Diagnostic and interventional radiology最新文献

{"title":"Clival and paraclival pathologies: imaging features and differential diagnosis.","authors":"Ahmet Bozer, Yeliz Pekçevik","doi":"10.4274/dir.2025.243148","DOIUrl":"https://doi.org/10.4274/dir.2025.243148","url":null,"abstract":"<p><p>Clival and paraclival pathologies encompass a broad spectrum of benign and malignant lesions, necessitating accurate imaging for precise diagnosis and management. Magnetic resonance imaging and computed tomography are pivotal in evaluating these lesions, facilitating differentiation, and guiding therapeutic decisions. This study reviews the imaging characteristics, differential diagnoses, and clinical significance of clival and paraclival pathologies.</p>","PeriodicalId":11341,"journal":{"name":"Diagnostic and interventional radiology","volume":" ","pages":""},"PeriodicalIF":1.4,"publicationDate":"2025-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143718343","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Evaluating artificial intelligence for a focal nodular hyperplasia diagnosis using magnetic resonance imaging: preliminary findings.","authors":"Mecit Kantarcı, Volkan Kızılgöz, Ramazan Terzi, Ahmet Enes Kılıç, Halime Kabalcı, Önder Durmaz, Nil Tokgöz, Mustafa Harman, Ayşegül Sağır Kahraman, Ali Avanaz, Sonay Aydın, Gülsüm Özlem Elpek, Merve Yazol, Bülent Aydınlı","doi":"10.4274/dir.2025.243095","DOIUrl":"https://doi.org/10.4274/dir.2025.243095","url":null,"abstract":"<p><strong>Purpose: </strong>This study aimed to evaluate the effectiveness of artificial intelligence (AI) in diagnosing focal nodular hyperplasia (FNH) of the liver using magnetic resonance imaging (MRI) and compare its performance with that of radiologists.</p><p><strong>Methods: </strong>In the first phase of the study, the MRIs of 60 patients (30 patients with FNH and 30 patients with no lesions or lesions other than FNH) were processed using a segmentation program and introduced to an AI model. After the learning process, the MRIs of 42 different patients that the AI model had no experience with were introduced to the system. In addition, a radiology resident and a radiology specialist evaluated patients with the same MR sequences. The sensitivity and specificity values were obtained from all three reviews.</p><p><strong>Results: </strong>The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of the AI model were found to be 0.769, 0.966, 0.909, and 0.903, respectively. The sensitivity and specificity values were higher than those of the radiology resident and lower than those of the radiology specialist. The results of the specialist versus the AI model revealed a good agreement level, with a kappa (κ) value of 0.777.</p><p><strong>Conclusion: </strong>For the diagnosis of FNH, the sensitivity, specificity, PPV, and NPV of the AI device were higher than those of the radiology resident and lower than those of the radiology specialist. With additional studies focused on different specific lesions of the liver, AI models are expected to be able to diagnose each liver lesion with high accuracy in the future.</p><p><strong>Clinical significance: </strong>AI is studied to provide assisted or automated interpretation of radiological images with an accurate and reproducible imaging diagnosis.</p>","PeriodicalId":11341,"journal":{"name":"Diagnostic and interventional radiology","volume":" ","pages":""},"PeriodicalIF":1.4,"publicationDate":"2025-03-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143709181","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Revolutionizing cardiac imaging: how photon-counting computed tomography is redefining coronary artery stent assessment.","authors":"Çağdaş Topel, Furkan Ufuk","doi":"10.4274/dir.2025.253274","DOIUrl":"https://doi.org/10.4274/dir.2025.253274","url":null,"abstract":"","PeriodicalId":11341,"journal":{"name":"Diagnostic and interventional radiology","volume":" ","pages":""},"PeriodicalIF":1.4,"publicationDate":"2025-03-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143709197","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The role of the Kaiser score system in uncertain malignant potential (B3) breast lesions: a pilot study.","authors":"Fatma Çelik Yabul, Hafize Otçu Temur, Bahar Atasoy, Serdar Balsak, Alpay Alkan, Şeyma Yıldız","doi":"10.4274/dir.2025.242401","DOIUrl":"https://doi.org/10.4274/dir.2025.242401","url":null,"abstract":"<p><strong>Purpose: </strong>This study aims to evaluate the effectiveness of the Kaiser score (KS) system in assessing breast lesions with uncertain malignant potential (B3).