Academic Radiology最新文献

{"title":"Vision Transformer for Pretreatment Prediction of Pathologic Complete Response in Multi-center Breast DCE-MRI.","authors":"Naomi Fridman, Bubby Solway, Tomer Fridman, Itamar Barnea, Anat Goldstein","doi":"10.1016/j.acra.2026.09.024","DOIUrl":"https://doi.org/10.1016/j.acra.2026.09.024","url":null,"abstract":"<p><strong>Rationale and objectives: </strong>Pretreatment prediction of pathologic complete response (pCR) is clinically important for guiding neoadjuvant chemotherapy (NAC), but robust multi-center benchmarks with standardized evaluation protocols remain limited. We evaluated a Vision Transformer (ViT) approach for pCR prediction from breast dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI).</p><p><strong>Materials and methods: </strong>We used BreastDCEDL, a public multi-center dataset comprising 2070 patients from the I-SPY1, I-SPY2, and Duke cohorts with predefined splits. Pre-contrast, early post-contrast, and late post-contrast phases were mapped to RGB channels for transfer learning from ImageNet-pretrained backbones. The held-out test set (n=175) included 32 pCR+ in I-SPY2 (99), 8 in Duke (41), and 12 in I-SPY1 (35). For I-SPY2, ViT scores were combined with clinical variables (age, HR/HER2 status, tumor volume, genomic score, treatment arm) via a Random Forest classifier.</p><p><strong>Results: </strong>The ViT achieved AUC 0.72 (95% CI 0.64-0.80), sensitivity 0.27 (95% CI 0.15-0.40), and specificity 0.96 (95% CI 0.92-0.99). Performance varied across cohorts (AUC: I-SPY2 0.78, 95% CI 0.68-0.86, I-SPY1 0.68, 95% CI 0.46-0.89, Duke 0.54, 95% CI 0.27-0.80). In I-SPY2, clinical variables increased AUC to 0.85 overall (n=99), a difference that was not statistically significant (DeLong P=0.06).</p><p><strong>Conclusion: </strong>Cross-cohort variation highlights challenges in developing generalizable pretreatment response models under real-world heterogeneity. The high specificity supports potential utility for identifying non-responders, enabling earlier transition to alternative strategies. Code, data, and predefined splits are publicly available, establishing a reproducible benchmark.</p>","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":" ","pages":""},"PeriodicalIF":4.7,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149118146","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Model-based Pressure-threshold Flow-rate Estimation for Contrast Injection via Upper-arm Central Venous Ports.","authors":"Yusuke Sato, Yasuhiro Fukushima, Soma Kumasaka, Tomokazu Takeuchi, Daisuke Ozaki, Yoshiyuki Sakai, Takayuki Suto, Yoshito Tsushima","doi":"10.1016/j.acra.2026.09.033","DOIUrl":"https://doi.org/10.1016/j.acra.2026.09.033","url":null,"abstract":"<p><strong>Rationale and objectives: </strong>To characterize the association between recorded maximum delivered injection rate and peak pressure during contrast-enhanced computed tomography (CT) through upper-arm central venous (CV) ports and estimate port- and viscosity-specific flow-rate thresholds corresponding to 13 kg/cm².</p><p><strong>Materials and methods: </strong>This retrospective study included 205 CT examinations from 102 patients with 103 implanted upper-arm CV ports. Three port systems and eight iodinated contrast media (ICM) formulations grouped by viscosity were evaluated. A no-intercept linear model estimated separate slopes and 13 kg/cm² thresholds for nine port-viscosity combinations using the recorded maximum delivered rate with port-clustered robust inference. Sensitivity analyses excluded pressure-limit activation and flow-reduction cases and examined the 13 kg/cm² preset subgroup.</p><p><strong>Results: </strong>All subgroup pressure-flow slopes were statistically significant (all p < 0.001), with uncentered R² values ranging from 0.962 to 0.993. Pressure-limit activation and flow reduction occurred in 38 (18.5%) and 26 (12.7%) examinations, respectively. Pressure-flow slopes differed by port and viscosity (both p < 0.001); the port-by-viscosity interaction was not significant in the primary analysis (p = 0.336). Residual formulation-level heterogeneity was detected within viscosity groups (p < 0.001). Thresholds ranged from 2.7 to >5.0 mL/s and changed by at most 5.6% in exclusion analyses. Similar patterns persisted in the 13 kg/cm² preset subgroup despite sparse cells.