Academic Radiology最新文献

{"title":"Prediction of Early Recurrence After Bronchial Arterial Chemoembolization in Non-small Cell Lung Cancer Patients Using Dual-energy CT: An Interpretable Model Based on SHAP Methodology","authors":"Ye Feng ,&nbsp;Yinuo Xu ,&nbsp;Jian Wang ,&nbsp;Zhenyu Cao ,&nbsp;Bojun Liu ,&nbsp;Zeliu Du ,&nbsp;Lingling Zhou ,&nbsp;Haokai Hua ,&nbsp;Wenjie Wang ,&nbsp;Jie Mei ,&nbsp;Linqiang Lai ,&nbsp;Jianfei Tu","doi":"10.1016/j.acra.2025.07.039","DOIUrl":"10.1016/j.acra.2025.07.039","url":null,"abstract":"<div><h3>Rationale and Objectives</h3><div>Bronchial artery chemoembolization (BACE) is a new treatment method for lung cancer. This study aimed to investigate the ability of dual-energy computed tomography (DECT) to predict early recurrence (ER) after BACE among patients with non-small cell lung cancer (NSCLC) who failed first-line therapy.</div></div><div><h3>Materials and Methods</h3><div>Clinical and imaging data from NSCLC patients undergoing BACE at Wenzhou Medical University Affiliated Fifth *** Hospital (10/2023–06/2024) were retrospectively analyzed. Logistic regression (LR) machine learning models were developed using 5 arterial-phase (AP) virtual monoenergetic images (VMIs; 40, 70, 100, 120, and 150 keV), while deep learning models utilized ResNet50/101/152 architectures with iodine maps. A combined model integrating optimal Rad-score, DL-score, and clinical features was established. Model performance was assessed via area under the receiver operating characteristic curve analysis (AUC), with SHapley Additive exPlanations (SHAP) framework applied for interpretability.</div></div><div><h3>Results</h3><div>A total of 196 patients were enrolled in this study (training cohort: <em>n<!--> </em>=<!--> <!-->158; testing cohort: <em>n<!--> </em>=<!--> <!-->38). The 100 keV machine learning model demonstrated superior performance (AUC<!--> <!-->=<!--> <!-->0.751) compared to other VMIs. The deep learning model based on the ResNet101 method (AUC<!--> <!-->=<!--> <!-->0.791) performed better than other approaches. The hybrid model combining Rad-score-100keV-A, Rad-score-100keV-V, DL-score-ResNet101-A, DL-score-ResNet101-V, and clinical features exhibited the best performance (AUC<!--> <!-->=<!--> <!-->0.798) among all models.</div></div><div><h3>Conclusion</h3><div>DECT holds promise for predicting ER after BACE among NSCLC patients who have failed first-line therapy, offering valuable guidance for clinical treatment planning.</div></div>","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":"32 10","pages":"Pages 6320-6329"},"PeriodicalIF":3.9,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144818083","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 Morally Distressed Radiologist: In Her Own Words","authors":"Richard B. Gunderman MD, PhD","doi":"10.1016/j.acra.2025.06.015","DOIUrl":"10.1016/j.acra.2025.06.015","url":null,"abstract":"","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":"32 10","pages":"Pages 5679-5680"},"PeriodicalIF":3.9,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144568111","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 Deep Learning Model for Preoperative Prediction of Lymph Node Metastasis in cT1-Stage Lung Adenocarcinoma: A Multicenter External Validation Study","authors":"Qinqin Yan ,&nbsp;Ying Wei ,&nbsp;Zenghui Cheng ,&nbsp;Zhong Xue ,&nbsp;Feng Shi ,&nbsp;Shan Yang ,&nbsp;Zhiyong Zhang ,&nbsp;Fuhua Yan ,&nbsp;Fei Shan","doi":"10.1016/j.acra.2025.06.001","DOIUrl":"10.1016/j.acra.2025.06.001","url":null,"abstract":"<div><h3>Purpose</h3><div><span>To develop and validate a deep learning (DL) model for preoperative prediction of lymph node metastasis (LNM) in clinical T1-stage </span>lung adenocarcinoma<span> (LUAD), and to compare its performance with conventional semantic and radiomics signatures.