</p><p><strong>Methods: </strong>Breast magnetic resonance imaging (MRI) scans from a total of 76 patients with histologically proven B3 lesions were included in this study. The KS was recorded for each MRI scan. The patients were classified based on biopsy results, and upgraded lesions were identified. Statistical analysis was conducted to evaluate the association between high KS values and upgraded lesions.</p><p><strong>Results: </strong>The mean age of the 76 patients was calculated as 49.6 ± 10.1. A significant association was observed between the KS system and the prediction of malignancy upgrade (<i>P</i> < 0.001). Furthermore, among the descriptors, spiculation, margin, and upgrading prediction demonstrated a statistically significant difference (<i>P</i> < 0.001). Additionally, the specificity improved when the accepted KS cut-off value was set at seven instead of five. A significant association was also observed between the KS system and the papilloma upgrade rate within the B3 lesion subgroups (<i>P</i> < 0.001).</p><p><strong>Conclusion: </strong>Breast radiology plays a crucial role in the diagnosis of B3 lesions. Our findings suggest that the KS system holds promise as a tool for predicting the upgrade potential of B3 lesions.</p><p><strong>Clinical significance: </strong>This study demonstrated that the KS system may assist in predicting the upgrade potential of B3 breast lesions. It also demonstrated that spiculation and margin descriptors within the KS system possess a high positive predictive value for upgrade prediction. Additionally, we believe that the KS system can help prevent unnecessary surgeries in patients with B3 lesions.</p>","PeriodicalId":11341,"journal":{"name":"Diagnostic and interventional radiology","volume":" ","pages":""},"PeriodicalIF":1.4,"publicationDate":"2025-03-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143709206","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Dual-energy computed tomography-based volumetric thyroid iodine quantification: correlation with thyroid hormonal status, pathologic diagnosis, and phantom validation.","authors":"Younghen Lee","doi":"10.4274/dir.2025.243132","DOIUrl":"https://doi.org/10.4274/dir.2025.243132","url":null,"abstract":"<p><strong>Purpose: </strong>To investigate the relationship between intrathyroidal iodine concentration (IC) (mg I/mL) and thyroid hormonal status or pathologic diagnosis with the use of dual-energy computed tomography (DECT).</p><p><strong>Methods: </strong>We retrospectively included patients who underwent neck CT examination between September 2016 and August 2021 using a dual-layer DECT scanner (120 kilovolt peak) for preoperative thyroid imaging. We performed volumetric IC measurements at the thyroid parenchyma on the additional iodine map generated from non-contrast images. We then compared the mean IC of thyroid parenchyma based on thyroid hormonal status (hypothyroid, euthyroid, and hyperthyroid) and diffuse thyroid disease (DTD). Additionally, we determined the accuracy of iodine quantification with our site-specific DECT acquisition protocol using a Gammex^TM phantom containing seven iodine inserts with different ICs ranging from 2 to 20 mgI/mL.</p><p><strong>Results: </strong>Among the 578 patients (M:F: 87:491, age: 48.6 ± 11.7 years) who were finally selected, the mean thyroid parenchymal ICs was the lowest in the hyperthyroid group, followed by the hypothyroid group, and then the euthyroid group (0.68 ± 0.37, n = 44 vs. 1.13 ± 0.42, n = 61 vs. 1.32 ± 0.43, n = 473, <i>P</i> < 0.01, respectively). In the patients with euthyroidism, the mean parenchymal IC was already lower in the patients with pathologically proven DTD than in those without DTD (1.22 ± 0.44 mgI/mL vs. 1.45 ± 0.37 mgI/mL, <i>P</i> < 0.01). Based on the phantom study, the median percentage deviations from the expected values were 5.1% for ICs of 2-20 mgI/mL.</p><p><strong>Conclusion: </strong>DECT-based IC quantification could be a potentially useful method for identifying patients with thyroid hormone dysfunction or DTD without the use of contrast media.</p><p><strong>Clinical significance: </strong>Without the need for intravenous administration, DECT-based intrathyroidal IC quantification provides potentially valuable information from the non-contrast CT image of the thyroid parenchyma.