</p><p><strong>Conclusion: </strong>Under the evaluated clinical conditions, pressure-flow relationships through upper-arm CV ports differed by port system and ICM viscosity group. Model-based descriptive thresholds provide port- and viscosity-specific pressure-flow reference information for the evaluated combinations.</p>","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":" ","pages":""},"PeriodicalIF":4.7,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149118054","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Updated Pediatric Sonographic Percentile Curves for Abdominal Organ Size in a Large European Cohort.","authors":"Elena Höhne, Sebastian Fischer, Jennifer Gotta, Philipp Reschke, Vitali Koch, Scherwin Mahmoudi, Ibrahim Yel, Katrin Eichler, Jan-Henning Klusmann, Thomas Vogl, Leon Grünewald","doi":"10.1016/j.acra.2026.09.013","DOIUrl":"https://doi.org/10.1016/j.acra.2026.09.013","url":null,"abstract":"<p><strong>Purpose: </strong>Accurate measurement of liver, spleen, and kidney size plays a central role in pediatric diagnostics. Contemporary sonographic percentile data are needed to support interpretation of organ dimensions in clinical practice. This study aimed to generate age-specific clinical reference percentiles for abdominal organ size based on a large real-world pediatric ultrasound cohort.</p><p><strong>Methods and materials: </strong>This retrospective analysis included 27,696 clinically indicated pediatric abdominal ultrasound examinations conducted between 2000 and 2025 at a major urban tertiary-care center in Germany. Standardized organ measurements were obtained by experienced pediatric radiologists. Patients were not excluded on the basis of underlying pathology. Age-specific clinical percentile curves (P3-P97) were generated using Generalized Additive Models for Location, Scale, and Shape (GAMLSS) to capture non-linear growth patterns.</p><p><strong>Results: </strong>From 27,696 examinations of children aged 0-18, organ-specific measurements comprised 12,254 livers, 12,703 spleens, 16,254 right kidneys, and 16,516 left kidneys. All organs demonstrated non-linear growth trajectories. No overall statistically significant sex differences were observed in this cohort for spleen (p = 0.195), liver (p = 0.517), right kidney (p = 0.712), or left kidney (p = 0.797). Owing to the retrospective design, sex-specific variation within individual age strata cannot be excluded. The left kidney was consistently larger than the right (p<0.0001), with a mean difference of 2-5 mm.</p><p><strong>Conclusion: </strong>This study presents contemporary clinical percentile curves for liver, spleen, and kidney dimensions derived from one of the largest Central European pediatric ultrasound datasets published to date. The curves provide population-specific benchmarks for interpreting pediatric abdominal organ dimensions in real-world clinical practice and aim to support standardized assessment across the full pediatric age range.</p>","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":" ","pages":""},"PeriodicalIF":4.7,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149118127","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Intratumoral and Peritumoral Ultrasoundomics Models for Preoperative Prediction of Pathological Complete Response to Neoadjuvant Chemotherapy in Breast Cancer.","authors":"Pu Zhou, Li Yang, Ying He, Libao Wang, Rong Zhao, Mei Gu, Pengfei Zhu, Xu Luo, Xiangyun Kong, Pei Du","doi":"10.1016/j.acra.2026.09.038","DOIUrl":"https://doi.org/10.1016/j.acra.2026.09.038","url":null,"abstract":"<p><strong>Rationale and objectives: </strong>This study investigated the clinical utility of a machine learning (ML) model based on ultrasoundomics features (UFs) from primary foci to predict pathological complete response (pCR) in patients with breast cancer (BC) undergoing neoadjuvant chemotherapy (NAC).