using t-test for continuous data and Chi-square test for categorical data. Adjusted multivariate logistic regression models and predictive performance of AI parameters were calculated.</div></div><div><h3>Results</h3><div>Ninety-eight of 120 biopsies (81.7 %) were technically successful. Peri-interventional pneumothorax occurred in 65 % of cases, 26.7 % needed drainage. Alveolar hemorrhage was documented in 53.3 %, high-grade hemorrhage in 30.8 %. Adjusted regression models showed significant association of AI-derived lesion volume with technical success (OR = 1.30, CI 1.00; 1.69), AI malignancy chance with histopathologically confirmed malignancy (OR = 1.17, CI 1.08; 1.28) and AI emphysema ratio with increased risk of pneumothorax requiring chest tube insertion (OR = 1.29, CI 1.12; 1.48). For alveolar hemorrhage, only AI lesion volume (p = 0.011) showed a significant inverse correlation in unadjusted models, while adjusted models identified emphysema ratio as the relevant AI feature.</div></div><div><h3>Conclusions</h3><div>AI derived imaging features show significant association with complication risks in CT-guided lung biopsies, like pneumothorax and alveolar hemorrhage. This may allow stricter patient selection and better execution of biopsies, which may lead to improved outcomes and patient safety. Additionally, software’s high association with histopathological malignancy supports reconsideration of its role in guiding the indication for lung biopsy in the assessment of pulmonary lesions.</div></div><div><h3>Clinical Relevance Statement</h3><div>With the advancing integration of AI tools into radiological workflows, this study highlights AI’s potential to provide pre-interventional risk stratification and outcome predictions for CT-guided lung biopsies, which are the gold standard for sampling peripheral lung lesions.</div></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":"195 ","pages":"Article 112642"},"PeriodicalIF":3.3,"publicationDate":"2025-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145921187","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-31DOI: 10.1016/j.ejrad.2025.112640
Liming Li , Yang Yao , Kehui Nie , Mengchen Yuan , Songmei Fan , Jianbo Gao
Purpose
To determine which keV PureCa images can effectively substitute TNC images for abdominal imaging.
Methods
A total of 50 consecutive patients who underwent thoracoabdominal scans on Photon-Counting CT from January to April 2025 were included. A true non-contrast (TNC), arterial (CTA) and venous (CTV) phase images were analyzed. Conventional virtual non-contrast (VNC) and 5 PureCa images were reconstructed from each contrast phase. Attenuation values and noise on 12 regions of interest were measured. Overall image quality, contrast removal and calcium preservation were analyzed using a 5-point scale. The paired t-tests and Friedman tests were applied to compare objective and subjective results.
Results
70 keV yielded the smallest HUerror in the majority of abdominal tissues during the arterial and venous phase. The overall trend demonstrated a plateau in noise reduction beyond 65 keV. Compared with VNC images, PureCa demonstrated significantly lower HUerror in key soft tissues: Liver: 2 (1–6) HU vs 5 (3–8) HU (P = 0.001), Muscle: 2 (1–4) HU vs 10 (7–14) HU (P < 0.001), subcutaneous adipose tissue: 2.5 (1–4) HU vs 39 (33–42) HU (P < 0.001). All patients had a subjective score of 3 or above. 65–70 keV PureCa images were rated significantly higher than 50–60 keV images, and comparable to VNC, but still inferior to TNC (P < 0.05).
Conclusions
PureCa images reconstructed at 70 keV on photon-counting CT systems provide accurate attenuation values and satisfactory image quality for abdominal tissues.