</span></div></div><div><h3>Methods</h3><div>This multicenter retrospective study enrolled 2503 patients with 2568 pathologically confirmed cT1-stage LUAD nodules from eight institutions. Data from six centers (1994 patients/2059 nodules) were randomly divided into training (1600 patients/1664 nodules) and internal test (394 patients/395 nodules) cohorts. Two independent external validation cohorts (Set-1: 283 patients/nodules; Set-2: 226 patients/nodules) were included. Three predictive models were developed as follows: 1) a semantic model incorporating spiculation, pleural traction, air bronchogram, and vacuole<span> signs; 2) a radiomics<span> model; and 3) ResLNM—a residual network-based DL model. Diagnostic performance was evaluated using the area under the receiver operating characteristic curve (AUC) and decision curve analysis (DCA).</span></span></div></div><div><h3>Results</h3><div>Both ResLNM and radiomics models significantly outperformed the semantic model in predicting LNM (AUC range: 0.71–0.85 and 0.72–0.84 vs. 0.58–0.74, respectively; <em>P</em>&lt;0.05). While ResLNM demonstrated comparable performance to the radiomics model in the internal test set (AUC: 0.85, 0.81–0.89 vs. 0.84, 0.80–0.88; <em>P<!--> </em>=<!--> <!-->0.624) and external validation set-2 (0.71, 0.63–0.79 vs. 0.72, 0.64–0.80; <em>P<!--> </em>=<!--> <!-->0.472), it achieved superior accuracy in external validation set-1 (0.82, 0.76–0.88 vs. 0.77, 0.71–0.83; <em>P<!--> </em>=<!--> <!-->0.039). DCA confirmed the clinical superiority of ResLNM. Notably, integrating ResLNM with either semantic or radiomics signatures provided no incremental value (<em>P</em><span>&gt;0.05), whereas incorporating radiologically mediastinal enlarged lymph node status (short-axis diameter ≥10</span> <!-->mm) significantly enhanced predictive performance, achieving AUCs of 0.88 (95%CI:0.85–0.92), 0.89 (95%CI:0.84–0.94), and 0.76 (95%CI:0.68–0.83) in the test set and two validation cohorts, respectively.</div></div><div><h3>Conclusion</h3><div>The ResLNM model provides a clinically feasible tool for preoperative LNM prediction in cT1-stage LUAD, outperforming conventional semantic and radiomics approaches. Its performance can be further optimized by integrating routinely available lymph node size criteria, offering potential to refine surgical decision-making and reduce overtreatment.</div></div>","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":"32 10","pages":"Pages 6272-6283"},"PeriodicalIF":3.9,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144621104","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":"Association Between MRI Findings of Lumbar Morphometric Changes and the Characteristics of Low Back Pain, Pain-Related Disability, and Quality of Life: A Cross-sectional Study","authors":"Daishui Yang MD ,&nbsp;Luis Becker MD ,&nbsp;Bernhard U. Hoehl MD ,&nbsp;Lukas Mödl MSc ,&nbsp;Tianwei Zhang MD ,&nbsp;Sihai Liu MD ,&nbsp;Torsten Diekhoff MD ,&nbsp;Sandra Reitmaier Med.vet ,&nbsp;Matthias Pumberger MD ,&nbsp;Hendrik Schmidt PhD","doi":"10.1016/j.acra.2025.06.026","DOIUrl":"10.1016/j.acra.2025.06.026","url":null,"abstract":"<div><h3>Rationale and Objectives</h3><div>This research explored the association between common lumbar morphometric changes—including intervertebral disc degeneration (IDD), intervertebral disc herniation (IDH), high-intensity zone (HIZ), facet joint degeneration (FJD), and Modic changes (MCs)—and their combined effect on the characteristics of low back pain (LBP), disability and quality of life.</div></div><div><h3>Methods</h3><div>712 participants were included in this study, including 254 no back pain (no-BP), 159 intermittent LBP (iLBP), and 299 chronic LBP (cLBP). All recruited participants underwent questionnaire completion, clinical examination, and MRI scanning. Binary logistic regression and linear models were conducted to assess the relationship between lumbar morphometric changes and the characteristics of LBP, disability caused by LBP and quality of life.</div></div><div><h3>Results</h3><div>Participants with single MRI abnormalities, including IDD, IDH, and MCs, were found to be associated with greater odds of cLBP and iLBP. The greater number of structural changes observed in MRI findings was associated with greater odds of cLBP and iLBP. In addition, participants with IDD and HIZ were found to experience a longer episode of LBP (β 2.5, 95% CI 0.7–4.3, p<!