</p>","PeriodicalId":11341,"journal":{"name":"Diagnostic and interventional radiology","volume":" ","pages":""},"PeriodicalIF":1.4,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143656431","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Efficacy of endovascular circulating false lumen occlusion in chronic aneurysmal descending aortic dissections.","authors":"Emeric Gremen, Mathieu Finas, Eliott Mathieu, Frédéric Thony, Mathieu Rodiere, Julien Ghelfi","doi":"10.4274/dir.2025.242986","DOIUrl":"https://doi.org/10.4274/dir.2025.242986","url":null,"abstract":"<p><strong>Purpose: </strong>To evaluate the efficacy of endovascular circulating false lumen occlusion (CFLO) in inducing positive aortic remodeling in chronic aneurysmal descending aortic dissection (AD).</p><p><strong>Methods: </strong>This retrospective monocentric study included patients treated by CFLO between 2003 and 2022 in the context of chronic AD with progressive descending aneurysmal evolution and persistent circulating false lumen (FL). The procedure was achieved with coils, plugs, and/or glue at the entry tear or in the FL and/or with covered stenting in the supra-aortic trunk. The primary endpoint evaluated the positive aortic remodeling, defined as stabilization or a decrease in the aortic diameter on a computed tomography scan at the 1-year follow-up after the procedure. The FL circulating status, safety, and occurrence of aneurysm events during follow-up were also evaluated.</p><p><strong>Results: </strong>Twenty patients [median age: 65.4 years, interquartile range (IQR): 58.4-69.9; 13 men] were included, with a median duration from an acute AD of 32.5 months (IQR: 8.8-76.5). Twelve patients (60%) achieved complete FL thrombosis after CFLO, whereas 8/20 patients (40.0%) experienced partial thrombosis. Additionally, positive aortic remodeling was observed in 13 patients (65%). Following the procedure, the aneurysmal aortic diameter decreased in 8/20 patients (40.0%) and remained stable in 5/20 patients (25.0%). Two patients (10%) had complications related to the procedure. Two patients (10%) had secondary aneurysm events during follow-up.</p><p><strong>Conclusion: </strong>CFLO is a feasible and efficient method to induce FL thrombosis and reduce aneurysmal progression in chronic AD.</p><p><strong>Clinical significance: </strong>The positive outcomes observed highlight the potential of this technique to improve patient management in complex aortic pathologies. This approach offers a valuable option in the management of chronic AD and emphasizes the importance of endovascular interventions in enhancing patient outcomes.</p>","PeriodicalId":11341,"journal":{"name":"Diagnostic and interventional radiology","volume":" ","pages":""},"PeriodicalIF":1.4,"publicationDate":"2025-03-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143647683","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Automatic bone age assessment: a Turkish population study.","authors":"Samet Öztürk, Murat Yüce, Gül Gizem Pamuk, Candan Varlık, Ahmet Tan Cimilli, Musa Atay","doi":"10.4274/dir.2025.242999","DOIUrl":"https://doi.org/10.4274/dir.2025.242999","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Purpose: &lt;/strong&gt;Established methods for bone age assessment (BAA), such as the Greulich and Pyle atlas, suffer from variability due to population differences and observer discrepancies. Although automated BAA offers speed and consistency, limited research exists on its performance across different populations using deep learning. This study examines deep learning algorithms on the Turkish population to enhance bone age models by understanding demographic influences.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Methods: &lt;/strong&gt;We analyzed reports from Bağcılar Hospital's Health Information Management System between April 2012 and September 2023 using \"bone age\" as a keyword. Patient images were re-evaluated by an experienced radiologist and anonymized. A total of 2,730 hand radiographs from Bağcılar Hospital (Turkish population), 12,572 from the Radiological Society of North America (RSNA), and 6,185 from the Radiological Hand Pose Estimation (RHPE) public datasets were collected, along with corresponding bone ages and gender information. A random set of 546 radiographs (273 from Bağcılar, 273 from public datasets) was initially randomly split for an internal test set with bone age stratification; the remaining data were used for training and validation. BAAs were generated using a modified InceptionV3 model on 500 × 500-pixel images, selecting the model with the lowest mean absolute error (MAE) on the validation set.