American Journal of Roentgenology最新文献

{"title":"Deep Learning for Synthetic Postcontrast T1-Weighted MRI: A Systematic Review With Targeted Meta-Analysis of Brain Tumor Studies.","authors":"Siddhant Dogra, Emmy Hu, Stella K Kang","doi":"10.2214/AJR.26.34673","DOIUrl":"https://doi.org/10.2214/AJR.26.34673","url":null,"abstract":"<p><p><b>Background.</b> Gadolinium-based contrast agents remain essential for MRI but carry risks. Deep learning (DL) methods have emerged as a potential approach for synthesizing postcontrast T1-weighted images from precontrast sequences alone. <b>Objective.</b> The objective of this study was to systematically review DL-based synthesis of postcontrast T1-weighted MRI, characterize model architectures and evaluation practices across subspecialties, and perform targeted meta-analysis where sufficient literature existed. <b>Evidence Acquisition.</b> A systematic search of PubMed, Embase, Cochrane Central, Scopus, and Web of Science (through January 16, 2025) identified peer-reviewed studies using DL to synthesize postcontrast T1-weighted MRI from precontrast sequences in adults. Two reviewers independently screened studies, extractingdata on subspecialty, architecture, quantitative metrics, pathology-specific evaluation, and reader studies. Risk of bias was assessed using modified QUADAS-2. Random-effects meta-analysis was performed for brain tumor studies. <b>Evidence Synthesis.</b> Of 268 records after deduplication, 41 met inclusion criteria. Most studies focused on neuroimaging (n = 24, 59%), followed by breast (n = 7, 17%) and body imaging (n = 6, 15%). Generative adversarial networks (n = 20, 45%) and convolutional neural networks (n = 19, 43%) predominated. Structural similarity index measure (SSIM, n = 31, 76%) and peak SNR (PSNR, n = 28, 68%) were the most common metrics. Fifty-one percent (n = 21) of studies performed pathology-specific evaluation, which showed substantially lower SSIM and PSNR compared with whole-image metrics. Thirty-seven percent (n = 15) included reader studies, 29% (n = 12) released code, and 61% (n = 25) used single-institution data. Meta-analysis of 15 brain tumor studies (30 models) yielded pooled SSIM of 0.92 (95% CI, 0.90-0.93) and PSNR of 30.6 dB (95% CI, 28.6-32.6). Given extreme heterogeneity (I² > 99%), pooled estimates should be interpreted as descriptive. <b>Conclusion.</b> DL-based postcontrast MRI synthesis shows technical feasibility across subspecialties but suffers from substantial heterogeneity in study design, inconsistent quantitative metric computation, and limited clinical validation. <b>Clinical Impact.</b> Limited rates of reader studies and external validation represent key barriers to clinical translation of DL-based postcontrast MRI synthesis. Standardized evaluation workflows incorporating whole-image metrics, pathology-specific assessment, and reader studies are essential before these techniques can be translated into clinical practice.</p>","PeriodicalId":55529,"journal":{"name":"American Journal of Roentgenology","volume":" ","pages":""},"PeriodicalIF":6.1,"publicationDate":"2026-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147846621","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":"The Future of Quantitative Breast Imaging: Integrated to Integral Imaging Biomarkers for Precision Cancer Care, From the <i>AJR</i> Special Series on Quantitative Imaging.","authors":"Elizabeth S McDonald, Anum S Kazerouni, Pouya Metanet, Lev Barinov, Savannah C Partridge, Nola M Hylton","doi":"10.2214/AJR.25.32958","DOIUrl":"https://doi.org/10.2214/AJR.25.32958","url":null,"abstract":"<p><p>This review explores the potential of quantitative imaging biomarkers (QIBs) for guiding personalized breast cancer treatment through prognostic stratification and measurement of treatment response. Accumulating data from prospective clinical trials show that QIBs are more predictive of treatment outcome than conventional sizebased measures (e.g., change in tumor diameter), highlighting these markers' potential clinical impact. Although QIBs have shown utility in the trial setting, clinical adoption has remained limited. In this article, we present biomarkers derived from dynamic contrast-enhanced MRI, DWI, [18F]FDG PET, and [18F]FES PET, and highlight implementation challenges including variable acquisition protocols, lack of standardization, and uncertainty around optimal timing of imaging during treatment. We distinguish between integrated biomarkers-those used for correlative or exploratory purposes within a trial-and integral biomarkers-those that are essential to trial design and directly inform patient stratification or therapeutic decisions. Standardization efforts by national and international organizations are underway to ensure imaging marker reliability, and commercial tools supporting clinical translation have become available. Dedicated end-to-end solutions remain needed to integrate QIBs into clinical workflows and ultimately into standard care pathways, to promote these markers' role in patient stratification and adaptive treatment decisions.