Japanese Journal of Radiology最新文献

{"title":"Gadoxetic acid-enhanced MRI in hepatocellular carcinoma: a comprehensive review of diagnostic, surveillance, and treatment response prediction and assessment.","authors":"Kumi Ozaki, Yukichi Tanahashi, Satoshi Goshima","doi":"10.1007/s11604-025-01870-x","DOIUrl":"https://doi.org/10.1007/s11604-025-01870-x","url":null,"abstract":"<p><p>Gadoxetic acid-enhanced magnetic resonance imaging (MRI) has become a pivotal imaging modality in hepatocellular carcinoma (HCC) management, offering unique advantages owing to its hepatocyte-specific contrast properties. Its technical foundation includes optimized dynamic phase imaging and hepatobiliary phase (HBP) acquisition, which together provide functional information unattainable with conventional extracellular contrast agents. This modality enhances sensitivity in detecting HCC and enables superior characterization of focal liver lesions based on hepatocyte-specific uptake patterns. In high-risk patients with chronic liver disease, gadoxetic acid-enhanced MRI facilitates the early detection of small and early-stage HCCs, enabling timely intervention and potentially improving clinical outcomes. Beyond diagnosis, gadoxetic acid-enhanced MRI aids in predicting treatment response by evaluating tumor biological characteristics. Key imaging biomarkers include: hyperintense or heterogeneous HCC on HBP, suggesting tumor immune microenvironment; peritumoral hypointensity on HBP, suggesting microvascular invasion; and clear hypointensity on HBP with several other findings, indicating vessels encapsulating tumor clusters, characteristic of the macrotrabecular-massive HCC subtype. These biomarkers support a comprehensive evaluation of histological differentiation and biological aggressiveness. Furthermore, this modality demonstrates superior accuracy in assessing local therapy effectiveness and monitoring systemic treatment responses compared to conventional imaging. Major international hepatology societies have incorporated gadoxetic acid-enhanced MRI into their HCC diagnostic algorithms, albeit with regional differences in emphasis. Eastern guidelines (e.g., from the Japan Society of Hepatology and the Asian Pacific Association for the Study of the Liver) prioritize sensitivity, whereas Western guidelines (e.g., from the European Association for the Study of the Liver and the Liver Imaging Reporting and Data System) emphasize specificity. Despite certain limitations, including potential suboptimal arterial phase visualization, challenges in interpreting the transitional phase, higher cost, and longer examination times, gadoxetic acid-enhanced MRI remains an indispensable tool in precision oncology, enabling personalized treatment strategies and supporting optimal patient outcomes through comprehensive HCC characterization and accurate treatment monitoring.</p>","PeriodicalId":14691,"journal":{"name":"Japanese Journal of Radiology","volume":" ","pages":""},"PeriodicalIF":2.1,"publicationDate":"2025-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145064595","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":"MRI and CT features of head and neck myoepithelioma: comparison with parotid pleomorphic adenoma.","authors":"Hiroki Kato, Takuya Seko, Hirofumi Shibata, Takenori Ogawa, Tomohiro Ando, Masaya Kawaguchi, Yoshifumi Noda, Abdelazim Elsayed Elhelaly, Hirohiko Imai, Masayuki Matsuo","doi":"10.1007/s11604-025-01867-6","DOIUrl":"https://doi.org/10.1007/s11604-025-01867-6","url":null,"abstract":"<p><strong>Purpose: </strong>To evaluate the MRI and CT features of head and neck myoepithelioma in comparison with parotid pleomorphic adenoma.</p><p><strong>Methods: </strong>This retrospective study included 11 patients with histopathologically confirmed myoepithelioma of the head and neck and 103 patients with pleomorphic adenoma of the parotid gland, all of whom underwent preoperative MRI. Among them, seven patients with myoepithelioma and 29 with pleomorphic adenoma also underwent preoperative CT. MRI and CT findings were compared between the two groups.</p><p><strong>Results: </strong>Multinodular configuration (27% vs. 4%), mild hyperintensity relative to the spinal cord on T2-weighted images (91% vs. 48%), and focal unenhanced areas on fat-suppressed contrast-enhanced T1-weighted images (100% vs. 47%) were significantly more frequent in myoepithelioma than in pleomorphic adenoma, respectively (p < 0.05). In contrast, marked hyperintensity relative to the spinal cord on T2-weighted images (46% vs. 9%), higher signal intensity ratios on T2-weighted images (1.68 ± 0.47 vs. 1.40 ± 0.39), and higher apparent diffusion coefficient (ADC) values (1.68 ± 0.36 vs. 1.38 ± 0.23 × 10^-3 mm²/s) were significantly more common in pleomorphic adenoma than in myoepithelioma, respectively (p < 0.05). Contrast-enhanced CT attenuation was significantly higher in myoepithelioma than in pleomorphic adenoma (93.3 ± 10.5 vs. 59.2 ± 22.8 HU, p < 0.05).