--> <!-->=<!--> <!-->0.008 and β 3.5, 95% CI 0.8–6.1, p<!--> <!-->=<!--> <!-->0.010, respectively). MCs were the only MRI abnormalities associated with maximum intensity of LBP (β 0.5, 95% CI 0.1–1.0, p<!--> <!-->=<!--> <!-->0.026). FJD was found to be associated with LBP onset patterns occurring under stress (OR 1.9, 95% CI 1.1–3.2, p<!--> <!-->=<!--> <!-->0.014). There was an association between the number of MRI abnormalities and the maximum intensity of LBP (p for trend<!--> <!-->=<!--> <!-->0.001). Furthermore, MCs were also observed associated with greater disability and physical health (β 1.2, 95% CI 0.2–2.1, p<!--> <!-->=<!--> <!-->0.009 and β −7.1, 95% CI −11 to −3.6, p&lt;0.001). Participants with greater number of morphometric changes shown increased disability (p for trend<!--> <!-->=<!--> <!-->0.005) and reduced quality of life (p for trend &lt;0.001).</div></div><div><h3>Conclusion</h3><div>Our study shown an association between single morphometric changes and the characteristics of LBP, functional impairment and quality of life. These associations were more pronounced with the greater number of morphometric changes. Further work should aim to clarify the underlying causal mechanisms and assess the effectiveness of MRI findings as indicators for diagnosing LBP.</div></div>","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":"32 10","pages":"Pages 6000-6008"},"PeriodicalIF":3.9,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144610257","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":"Multi-modal Risk Stratification in Heart Failure with Preserved Ejection Fraction Using Clinical and CMR-derived Features: An Approach Incorporating Model Explainability","authors":"Shiwen Zhang ,&nbsp;Yuan Lin ,&nbsp;Dongqi Han ,&nbsp;Yifeng Pan ,&nbsp;Tianyu Geng ,&nbsp;Haitao Ge ,&nbsp;Jie Zhao","doi":"10.1016/j.acra.2025.06.048","DOIUrl":"10.1016/j.acra.2025.06.048","url":null,"abstract":"<div><h3>Rationale and Objectives</h3><div>Heart failure with preserved ejection fraction (HFpEF) poses significant diagnostic and prognostic challenges due to its clinical heterogeneity. This study proposes a multi-modal, explainable machine learning framework that integrates clinical variables and cardiac magnetic resonance (CMR)-derived features, particularly epicardial adipose tissue (EAT) volume, to improve risk stratification and outcome prediction in patients with HFpEF.</div></div><div><h3>Materials and Methods</h3><div>A retrospective cohort of 301 participants (171 in the HFpEF group and 130 in the control group) was analyzed. Baseline characteristics, CMR-derived EAT volume, and laboratory biomarkers were integrated into machine learning models. Model performance was evaluated using accuracy, precision, recall, and F1-score. Additionally, receiver operating characteristic area under the curve (ROC-AUC) and precision-recall area under the curve (PR-AUC) were employed to assess discriminative power across varying decision thresholds. Hyperparameter optimization and ensemble techniques were applied to enhance predictive performance.</div></div><div><h3>Results</h3><div>HFpEF patients exhibited significantly higher EAT volume (70.9±27.3 vs. 41.9±18.3 mL, p&lt;0.001) and NT-proBNP levels (1574 [963,2722] vs. 33 [10,100] pg/mL, p&lt;0.001), along with a greater prevalence of comorbidities. The voting classifier demonstrated the highest accuracy for HFpEF diagnosis (0.94), with a precision of 0.96, recall of 0.94, and an F1-score of 0.95. For prognostic tasks, AdaBoost, XGBoost and Random Forest yielded superior performance in predicting adverse clinical outcomes, including rehospitalization and all-cause mortality (accuracy: 0.95). Key predictive features identified included EAT volume, right atrioventricular groove (Right AVG), tricuspid regurgitation velocity (TRV), and metabolic syndrome.