目的：确定哪些keV PureCa图像可以有效地替代TNC图像进行腹部成像。方法：选取自2025年1月至4月连续50例行光子计数CT胸腹扫描的患者。对真实无对比期（TNC）、动脉期（CTA）和静脉期（CTV）图像进行分析。从每个对比阶段重建常规虚拟非对比（VNC）和5张PureCa图像。测量了12个感兴趣区域的衰减值和噪声。整体图像质量、对比度去除和钙保存采用5分制进行分析。采用配对t检验和Friedman检验比较客观和主观结果。结果：在动脉和静脉期，70 keV在大多数腹部组织中产生最小的HUerror。总体趋势表明，在65 keV以上，降噪趋于平稳。与VNC图像相比，PureCa显示关键软组织的HUerror明显降低：肝脏：2 (1-6)HU vs 5 (3-8) HU (P = 0.001)，肌肉：2 (1-4)HU vs 10 (7-14) HU (P < 0.001)，皮下脂肪组织：2.5 (1-4)HU vs 39 (33-42) HU （P < 0.001）。所有患者主观评分均在3分及以上。65-70 keV的PureCa图像评分明显高于50-60 keV的图像，与VNC相当，但仍低于TNC (P)结论：在光子计数CT系统上重建的PureCa图像在70 keV下可以提供准确的衰减值和令人满意的图像质量。
{"title":"Evaluation of multi-keV PureCalcium images from photon-counting CT for abdominal imaging: A comparison with true and conventional virtual non-contrast images","authors":"Liming Li , Yang Yao , Kehui Nie , Mengchen Yuan , Songmei Fan , Jianbo Gao","doi":"10.1016/j.ejrad.2025.112640","DOIUrl":"10.1016/j.ejrad.2025.112640","url":null,"abstract":"<div><h3>Purpose</h3><div>To determine which keV PureCa images can effectively substitute TNC images for abdominal imaging.</div></div><div><h3>Methods</h3><div>A total of 50 consecutive patients who underwent thoracoabdominal scans on Photon-Counting CT from January to April 2025 were included. A true non-contrast (TNC), arterial (CTA) and venous (CTV) phase images were analyzed. Conventional virtual non-contrast (VNC) and 5 PureCa images were reconstructed from each contrast phase. Attenuation values and noise on 12 regions of interest were measured. Overall image quality, contrast removal and calcium preservation were analyzed using a 5-point scale. The paired t-tests and Friedman tests were applied to compare objective and subjective results.</div></div><div><h3>Results</h3><div>70 keV yielded the smallest HUerror in the majority of abdominal tissues during the arterial and venous phase. The overall trend demonstrated a plateau in noise reduction beyond 65 keV. Compared with VNC images, PureCa demonstrated significantly lower HUerror in key soft tissues: Liver: 2 (1–6) HU vs 5 (3–8) HU (P = 0.001), Muscle: 2 (1–4) HU vs 10 (7–14) HU (P < 0.001), subcutaneous adipose tissue: 2.5 (1–4) HU vs 39 (33–42) HU (P < 0.001). All patients had a subjective score of 3 or above. 65–70 keV PureCa images were rated significantly higher than 50–60 keV images, and comparable to VNC, but still inferior to TNC (P < 0.05).</div></div><div><h3>Conclusions</h3><div>PureCa images reconstructed at 70 keV on photon-counting CT systems provide accurate attenuation values and satisfactory image quality for abdominal tissues.</div></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":"195 ","pages":"Article 112640"},"PeriodicalIF":3.3,"publicationDate":"2025-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145910798","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-31DOI: 10.1016/j.ejrad.2025.112639
Ziman Chen , Mengting Ye , Jieyi Liang , Fei Chen , Michael Tin Cheung Ying
Purpose
Recent advances in multimodal large language models (LLMs) have demonstrated promising potential for medical image analysis, yet their diagnostic capability in thyroid ultrasound remains unverified. This study explored the feasibility of ChatGPT-5, the latest multimodal LLM, for thyroid nodule classification and contextualized its diagnostic performance against S-Detect, an FDA-approved commercial computer-aided diagnosis system.
Methods
In this prospective study, 141 patients with 186 nodules who underwent preoperative ultrasound and subsequent surgery were enrolled. For S-Detect, the largest transverse grayscale ultrasound image of each nodule was analyzed with automated contouring for binary classification. For ChatGPT-5, cropped transverse and longitudinal nodule ultrasound images were analyzed using a standardized diagnostic prompt for binary classification. Agreement with histopathology was assessed using Kappa statistics; sensitivity, specificity, accuracy, and area under the receiver operating characteristic curve (AUC) were calculated.
Results
Both systems showed statistically significant ability to distinguish benign from malignant nodules (P < 0.05). Agreement with histopathology was fair for ChatGPT-5 (Kappa = 0.224) and moderate for S-Detect (Kappa = 0.579). ChatGPT-5 demonstrated sensitivity 50.8 %, specificity 75.8 %, and accuracy 59.1 %, whereas S-Detect achieved higher sensitivity (91.9 %) and accuracy (82.3 %) but lower specificity (62.9 %). The AUC for S-Detect (77.4 %) was significantly greater than that for ChatGPT-5 (63.3 %, P < 0.001).