</p><p><strong>Conclusions: </strong>Although MRI and CT features of myoepithelioma and pleomorphic adenoma can overlap, the presence of a multinodular configuration, focal unenhanced areas, lower T2 signal intensity, lower ADC values, and higher contrast-enhanced CT attenuation may aid in differentiating myoepithelioma from pleomorphic adenoma.</p>","PeriodicalId":14691,"journal":{"name":"Japanese Journal of Radiology","volume":" ","pages":""},"PeriodicalIF":2.1,"publicationDate":"2025-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145064641","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":"Arterial and venous phase evaluation in fast kV-switching dual-energy CT for detecting acute small bowel ischemia caused by small bowel obstruction.","authors":"Masazumi Matsuda, Motoko Konno, Takahiro Otani, Tomoki Tozawa, Kento Hatakeyama, Toshiki Murata, Emika Murasawa, Daichi Sugawara, Junichi Arita, Hajime Nakae, Naoko Mori","doi":"10.1007/s11604-025-01873-8","DOIUrl":"https://doi.org/10.1007/s11604-025-01873-8","url":null,"abstract":"<p><strong>Purpose: </strong>To compare the diagnostic performance of fast kV-switching dual-energy CT (DECT) parameters, including virtual monochromatic imaging (VMI) and iodine mapping, in the arterial and venous phases for detecting surgically confirmed acute small bowel ischemia (ASBI) in cases suspected as ASBI on conventional visual CT findings.</p><p><strong>Materials and methods: </strong>Thirty-two patients with conventional visual CT findings suggesting possible or suspected ASBI caused by small bowel obstruction (SBO) were included. Two radiologists independently placed ten cursors on visually hypo-perfused bowel wall regions to measure CT values at 70 keV, 40 keV, and iodine quantity. Patients were categorized into surgically confirmed ASBI (n = 12) and non-confirmed ASBI (n = 20). ROC analysis assessed diagnostic performance, and inter-observer reliability was evaluated using intra-class correlation coefficients (ICC).</p><p><strong>Results: </strong>For both observers, the CT value at 40 keV and iodine quantity in the arterial phase, as well as iodine quantity in the venous phase, was significantly different between surgically confirmed ASBI and non-confirmed ASBI groups (p < 0.05). Iodine quantity consistently yielded the highest AUC among the evaluated parameters in each phase although the differences compared to 70-keV VMI were not statistically significant. The parameters in the arterial phase tended to demonstrate higher AUCs than those in the venous phase. Inter-observer agreement was moderate to substantial (ICC 0.585-0.741), while intra-observer agreement was substantial to almost perfect (ICC 0.733-0.940).</p><p><strong>Conclusions: </strong>DECT parameters, such as the CT value at 40-keV and the iodine quantity, were effective in differentiating surgically confirmed ASBI from non-confirmed ASBI in SBO cases with suspected ischemia. Iodine quantity showed the highest diagnostic performance among all evaluated parameters. Although the differences were not statistically significant, arterial phase parameters generally yielded higher AUCs than those in the venous phase, suggesting the potential utility of arterial phase DECT in the detection of ASBI.</p>","PeriodicalId":14691,"journal":{"name":"Japanese Journal of Radiology","volume":" ","pages":""},"PeriodicalIF":2.1,"publicationDate":"2025-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145053634","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":"Assessing accuracy and legitimacy of multimodal large language models on Japan Diagnostic Radiology Board Examination.","authors":"Yuichiro Hirano, Soichiro Miki, Yosuke Yamagishi, Shouhei Hanaoka, Takahiro Nakao, Tomohiro Kikuchi, Yuta Nakamura, Yukihiro Nomura, Takeharu Yoshikawa, Osamu Abe","doi":"10.1007/s11604-025-01861-y","DOIUrl":"https://doi.org/10.1007/s11604-025-01861-y","url":null,"abstract":"<p><strong>Purpose: </strong>To assess and compare the accuracy and legitimacy of multimodal large language models (LLMs) on the Japan Diagnostic Radiology Board Examination (JDRBE).