</p><p><strong>Materials and methods: </strong>Ultrasound images from 333 patients with BC treated with NAC were retrospectively analyzed. Patients in the internal training cohort and external validation cohort were from two different hospitals, respectively. Tumor regions of interest (ROI) were delineated, and outward segmentation of the peritumor area was performed, using the ROI edge as a baseline and expanding it by 3 mm and 5 mm, respectively. This resulted in three distinct segmentation images: ROI (intratumoral region), ROI (intratumoral region + 3-mm peritumoral region), and ROI (intratumoral region + 5-mm peritumoral region). Three sets of UFs were extracted, normalized using the Z-score method, and further refined through Spearman correlation analysis and LASSO for feature selection. Eight ML models (R0), based on key UFs in ROI (intratumoral region), were compared: support vector machine, K-nearest neighbor, random forest, Naïve Bayes, logistic regression, AdaBoost (AB), multilayer perceptron, and light gradient boosting machine learning. The optimal ML models were evaluated using diagnostic performance metrics, including AUC, accuracy, sensitivity, and specificity. The best-performing algorithm was then applied to construct ML models (R3 and R5) incorporating ROI (intratumoral region + 3-mm peritumoral region) and ROI (intratumoral region + 5-mm peritumoral region). Predictive efficacy and clinical utility were assessed through decision curve analysis (DCA) and Shapley Additive exPlanation (SHAP), which quantified feature importance and facilitated model visualization and interpretation.</p><p><strong>Results: </strong>Among the 333 patients, 119 (35.7%) achieved pCR, while 214 (64.3%) did not. Among the eight ML algorithms, the R0 model based on AB demonstrated the best performance, with an AUC of 0.811 (Internal training cohort) and 0.632 (External validation cohort). The R3 model also showed improved performance, with an AUC of 0.830 (Internal training cohort) and 0.699 (External validation cohort). The R5 model yielded the highest predictive accuracy, with an AUC of 0.871 (Internal training cohort) and 0.779 (External validation cohort). SHAP analysis was conducted for R5, and SHAP waterfall charts enable single-sample predictions to validate the model's clinical utility. DCA revealed that the combination of intratumoral region and 5-mm peritumoral region in R5 offered superior clinical benefit.</p><p><strong>Conclusion: </strong>ML models integrating intratumoral and peritumoral ultrasoundomics offer valuable assistance in assessing NAC efficacy in patients with BC.</p>","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":" ","pages":""},"PeriodicalIF":4.7,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149117903","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Corrigendum to \"Reliability and Educational Quality of Online Video-based CBCT Interpretation Training: Implications for Radiology Education\" (Academic Radiology, Volume 33, Issue 7 [2026] 2790-2795).","authors":"Murat Mert Atapek, Sedef Ayşe Taşyapan","doi":"10.1016/j.acra.2026.09.043","DOIUrl":"https://doi.org/10.1016/j.acra.2026.09.043","url":null,"abstract":"","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":" ","pages":""},"PeriodicalIF":4.7,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149117637","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Ultrasound-Based Habitat Radiomics for Differentiating Phyllodes Tumor from Breast Fibroadenoma.","authors":"Qinwen Xue, Wenzhen Ding, Zhonglin Lu, Fang Chen, XiaoHong Jia, Jie Liu, Jiahong Sun, Shuao Li, Jianqiao Zhou, Cheng Zhao, Jie Yu, Ping Liang","doi":"10.1016/j.acra.2026.09.010","DOIUrl":"https://doi.org/10.1016/j.acra.2026.09.010","url":null,"abstract":"<p><strong>Rationale and objectives: </strong>Phyllodes tumors (PTs) and fibroadenomas (FAs) have overlapping ultrasound (US) appearances. This study aimed to develop and evaluate an interpretable machine-learning model integrating US radiomics, habitat features, and clinical variables.