Conclusions
ChatGPT-5 demonstrated feasibility for thyroid nodule classification but showed lower diagnostic performance than the licensed, pre-trained S-Detect system and is not yet adequate for medical imaging applications.
{"title":"ChatGPT-5–Based large language model analysis versus an FDA-approved AI-CAD system for thyroid nodule ultrasound evaluation","authors":"Ziman Chen , Mengting Ye , Jieyi Liang , Fei Chen , Michael Tin Cheung Ying","doi":"10.1016/j.ejrad.2025.112639","DOIUrl":"10.1016/j.ejrad.2025.112639","url":null,"abstract":"<div><h3>Purpose</h3><div>Recent advances in multimodal large language models (LLMs) have demonstrated promising potential for medical image analysis, yet their diagnostic capability in thyroid ultrasound remains unverified. This study explored the feasibility of ChatGPT-5, the latest multimodal LLM, for thyroid nodule classification and contextualized its diagnostic performance against S-Detect, an FDA-approved commercial computer-aided diagnosis system.</div></div><div><h3>Methods</h3><div>In this prospective study, 141 patients with 186 nodules who underwent preoperative ultrasound and subsequent surgery were enrolled. For S-Detect, the largest transverse grayscale ultrasound image of each nodule was analyzed with automated contouring for binary classification. For ChatGPT-5, cropped transverse and longitudinal nodule ultrasound images were analyzed using a standardized diagnostic prompt for binary classification. Agreement with histopathology was assessed using <em>Kappa</em> statistics; sensitivity, specificity, accuracy, and area under the receiver operating characteristic curve (AUC) were calculated.</div></div><div><h3>Results</h3><div>Both systems showed statistically significant ability to distinguish benign from malignant nodules (<em>P</em> < 0.05). Agreement with histopathology was fair for ChatGPT-5 (<em>Kappa</em> = 0.224) and moderate for S-Detect (<em>Kappa</em> = 0.579). ChatGPT-5 demonstrated sensitivity 50.8 %, specificity 75.8 %, and accuracy 59.1 %, whereas S-Detect achieved higher sensitivity (91.9 %) and accuracy (82.3 %) but lower specificity (62.9 %). The AUC for S-Detect (77.4 %) was significantly greater than that for ChatGPT-5 (63.3 %, <em>P</em> < 0.001).</div></div><div><h3>Conclusions</h3><div>ChatGPT-5 demonstrated feasibility for thyroid nodule classification but showed lower diagnostic performance than the licensed, pre-trained S-Detect system and is not yet adequate for medical imaging applications.</div></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":"195 ","pages":"Article 112639"},"PeriodicalIF":3.3,"publicationDate":"2025-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145910782","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-31DOI: 10.1016/j.ejrad.2025.112643
Mika Donabauer , Stefan Sawall , Isabelle Ayx , Carmen Wängler , Stefan O. Schoenberg , Björn Wängler
Purpose
The development of novel contrast agents is a critical step in unleashing the full potential of Photon-Counting Computed Tomography (PCCT), as clinically approved iodine contrast agents do not exhibit optimal spectral properties to harness the full potential of photon-counting detectors. Moreover, material decomposition requires the application of different contrast agents with differing X-ray attenuation characteristics at different photon energies. However, the development of contrast agents being suitable for this purpose is in its infancy, limiting the imaging options of PCCT. In this study, we investigated which of the elements lanthanum, gadolinium, ytterbium, lead, bismuth and gold are suitable for use as the basis of contrast agents in PCCT as these should exhibit highly interesting X-ray attenuation characteristics at different tube voltages for this application.
Methods
New contrast agents based on the aforementioned elements were synthesized and characterized in terms of stability. Further, these substances were characterized by phantom studies on the first FDA-approved clinical PCCT Naeotom Alpha scanner with regard to contrast generation at different tube voltages of 70, 90, 120, and 140 kV in comparison to the approved iodine-based contrast agent Gastrolux to determine their general potential for PCCT imaging.