</div></div><div><h3>Conclusion</h3><div>Explainable models combining clinical and CMR-derived features, especially EAT volume, improve support for HFpEF diagnosis and outcome prediction. These findings highlight the value of a data-driven, interpretable approach to characterizing HFpEF phenotypes and may facilitate individualized risk assessment in selected populations.</div></div>","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":"32 10","pages":"Pages 5743-5753"},"PeriodicalIF":3.9,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144668920","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":"Comparing Habitat, Radiomics, and Fusion Models for Predicting Micropapillary/Solid Components in Stage I Lung Adenocarcinoma","authors":"Shaoyu Huang ,&nbsp;Xiuzhen Liang ,&nbsp;Kaihua Lou ,&nbsp;Jinlong Zhou ,&nbsp;Jie Wang ,&nbsp;Guodong Xu ,&nbsp;Shibo Wu ,&nbsp;Hongjie Hu ,&nbsp;Haibo Dong","doi":"10.1016/j.acra.2025.07.035","DOIUrl":"10.1016/j.acra.2025.07.035","url":null,"abstract":"<div><h3>Rationale and Objectives</h3><div>To comprehensively compare habitat, radiomics, and fusion models for the preoperative prediction of micropapillary/solid (MP/S) status in stage I lung adenocarcinoma (LAC).</div></div><div><h3>Materials and Methods</h3><div>In this retrospective study, we enrolled 345 patients postoperatively diagnosed with stage I LAC from two medical centers, dividing them into training (<em>n<!--> </em>=<!--> <!-->207), internal validation (<em>n<!--> </em>=<!--> <!-->69), and external validation (<em>n<!--> </em>=<!--> <!-->69) cohorts. Radiomics model (RM) was developed using CT images of the primary tumor. Habitat model (HM) was built by analyzing intra-tumor subregions identified via unsupervised K-means clustering algorithm. Fusion model employed two integration strategies as follows: feature-based pre-fusion model (pre-FM) and decision-based post-fusion model (post-FM). The predictive performance of all models was comprehensively evaluated by area under the curve (AUC) and integrated discrimination improvement (IDI). Additionally, correlations between clustering and radiomics features were analyzed with Spearman’s correlation analysis.</div></div><div><h3>Results</h3><div>The HM demonstrated superior predictive performance compared to the RM in the training cohort (AUC: 0.900 vs. 0.876, <em>p<!--> </em>=<!--> <!-->0.252). The pre-FM consistently outperformed the HM and RM across all study cohorts (AUC: 0.843–0.914 vs. 0.802–0.900 and 0.841–0.876, <em>p<!--> </em>=<!--> <!-->0.041–0.484 and 0.011–0.924, respectively). The post-FM further enhanced predictive performance, as evidenced by the highest AUCs in the training and internal validation cohorts (AUC: 0.952 vs. 0.862–0.914, <em>p<!--> </em>=<!--> <!-->0.001–0.116; 0.850 [0.724–0.922] vs. 0.770–0.843, <em>p<!--> </em>=<!--> <!-->0.102–0.922). and IDI values (14.2%–36.4% increase). Additionally, clustering and radiomics features displayed a higher number of correlated feature pairs in the MP/S (+) group than MP/S (-) group.</div></div><div><h3>Conclusion</h3><div>The post-FM, integrating clustering signature, radiomics signature, and clinical characteristics, has been established as a reliable predictor for MP/S status in stage I LAC.</div></div>","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":"32 10","pages":"Pages 6307-6319"},"PeriodicalIF":3.9,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144765775","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":"Non-cardiovascular Calcification Measures and Warranty Period of a Zero CAC in Young Adults: the Multi-ethnic Study of Atherosclerosis","authors":"Elena Ghotbi MD ,&nbsp;Khurram Nasir MD, MPH ,&nbsp;Michael P. Bancks PhD ,&nbsp;Nikhil Subhas ,&nbsp;Sepehr Akhtarkhavari ,&nbsp;Mahsima Shabani MD, MPH ,&nbsp;Roham Hadidchi ,&nbsp;David A. Bluemke MD, PhD ,&nbsp;Wendy S. Post MD, MS ,&nbsp;Matthew Budoff MD ,&nbsp;R. Graham Barr MD, DrPH ,&nbsp;João A.C. Lima MD ,&nbsp;Shadpour Demehri MD","doi":"10.1016/j.acra.2025.06.002","DOIUrl":"10.1016/j.acra.2025.06.002","url":null,"abstract":"<div><h3>Rationale and Objectives</h3><div>This study investigates the warranty period for downstream incident coronary artery calcium (CAC) &gt; 0 and cardiovascular disease (CVD) events in young adults (age &lt; 50 years) with baseline CAC<!