</p><p><strong>Materials and methods: </strong>This retrospective multicenter study included 1950 pathologically confirmed lesions (1029 FAs and 921 PTs) divided into training (n=918), validation (n=474), and test (n=558) cohorts. Nineteen voxelwise features were used for K-means habitat segmentation. The number of clusters was selected from K=2-10 in the training cohort and evaluated across 50 random seeds. Whole-lesion, habitat, and combined habitat-radiomics (HR) signatures were developed using a train-only feature-selection pipeline, and ExtraTrees was selected as the classifier. The final combined model was built by probability-level late fusion of the ExtraTrees HR model and a logistic clinical model based on age, lesion size, and orientation.</p><p><strong>Results: </strong>K=3 yielded the highest Calinski-Harabasz index (200,927.01). Clustering was stable across 50 random seeds, with mean adjusted Rand indices of 0.9942, 0.9948, and 0.9941 in the training, validation, and test cohorts, respectively. Feature selection retained 39 whole-lesion features and 34 habitat features (habitats 1-3: 1, 14, and 19). The combined model achieved AUCs of 0.921 (95% CI, 0.904-0.939), 0.797 (95% CI, 0.754-0.835), and 0.844 (95% CI, 0.813-0.877) in the training out-of-fold, validation, and test cohorts.</p><p><strong>Conclusion: </strong>In this multicenter cohort, a model integrating US habitat radiomics with age, lesion size, and orientation differentiated PTs from FAs with consistent discrimination across data partitions. Shapley additive explanations analysis provided global and case-level explanations of the combined prediction.</p>","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":" ","pages":""},"PeriodicalIF":4.7,"publicationDate":"2026-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149118155","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Assessment of Pulmonary Vascular Stent Morphology Using Intra-procedural Cone-Beam CT in Fibrosing Mediastinitis: Imaging Agreement and Clinical Implications.","authors":"Jielong Lin, Jiawei Long, Xinye Xia, Zhenyi Li, Wenliang Guo, Xiaofeng Wu, Cheng Hong","doi":"10.1016/j.acra.2026.09.004","DOIUrl":"https://doi.org/10.1016/j.acra.2026.09.004","url":null,"abstract":"<p><strong>Rationale and objectives: </strong>Percutaneous Pulmonary Vascular Stent Angioplasty (PPVSA) is a key treatment for pulmonary vascular stenosis in Fibrosing Mediastinitis (FM). This study evaluated the added value of intraoperative cone-beam computed tomography (CBCT) beyond conventional angiography for post-stenting assessment during PPVSA.</p><p><strong>Materials and methods: </strong>This retrospective single-center study included 54 consecutive FM patients who underwent PPVSA between August 2021 and May 2024, comprising 114 stents (99 arterial and 15 venous). All patients underwent intraoperative CBCT, angiography, and follow-up computed tomography pulmonary angiography (CTPA) for assessment of stent geometry. Agreement between imaging modalities was evaluated using correlation, Bland-Altman, and reproducibility analyses. Associations between CBCT-derived parameters, hemodynamic outcomes, and stent patency were further explored.</p><p><strong>Results: </strong>Among 54 patients (57.26 ± 11.47 years; 17 females [31.48%]), CBCT demonstrated excellent correlation with first follow-up CTPA. For maximum stent diameter, the correlation between CBCT and CTPA was r = 0.99, compared with r = 0.94 for angiography. For minimum stent diameter, the corresponding correlations were r = 0.99 and r = 0.85, respectively (all p < 0.001). Bland-Altman analysis showed smaller measurement bias and narrower limits of agreement for CBCT than for angiography. During a median imaging follow-up of 614.50 days, in-stent restenosis occurred in 15.79% of stents, whereas complete stent occlusion occurred in 1.75% of stents. Stents with a minimum-to-maximum diameter ratio <80% showed higher restenosis rates than those with a ratio ≥80% (24.00% vs. 9.38%, p = 0.017). In addition, among pulmonary arterial stents, a CBCT-derived stent relief index was positively correlated with trans-lesional systolic pressure gradient reduction (r = 0.249, p = 0.033).