</p>","PeriodicalId":55529,"journal":{"name":"American Journal of Roentgenology","volume":" ","pages":""},"PeriodicalIF":6.1,"publicationDate":"2026-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147846684","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":"Early-Career Momentum: The Method Behind the Madness-From Training to Practice (Episode 11).","authors":"Surbhi Raichandani, Patricia Balthazar","doi":"10.2214/AJR.26.35137","DOIUrl":"https://doi.org/10.2214/AJR.26.35137","url":null,"abstract":"","PeriodicalId":55529,"journal":{"name":"American Journal of Roentgenology","volume":" ","pages":""},"PeriodicalIF":6.1,"publicationDate":"2026-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147846633","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":"Probable Stricturing in Small Bowel Crohn Disease: In Support of SAR Consensus Recommendations.","authors":"Jonathan R Dillman, Florian Rieder, Mark E Baker, Joel G Fletcher, David H Bruining, Subra Kugathasan, Lee A Denson","doi":"10.2214/AJR.25.34245","DOIUrl":"10.2214/AJR.25.34245","url":null,"abstract":"<p><p>Crohn disease is a chronic immune-mediated gastrointestinal disorder affecting nearly 1 million people in the United States. Despite therapeutic advances, many patients do not achieve durable intestinal healing, and stricture-related complications, including internal penetrating disease, remain frequent. Disease often progresses radiologically from isolated active inflammation (characterized by bowel wall thickening), to associated luminal narrowing (i.e., > 50% diameter reduction), to associated upstream dilatation (i.e., overt stricture, with obstruction risk). The intermediate stage of wall thickening with luminal narrowing is associated with distinct biomarker profiles, microbial signatures, gene expression patterns, imaging features, and prognostic outcomes. Accordingly, consensus statements from the Society of Abdominal Radiology Inflammatory Bowel Disease Disease-Focused Panel indicate that patients with bowel wall thickening and fixed luminal narrowing without upstream dilatation should be considered to have probable strictures. Yet, current clinical classification systems do not fully incorporate this radiologic entity. For example, the Montreal and Paris classifications have historically classified patients with wall thickening and luminal narrowing as having the inflammatory rather than stricturing phenotype. This Perspective summarizes evidence supporting probable strictures as a distinct biologic and clinical entity that could aid individualized care and argues for integrating probable stricturing into radiology reports, clinical classification schemes, and treatment pathways.</p>","PeriodicalId":55529,"journal":{"name":"American Journal of Roentgenology","volume":" ","pages":"1-6"},"PeriodicalIF":6.1,"publicationDate":"2026-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145919211","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":"Artificial Intelligence Sees the Image, Radiologists See the Patient.","authors":"Yee Seng Ng, Antonio C Westphalen","doi":"10.2214/AJR.26.34823","DOIUrl":"https://doi.org/10.2214/AJR.26.34823","url":null,"abstract":"<p><p>Radiology is often portrayed as the medical specialty most vulnerable to replacement by artificial intelligence (AI), potentially impacting medical students' interest in entering the field. In practice, radiologists' work depends on clinical synthesis, judgment under uncertainty, and real-time collaboration within multidisciplinary teams-capabilities that current AI systems lack. Rather than being threatened by AI, radiologists are better served by coevolving with AI, deliberately adapting their skills to meet the demands of future radiology practice.</p>","PeriodicalId":55529,"journal":{"name":"American Journal of Roentgenology","volume":" ","pages":""},"PeriodicalIF":6.1,"publicationDate":"2026-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147846564","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":"Black-Blood CTA Using Photon-Counting Detector CT for Depiction of Small Plaque on the Middle Cerebral Artery Vessel Wall.","authors":"Xinying Huang, Zeyu Liu","doi":"10.2214/AJR.25.34291","DOIUrl":"10.2214/AJR.25.34291","url":null,"abstract":"","PeriodicalId":55529,"journal":{"name":"American Journal of Roentgenology","volume":" ","pages":"1"},"PeriodicalIF":6.1,"publicationDate":"2026-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145821917","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":"Leveraging Fine-Tuned Large Language Models for Interpretable Pancreatic Cystic Lesion Feature Extraction and Risk Categorization.","authors":"Ebrahim Rasromani, Stella K Kang, Yanqi Xu, Beisong Liu, Garvit Luhadia, Wan Fung Chui, Felicia L Pasadyn, Yu Chih Hung, Julie Y An, Edwin Mathieu, Zehui Gu, Carlos Fernandez-Granda, Ammar A Javed, Greg D Sacks, Tamas Gonda, Chenchan Huang, Yiqiu Shen","doi":"10.2214/AJR.25.34076","DOIUrl":"https://doi.org/10.2214/AJR.25.34076","url":null,"abstract":"<p><p><b>Background:</b> Manual extraction of pancreatic cystic lesion (PCL) features from radiology reports is labor-intensive, limiting large-scale studies needed to advance PCL research. <b>Objective:</b> The purpose of this study was to evaluate GPT-4o (closed source), Llama (open source), and DeepSeek (open