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;Three models were trained and tested based on dataset origin: Bağcılar (Turkish), public (RSNA-RHPE), and a Combined model. Internal test set predictions of the Combined model estimated bone age within less than 6, 12, 18, and 24 months at rates of 44%, 73%, 87%, and 94%, respectively. The MAE was 9.2 months in the overall internal test set, 7 months on the public test set, and 11.5 months on the Bağcılar internal test data. The Bağcılar-only model had an MAE of 12.7 months on the Bağcılar internal test data. Despite less training data, there was no significant difference between the combined and Bağcılar models on the Bağcılar dataset (&lt;i&gt;P&lt;/i&gt; &gt; 0.05). The public model showed an MAE of 16.5 months on the Bağcılar dataset, significantly worse than the other models (&lt;i&gt;P&lt;/i&gt; &lt; 0.05).&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Conclusion: &lt;/strong&gt;We developed an automatic BAA model including the Turkish population, one of the few such studies using deep learning. Despite challenges from population differences and data heterogeneity, these models can be effectively used in various clinical settings. Model accuracy can improve over time with cumulative data, and publicly available datasets may further refine them. Our approach enables more accurate and efficient BAAs, supporting healthcare professionals where traditional methods are time-consuming and variable.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Clinical significance: &lt;/strong&gt;The developed automated BAA model for the Turkish population offers a reliable and efficient alternative to traditional methods. By utilizing de","PeriodicalId":11341,"journal":{"name":"Diagnostic and interventional radiology","volume":" ","pages":""},"PeriodicalIF":1.4,"publicationDate":"2025-03-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143647669","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Grading portal vein stenosis following partial hepatectomy by high-frequency ultrasonography: an <i>in vivo</i> study of rats.","authors":"Lin Ma, Chihan Peng, Lulu Yang, Xiaoxia Zhu, Hongxia Fan, Jiali Yang, Hong Wang, Yan Luo","doi":"10.4274/dir.2024.242912","DOIUrl":"10.4274/dir.2024.242912","url":null,"abstract":"<p><strong>Purpose: </strong>To evaluate the diagnostic value of ultrasound in grading portal vein stenosis (PVS) in a rat model of 70% partial hepatectomy (PH).</p><p><strong>Methods: </strong>A total of 96 Sprague-Dawley rats were randomly divided into a PH group and PVS groups with mild, moderate, and severe PVS following PH. Hemodynamic parameters were measured using high-frequency ultrasound (5-12 MHz high-frequency linear transducer), including pre-stenotic, stenotic, and post-stenotic portal vein diameters (PVD_pre, PVD_s, PVDpost); pre-stenotic and stenotic portal vein velocity (PVVpre, PVVs); hepatic artery peak systolic velocity (PSV); end-diastolic velocity; and resistive index. The portal vein diameter ratio (PVDR) and portal vein velocity ratio (PVVR) were calculated using the following formulas: PVDR=PVD_pre/PVD_s and PVVR=PVVs/PVVpre. The value of these parameters in grading PVS was assessed.</p><p><strong>Results: </strong>Portal vein hemodynamics showed gradient changes as PVS aggravated. For identifying >50% PVS, PVD_s and PVDR were the best parameters, with areas under the curve (AUC) of 0.85 and 0.86, respectively. For identifying >65% PVS, PVD_s, PVDR, and PVVR were relatively better, with AUCs of 0.94, 0.85, and 0.88, respectively. The AUC of hepatic artery PSV for identifying >65% PVS was 0.733.</p><p><strong>Conclusion: </strong>High-frequency ultrasonography can be used to grade PVS in rats, with PVD_s, PVDR, and PVVR being particularly useful. Hepatic artery PSV may help in predicting >65% PVS. These findings provide valuable information for PVS rat model research and offer an experimental basis for further studies on PVS evaluation in living-donor liver transplantation (LDLT).</p><p><strong>Clinical significance: </strong>Ultrasonography serves as a first-line technology for diagnosing PVS following LDLT. However, the grading criteria for PVS severity remain unclear. Investigating the use of ultrasonic hemodynamics in the early diagnosis of PVS and grading stenosis severity is important for early postoperative intervention and improving recipient survival rates.