</p><p><strong>Conclusion: </strong>Intraoperative CBCT provides reliable assessment of stent geometry during PPVSA. In addition to demonstrating strong agreement with first follow-up CTPA, CBCT-derived morphologic parameters may be associated with procedural hemodynamic improvement and may provide supplemental information regarding subsequent stent patency and long-term outcomes.</p>","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":" ","pages":""},"PeriodicalIF":4.7,"publicationDate":"2026-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149118285","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A National County-Level Typology of Imaging Infrastructure and the Radiologist Workforce in the United States.","authors":"Adyasha Pradhan, Sharon D'Souza, Andrew B Rosenkrantz","doi":"10.1016/j.acra.2026.09.030","DOIUrl":"https://doi.org/10.1016/j.acra.2026.09.030","url":null,"abstract":"<p><strong>Rationale and objectives: </strong>This study aimed to perform a national county-level analysis integrating the local presence of both imaging equipment and radiologists into a unified typology of imaging access and to evaluate its association with imaging utilization.</p><p><strong>Materials and methods: </strong>Radiologists were identified in the June 2026 National Plan and Provider Enumeration System using six verified taxonomy codes and assigned to counties by practice ZIP code. Imaging infrastructure combined hospital presence, Medicare computed tomography or magnetic resonance imaging billing, and certified mammography facilities. Counties were classified into four groups based on local radiologist and imaging equipment presence: true imaging deserts (neither), equipment-only, radiologist-only, and fully served (both). Sensitivity definitions used diagnostic-only taxonomies, Medicare billing, and billing-volume thresholds. Medicare imaging utilization was compared across categories. Multivariable logistic regression modeled radiologist absence and true imaging desert status.</p><p><strong>Results: </strong>A total of 1692/3144 (53.8%) counties had no documented local radiologist. A total of 568 counties were true imaging deserts (1.9% of residents), 1124 were equipment-only (7.6%), 19 were radiologist-only (0.1%), and 1433 were fully served (90.4%). Imaging utilization was lower in true imaging deserts than fully served counties (3989 vs. 4242 per 1000, P < 0.001); this difference held across varying equipment definitions. The no-radiologist share ranged 53.8%-63.3% across three workforce definitions. At the commuting-zone level, the prevalence of no-radiologist counties decreased to 16.8% and of true imaging desert counties to 1.7%. Smaller population and lower education predicted both county-level outcomes.</p><p><strong>Conclusion: </strong>Local imaging access stratifies into four county types with measurable utilization differences. Equipment-only counties and true imaging deserts may warrant different policy responses.</p>","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":" ","pages":""},"PeriodicalIF":4.7,"publicationDate":"2026-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149118294","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"CT Radiomics of Epicardial Adipose Tissue for Predicting Atrial Fibrillation: A Systematic Review.","authors":"Yevgeniy Gladkiy, Amin Hasheminia, Adham Elsherbini, Mutaz Kassas, Daniel Boyes, Daniel Faddoul, Robert Gilkeson, Mohammad El Diasty","doi":"10.1016/j.acra.2026.08.071","DOIUrl":"https://doi.org/10.1016/j.acra.2026.08.071","url":null,"abstract":"<p><strong>Rationale and objectives: </strong>Atrial fibrillation (AF) is the most common arrhythmia, affecting 1-2% of the general population. Epicardial adipose tissue (EAT) may contribute to AF through its proximity to the myocardium and coronary vessels, as well as its vasocrine and paracrine functions. Previous studies linked EAT thickness and volume with AF incidence. This review consolidates evidence on the correlation between radiomic analysis of EAT and the incidence of AF.