source) large language models (LLMs) for PCL feature extraction, without and with chain-of-thought (CoT) reasoning. <b>Methods:</b> We curated a dataset of 6469 abdominal MRI or CT reports (2005-2024) from 5615 patients that described PCLs. Llama and DeepSeek were fine-tuned using Quantized Low-Rank Adaptation on GPT-4o-generated CoT labels for extracting PCL and main pancreatic duct features. Features were mapped to risk categories per institutional policy. Evaluation was performed on 285 held-out human-annotated reports from 281 patients. Model outputs for 100 cases were independently reviewed by three radiologists. Feature extraction was evaluated using exact match accuracy, risk categorization with macro-averaged F1 score, and radiologist-model agreement with Fleiss' kappa. Error analyses were performed to assess how and why models made mistakes. <b>Results:</b> CoT fine-tuned LLMs showed a feature extraction accuracy of 97% (95% CI, 97-98%) for Llama, 98% (95% CI, 97-98%) for DeepSeek, and 97% (95% CI, 97-98%) for GPT-4o. Risk categorization F1 scores were 0.93 (95% CI, 0.89-0.97) for Llama, 0.94 (95% CI, 0.90-0.98) for DeepSeek, and 0.97 (95% CI, 0.93-0.99) for GPT-4o. Radiologist interreader agreement was high (κ = 0.888) and showed no significant difference with the addition of Llama (κ = 0.882; p > .99), DeepSeek (κ = 0.893, p > .99), or GPT (κ = 0.897, p > .99). Across all models, object identification and clinical reasoning were the most frequent error types, accounting for 29.3-37.3% and 18.1-21.1% of total errors, respectively. <b>Conclusion:</b> LLMs show feasibility for automatically extracting PCL features from radiology reports. Fine-tuned open-source LLMs achieved performance comparable to GPT-4o. CoT reasoning improved accuracy and enabled interpretable error analysis. Model-assigned risk categories showed high agreement with abdominal radiologists. <b>Clinical Impact:</b> LLMs have the potential to enable creation of large structured registries from existing radiology reports to support population-level research on PCLs.</p>","PeriodicalId":55529,"journal":{"name":"American Journal of Roentgenology","volume":" ","pages":""},"PeriodicalIF":6.1,"publicationDate":"2026-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147846570","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":"Risk Stratification for Breast Cancer Screening: <i>AJR</i> Expert Panel Narrative Review.","authors":"Cody Schopf, Susan M Domchek, Jeffrey A Tice, Peter R Eby, Ritse Mann, Constance D Lehman, Andrea Cozzi, Hari Trivedi, Janie M Lee","doi":"10.2214/AJR.26.34641","DOIUrl":"https://doi.org/10.2214/AJR.26.34641","url":null,"abstract":"<p><p>Early detection of breast cancer reduces mortality and is influenced by screening strategies. The balance of benefits and harms within any screening program improves when screening is aligned with an individual patient's cancer risk profile, and specialized risk prediction tools can support this assessment. Although professional societies provide guidance on breast cancer risk evaluation, their recommendations vary. This variability creates uncertainty regarding standardized approaches to risk assessment, particularly with respect to the use of risk prediction tools, the optimal timing of risk assessment, and the translation of calculated risk into actionable screening decisions. This <i>AJR</i> Expert Panel Narrative Review provides an overview of breast cancer risk stratification, commonly used risk prediction models in current clinical practice, existing societal guidelines related to model use, and considerations for implementing these tools. The article applies such insights to propose a practical approach to incorporating risk assessment into clinical practice. Finally, the article highlights ongoing research aimed at improving breast cancer risk stratification and explores the potential role of deep learning-based prediction models in informing future screening strategies.</p>","PeriodicalId":55529,"journal":{"name":"American Journal of Roentgenology","volume":" ","pages":""},"PeriodicalIF":6.1,"publicationDate":"2026-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147846704","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":"Recognizing and Valuing Radiologic and Ultrasound Technologists' Role in Patient-Centered Radiologic Care: Perspective From a Medical Student in Rural Canada.","authors":"Zier Zhou","doi":"10.2214/AJR.26.35135","DOIUrl":"https://doi.org/10.2214/AJR.26.35135","url":null,"abstract":"","PeriodicalId":55529,"journal":{"name":"American Journal of Roentgenology","volume":" ","pages":""},"PeriodicalIF":6.1,"publicationDate":"2026-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147846734","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":"The Peripheral Echogenic Zone Is Not a Rind.","authors":"Ellen B Mendelson, Wendie A Berg","doi":"10.2214/AJR.25.34372","DOIUrl":"10.2214/AJR.25.34372","url":null,"abstract":"","PeriodicalId":55529,"journal":{"name":"American Journal of Roentgenology","volume":" ","pages":"1"},"PeriodicalIF":6.1,"publicationDate":"2026-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145769822","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}