--> <!-->=<!--> <!-->0, stratified by non-cardiovascular calcification. The absence of coronary calcification (CAC<!--> <!-->=<!--> <!-->0) is a strong negative risk predictor, especially in young, low-risk individuals, as coronary events without CAC are rare. Beyond known cardiovascular risk factors, propensity for systemic calcium deposition—reflected by non-cardiovascular calcification—may influence incident CAC &gt; 0. Refining risk prediction in young adults, who face cumulative harms from unnecessary imaging and radiation, is important.</div></div><div><h3>Materials and Methods</h3><div>Young adults with baseline CAC score<!--> <!-->=<!--> <!-->0 from Exam 1 (2000–2002) of the MESA cohort (<em>n<!--> </em>=<!--> <!-->402) were stratified by costal cartilage calcification (CCC) into below- and above-median groups (<em>n<!--> </em>=<!--> <!-->201 each). A Weibull parametric survival model estimated time-to-conversion to CAC &gt; 0 and CVD events, enabling estimation of the “warranty period” as the mean time to 25% incidence of CAC &gt; 0% and 2% incidence of CVD events.</div></div><div><h3>Results</h3><div>By Exam 5 (2010–2012), incident CAC &gt; 0 was observed in 27% of the below-median group and 42% of the above-median group. The warranty period for 25% CAC &gt; 0 incidence was shorter in the above-median group (8.0 [95% CI: 7.2, 8.7] vs. 9.6 [8.6, 10.6] years). Conversely, the warranty period for 2% CVD incidence was longer in the above-median group (11.2 [95% CI: 7.4, 18.1] vs. 9.3 [95% CI: 6.5, 14.1] years).</div></div><div><h3>Conclusion</h3><div>In young adults with CAC<!--> <!-->=<!--> <!-->0 at baseline, high non-cardiovascular calcification may indicate a shorter CAC &gt; 0 incidence warranty period and possibly less severe atherosclerotic stages, as reflected by a longer warranty period for CVD events.</div></div>","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":"32 10","pages":"Pages 5724-5732"},"PeriodicalIF":3.9,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144790652","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":"Advancing Neurovascular Imaging: Optimization of Reconstruction Kernel and Quantum Iterative Reconstruction for Ultra-high-resolution Photon-Counting Detector CT Angiography of the Head and Neck","authors":"Adrienn Tóth ,&nbsp;Jennifer Yongjoo Cho ,&nbsp;John Crow ,&nbsp;Evan Wilson ,&nbsp;Kimberly Kicielinski ,&nbsp;Sami Al Kasab ,&nbsp;Jennifer Joyce ,&nbsp;Maria Gisele Matheus ,&nbsp;Eric Bass ,&nbsp;Maria Vittoria Spampinato","doi":"10.1016/j.acra.2025.07.033","DOIUrl":"10.1016/j.acra.2025.07.033","url":null,"abstract":"<div><h3>Rationale and Objectives</h3><div>To find the optimal reconstruction parameters for ultra-high-resolution (UHR) photon-counting detector CT (PCD-CT) angiography of the head and neck, with a special focus on imaging intracranial aneurysms.</div></div><div><h3>Material and Methods</h3><div>18 patients with intracranial aneurysms were prospectively enrolled in this single-center study. CT angiograms were acquired in UHR mode on a clinical PCD-CT scanner. Images were reconstructed with six strength levels of a dedicated neurovascular kernel (Hv48–89) and with quantum iterative reconstruction (QIR) levels 1–4. Image noise, contrast-to-noise ratio, and vessel sharpness were determined for all reconstructions. Qualitative image quality was assessed by three readers using a 5-point Likert scale, for the best-performing reconstructions. Aneurysm dome and neck sizes were independently measured by two readers to assess inter-reader reliability.</div></div><div><h3>Results</h3><div>18 patients (mean age, 64.3 years ± 15 [SD], 5 men,) were evaluated. Three kernels (Hv56, Hv64, Hv72) were identified as best performing in the quantitative analysis. The qualitative analysis demonstrated a consistent preference for QIR level 4 across all kernels in each evaluated category (P&lt;0.001). Hv72 was the most frequently preferred, although it exhibited a higher incidence of image artifacts compared to the other kernels. Inter-reader reliability was high for dome measurements—especially on UHR images—while neck measurements showed greater variability.