</p><p><strong>Material and methods: </strong>This systematic review was conducted in accordance with PRISMA 2020 and registered in the International Prospective Register of Systematic Reviews (PROSPERO). Embase, PubMed, MEDLINE, Web of Science, and Google Scholar were searched for primary studies using CT-based EAT radiomics to predict AF. Case reports, animal studies, pediatric populations, and Magnetic Resonance Imaging (MRI)/ultrasound studies were excluded. Risk of bias was assessed with PROBAST, and methodological quality with the Radiomics Quality Score.</p><p><strong>Results: </strong>Ten studies, including 2952 patients (53.8-81.4% male; mean age range 42.2-74.1 years), met eligibility. Six were retrospective, and four were prospective. The studies evaluated postoperative AF, AF recurrence after catheter ablation, AF detection, and AF subtype classification. Reported machine learning models included random forest, logistic regression, Cox regression, and Least Absolute Shrinkage and Selection Operator (LASSO). Performance metrics ranged from AUC 0.73-0.92, sensitivity 0.50-0.90 and specificity 0.53-0.98. Segmentation methods and feature extraction varied, and external validation was rare.</p><p><strong>Conclusion: </strong>Machine learning models using EAT radiomic features show promise for AF prediction. However, this heterogeneity limits generalizability and precludes immediate incorporation into clinical practice.</p>","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":" ","pages":""},"PeriodicalIF":4.7,"publicationDate":"2026-10-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149117704","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Prespecified Structured CT and MRI Scores for Sinonasal Inverted Papilloma: A Dual-reader Diagnostic Accuracy Study.","authors":"Buket Yağcı, Tansu Pınarbaşılı, Erdinç Hakan İnan, Rezarta Taga Senirli, Döndü Nergiz, Özer Erdem Gür","doi":"10.1016/j.acra.2026.09.037","DOIUrl":"https://doi.org/10.1016/j.acra.2026.09.037","url":null,"abstract":"<p><strong>Rationale and objectives: </strong>To compare structured computed tomography (CT) and magnetic resonance imaging (MRI) scores for differentiating inverted papilloma (IP) from other benign and malignant causes of unilateral sinonasal opacification.</p><p><strong>Materials and methods: </strong>In this retrospective single-center study, 179 patients with histopathologically confirmed unilateral sinonasal opacification were classified as IP spectrum (n = 72) or non-IP (n = 107). Two blinded readers independently assessed CT (n = 172) and MRI (n = 109; both modalities, n = 102) using prespecified IP scores; Reader 1's final assessment served as the primary analysis. Diagnostic performance was compared using area under the curve (AUC) analysis with paired DeLong testing, and independent predictors were identified with multivariable, Firth-penalized, and parsimonious logistic regression.</p><p><strong>Results: </strong>In the paired cohort (n = 102), MRI demonstrated a numerically higher, but not statistically superior, performance than CT for distinguishing IP-spectrum from non-IP lesions (AUC 0.915 vs. 0.869; ΔAUC +0.045; p = 0.076). Combining CT and MRI scores did not improve discrimination beyond MRI alone (combined AUC 0.891; cross-validated AUC 0.883). Convoluted cerebriform pattern and bone destruction were the only independent predictors. MRI demonstrated a significant advantage over CT in differentiating IP from unilateral sinonasal polyps (AUC 0.965 vs. 0.811; q < 0.001), particularly in lesions lacking focal hyperostosis.</p><p><strong>Conclusion: </strong>CT and MRI provide complementary rather than interchangeable information in the evaluation of unilateral sinonasal opacification. Although overall diagnostic performance was comparable and the combined assessment offered no incremental benefit over MRI alone, MRI significantly outperformed CT in distinguishing IP from unilateral polyps. These findings support a structured multiparametric imaging approach, particularly in diagnostically challenging cases, and warrant prospective multicenter validation.</p>","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":" ","pages":""},"PeriodicalIF":4.7,"publicationDate":"2026-10-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149118113","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}