Results
We found that at lower tube voltages of 70 kV, iodine (48.1 ± 0.8 HU/(mg/mL)) and gadolinium (43.5 ± 1.0 HU/(mg/mL)) demonstrated the highest attenuation rates, whereas lanthanum, gold, ytterbium, and lead showed lower X-ray attenuation efficiency under these conditions (35.7 ± 0.7, 28.1 ± 0.6, 25.8 ± 0.2, and 25.3 ± 0.1 HU/(mg/mL), respectively). At a higher tube voltage of 120 kV, the situation changed, showing the highest attenuation rates for gadolinium, ytterbium, and iodine (33.4 ± 0.4, 28.1 ± 0.1, and 24.6 ± 0.3 HU/(mg/mL), respectively), followed by gold, bismuth, lanthanum, and lead (23.5 ± 0.3, 21.7 ± 0.2, 19.7 ± 0.3, and 18.1 ± 0.1 HU/(mg/mL), respectively).
Conclusions
Although all elements investigated showed a general suitability for implementation in contrast-enhanced CT imaging, iodine, gadolinium and ytterbium demonstrated the highest potential in this regard. However, also gold and bismuth could in general be suitable for material decomposition studies in combination with the aforementioned elements in cases necessitating a strongly varying contrasting behavior at chosen tube voltages.
{"title":"Development and characterization of new contrast agents for Photon-Counting CT","authors":"Mika Donabauer , Stefan Sawall , Isabelle Ayx , Carmen Wängler , Stefan O. Schoenberg , Björn Wängler","doi":"10.1016/j.ejrad.2025.112643","DOIUrl":"10.1016/j.ejrad.2025.112643","url":null,"abstract":"<div><h3>Purpose</h3><div>The development of novel contrast agents is a critical step in unleashing the full potential of Photon-Counting Computed Tomography (PCCT), as clinically approved iodine contrast agents do not exhibit optimal spectral properties to harness the full potential of photon-counting detectors. Moreover, material decomposition requires the application of different contrast agents with differing X-ray attenuation characteristics at different photon energies. However, the development of contrast agents being suitable for this purpose is in its infancy, limiting the imaging options of PCCT. In this study, we investigated which of the elements lanthanum, gadolinium, ytterbium, lead, bismuth and gold are suitable for use as the basis of contrast agents in PCCT as these should exhibit highly interesting X-ray attenuation characteristics at different tube voltages for this application.</div></div><div><h3>Methods</h3><div>New contrast agents based on the aforementioned elements were synthesized and characterized in terms of stability. Further, these substances were characterized by phantom studies on the first FDA-approved clinical PCCT Naeotom Alpha scanner with regard to contrast generation at different tube voltages of 70, 90, 120, and 140 kV in comparison to the approved iodine-based contrast agent Gastrolux to determine their general potential for PCCT imaging.</div></div><div><h3>Results</h3><div>We found that at lower tube voltages of 70 kV, iodine (48.1 ± 0.8 HU/(mg/mL)) and gadolinium (43.5 ± 1.0 HU/(mg/mL)) demonstrated the highest attenuation rates, whereas lanthanum, gold, ytterbium, and lead showed lower X-ray attenuation efficiency under these conditions (35.7 ± 0.7, 28.1 ± 0.6, 25.8 ± 0.2, and 25.3 ± 0.1 HU/(mg/mL), respectively). At a higher tube voltage of 120 kV, the situation changed, showing the highest attenuation rates for gadolinium, ytterbium, and iodine (33.4 ± 0.4, 28.1 ± 0.1, and 24.6 ± 0.3 HU/(mg/mL), respectively), followed by gold, bismuth, lanthanum, and lead (23.5 ± 0.3, 21.7 ± 0.2, 19.7 ± 0.3, and 18.1 ± 0.1 HU/(mg/mL), respectively).</div></div><div><h3>Conclusions</h3><div>Although all elements investigated showed a general suitability for implementation in contrast-enhanced CT imaging, iodine, gadolinium and ytterbium demonstrated the highest potential in this regard. However, also gold and bismuth could in general be suitable for material decomposition studies in combination with the aforementioned elements in cases necessitating a strongly varying contrasting behavior at chosen tube voltages.</div></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":"195 ","pages":"Article 112643"},"PeriodicalIF":3.3,"publicationDate":"2025-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145879948","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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