</div></div><div><h3>Conclusion</h3><div>Hv72 kernel with QIR level 4 proved to be the optimal combination of CTA parameters among the configurations evaluated. Our results could provide reference for protocol optimization on PCD-CT for neurovascular imaging.</div></div>","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":"32 10","pages":"Pages 6093-6103"},"PeriodicalIF":3.9,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144769259","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":"LLM-Based Extraction of Imaging Features from Radiology Reports: Automating Disease Activity Scoring in Crohn’s Disease","authors":"Reza Dehdab MD ,&nbsp;Fiona Mankertz MD ,&nbsp;Jan Michael Brendel MD ,&nbsp;Nour Maalouf MD ,&nbsp;Kenan Kaya MD ,&nbsp;Saif Afat MD ,&nbsp;Shadi Kolahdoozan MD, MPH, PhD ,&nbsp;Amir Reza Radmard MD","doi":"10.1016/j.acra.2025.07.041","DOIUrl":"10.1016/j.acra.2025.07.041","url":null,"abstract":"<div><h3>Rationale and Objectives</h3><div>Large Language Models (LLMs) offer a promising solution for extracting structured clinical information from free-text radiology reports. The Simplified Magnetic Resonance Index of Activity (sMARIA) is a validated scoring system used to quantify Crohn’s disease (CD) activity based on Magnetic Resonance Enterography (MRE) findings. This study aims to evaluate the performance of two advanced LLMs in extracting key imaging features and computing sMARIA scores from free-text MRE reports.</div></div><div><h3>Materials and Methods</h3><div>This retrospective study included 117 anonymized free-text MRE reports from patients with confirmed CD. ChatGPT (GPT-4o) and DeepSeek (DeepSeek-R1) were prompted using a structured input designed to extract four key radiologic features relevant to sMARIA: bowel wall thickness, mural edema, perienteric fat stranding, and ulceration. LLM outputs were evaluated against radiologist annotations at both the segment and feature levels. Segment-level agreement was assessed using accuracy, mean absolute error (MAE) and Pearson correlation. Feature-level performance was evaluated using sensitivity, specificity, precision, and F1-score. Errors including confabulations were recorded descriptively<em>.</em></div></div><div><h3>Results</h3><div>ChatGPT achieved a segment-level accuracy of 98.6%, MAE of 0.17, and Pearson correlation of 0.99. DeepSeek achieved 97.3% accuracy, MAE of 0.51, and correlation of 0.96. At the feature level, ChatGPT yielded an F1-score of 98.8% (precision 97.8%, sensitivity 99.9%), while DeepSeek achieved 97.9% (precision 96.0%, sensitivity 99.8%).</div></div><div><h3>Conclusions</h3><div>LLMs demonstrate near-human accuracy in extracting structured information and computing sMARIA scores from free-text MRE reports. This enables automated assessment of CD activity without altering current reporting workflows, supporting longitudinal monitoring and large-scale research. Integration into clinical decision support systems may be feasible in the future, provided appropriate human oversight and validation are ensured.</div></div>","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":"32 10","pages":"Pages 5869-5877"},"PeriodicalIF":3.9,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144812624","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":"Enhancing ESCC Grading Prediction: Tumor Markers Augmented CT Radiomics and Inflammatory Immune Model","authors":"Yanyan Xiao,&nbsp;Qingxu Liu","doi":"10.1016/j.acra.2025.05.029","DOIUrl":"10.1016/j.acra.2025.05.029","url":null,"abstract":"","PeriodicalId":50928,"journal":{"name":"Academic Radiology","volume":"32 10","pages":"Page 6352"},"PeriodicalIF":3.9,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144856965","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}