</p>","PeriodicalId":11341,"journal":{"name":"Diagnostic and interventional radiology","volume":" ","pages":"68-74"},"PeriodicalIF":1.4,"publicationDate":"2025-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11880861/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142709178","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Artificial intelligence in musculoskeletal applications: a primer for radiologists.","authors":"Michelle W Tong, Jiamin Zhou, Zehra Akkaya, Sharmila Majumdar, Rupsa Bhattacharjee","doi":"10.4274/dir.2024.242830","DOIUrl":"10.4274/dir.2024.242830","url":null,"abstract":"<p><p>As an umbrella term, artificial intelligence (AI) covers machine learning and deep learning. This review aimed to elaborate on these terms to act as a primer for radiologists to learn more about the algorithms commonly used in musculoskeletal radiology. It also aimed to familiarize them with the common practices and issues in the use of AI in this domain.</p>","PeriodicalId":11341,"journal":{"name":"Diagnostic and interventional radiology","volume":" ","pages":"89-101"},"PeriodicalIF":1.4,"publicationDate":"2025-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11880867/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141999601","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Evaluating text and visual diagnostic capabilities of large language models on questions related to the Breast Imaging Reporting and Data System Atlas 5^th edition.","authors":"Yasin Celal Güneş, Turay Cesur, Eren Çamur, Leman Günbey Karabekmez","doi":"10.4274/dir.2024.242876","DOIUrl":"10.4274/dir.2024.242876","url":null,"abstract":"<p><strong>Purpose: </strong>This study aimed to evaluate the performance of large language models (LLMs) and multimodal LLMs in interpreting the Breast Imaging Reporting and Data System (BI-RADS) categories and providing clinical management recommendations for breast radiology in text-based and visual questions.</p><p><strong>Methods: </strong>This cross-sectional observational study involved two steps. In the first step, we compared ten LLMs (namely ChatGPT 4o, ChatGPT 4, ChatGPT 3.5, Google Gemini 1.5 Pro, Google Gemini 1.0, Microsoft Copilot, Perplexity, Claude 3.5 Sonnet, Claude 3 Opus, and Claude 3 Opus 200K), general radiologists, and a breast radiologist using 100 text-based multiple-choice questions (MCQs) related to the BI-RADS Atlas 5^th edition. In the second step, we assessed the performance of five multimodal LLMs (ChatGPT 4o, ChatGPT 4V, Claude 3.5 Sonnet, Claude 3 Opus, and Google Gemini 1.5 Pro) in assigning BI-RADS categories and providing clinical management recommendations on 100 breast ultrasound images. The comparison of correct answers and accuracy by question types was analyzed using McNemar's and chi-squared tests. Management scores were analyzed using the Kruskal- Wallis and Wilcoxon tests.</p><p><strong>Results: </strong>Claude 3.5 Sonnet achieved the highest accuracy in text-based MCQs (90%), followed by ChatGPT 4o (89%), outperforming all other LLMs and general radiologists (78% and 76%) (<i>P</i> < 0.05), except for the Claude 3 Opus models and the breast radiologist (82%) (<i>P</i> > 0.05). Lower-performing LLMs included Google Gemini 1.0 (61%) and ChatGPT 3.5 (60%). Performance across different categories of showed no significant variation among LLMs or radiologists (<i>P</i> > 0.05). For breast ultrasound images, Claude 3.5 Sonnet achieved 59% accuracy, significantly higher than other multimodal LLMs (<i>P</i> < 0.05). Management recommendations were evaluated using a 3-point Likert scale, with Claude 3.5 Sonnet scoring the highest (mean: 2.12 ± 0.97) (<i>P</i> < 0.05). Accuracy varied significantly across BI-RADS categories, except Claude 3 Opus (<i>P</i> < 0.05). Gemini 1.5 Pro failed to answer any BI-RADS 5 questions correctly. Similarly, ChatGPT 4V failed to answer any BI-RADS 1 questions correctly, making them the least accurate in these categories (<i>P</i> < 0.05).</p><p><strong>Conclusion: </strong>Although LLMs such as Claude 3.5 Sonnet and ChatGPT 4o show promise in text-based BI-RADS assessments, their limitations in visual diagnostics suggest they should be used cautiously and under radiologists' supervision to avoid misdiagnoses.</p><p><strong>Clinical significance: </strong>This study demonstrates that while LLMs exhibit strong capabilities in text-based BI-RADS assessments, their visual diagnostic abilities are currently limited, necessitating further development and cautious application in clinical practice.</p>","PeriodicalId":11341,"journal":{"name":"Diagnostic and interventional radiology","volume":" ","pages":"111-129"},"PeriodicalIF":1.4,"publicationDate":"2025-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11880873/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142153440","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}