</p><p><strong>Materials and methods: </strong>The dataset comprised questions from JDRBE 2021, 2023, and 2024, with ground-truth answers established through consensus among multiple board-certified diagnostic radiologists. Questions without associated images and those lacking unanimous agreement on answers were excluded. Eight LLMs were evaluated: GPT-4 Turbo, GPT-4o, GPT-4.5, GPT-4.1, o3, o4-mini, Claude 3.7 Sonnet, and Gemini 2.5 Pro. Each model was evaluated under two conditions: with inputting images (vision) and without (text-only). Performance differences between the conditions were assessed using McNemar's exact test. Two diagnostic radiologists (with 2 and 18 years of experience) independently rated the legitimacy of responses from four models (GPT-4 Turbo, Claude 3.7 Sonnet, o3, and Gemini 2.5 Pro) using a five-point Likert scale, blinded to model identity. Legitimacy scores were analyzed using Friedman's test, followed by pairwise Wilcoxon signed-rank tests with Holm correction.</p><p><strong>Results: </strong>The dataset included 233 questions. Under the vision condition, o3 achieved the highest accuracy at 72%, followed by o4-mini (70%) and Gemini 2.5 Pro (70%). Under the text-only condition, o3 topped the list with an accuracy of 67%. Addition of image input significantly improved the accuracy of two models (Gemini 2.5 Pro and GPT-4.5), but not the others. Both o3 and Gemini 2.5 Pro received significantly higher legitimacy scores than GPT-4 Turbo and Claude 3.7 Sonnet from both raters.</p><p><strong>Conclusion: </strong>Recent multimodal LLMs, particularly o3 and Gemini 2.5 Pro, have demonstrated remarkable progress on JDRBE questions, reflecting their rapid evolution in diagnostic radiology. Eight multimodal large language models were evaluated on the Japan Diagnostic Radiology Board Examination. OpenAI's o3 and Google DeepMind's Gemini 2.5 Pro achieved high accuracy rates (72% and 70%) and received good legitimacy scores from human raters, demonstrating steady progress.</p>","PeriodicalId":14691,"journal":{"name":"Japanese Journal of Radiology","volume":" ","pages":""},"PeriodicalIF":2.1,"publicationDate":"2025-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145040205","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":"Association between FDG accumulation in interstitial lesions and acute exacerbation risk in lung cancer: multicenter analysis.","authors":"Yuriko Ishida, Shiro Watanabe, Jun Sakakibara-Konishi, Yasuyuki Ikezawa, Hajime Kikuchi, Yasutaka Kawai, Hirokazu Kimura, Sho Nakakubo, Kenji Hirata, Kohsuke Kudo, Satoshi Konno","doi":"10.1007/s11604-025-01869-4","DOIUrl":"https://doi.org/10.1007/s11604-025-01869-4","url":null,"abstract":"<p><strong>Purpose: </strong>Interstitial pneumonia (IP) is associated with poor prognosis in lung cancer and increases the risk of acute exacerbation (AE). Few studies analyzed the relationship between fluorodeoxyglucose (FDG) accumulation in IP with lung cancer complicated with IP and the incidence of AE. This study investigates the association between FDG accumulation in the interstitial lesions and the AE incidence in patients with lung cancer complicated with IP.</p><p><strong>Materials and methods: </strong>This multicenter, retrospective study included patients with lung cancer complicated with IP who received chemotherapy. All CTs at baseline and the onset of AE were centrally adjudicated. The SUVpeak and FDGscore for interstitial lesions were calculated from FDG positron emission tomography images before chemotherapy, and these values were corrected using reference uptake. To determine the association with AE risk, clinical characteristics and imaging findings were compared between patients who developed AE and those who did not. Subsequently, logistic regression analysis was performed to identify risk factors for the development of AE.</p><p><strong>Results: </strong>One hundred and thirteen patients who met the eligibility criteria were enrolled from three centers. However, 9 patients with a clinical diagnosis of collagen-related interstitial pneumonia were excluded due to predominant FDG accumulation in the interstitial lesions, and 104 patients were analyzed. Of those patients, 31.7% (33/104) developed all grade AE and 18.3% (19/104) developed grade 3 or higher. There were no significant differences in patient characteristics and imaging patterns between those with and without AE. SUVpeak in the ipsilateral and contralateral interstitial lesions to the tumor and the FDGscore did not differ between those with or without AE.</p><p><strong>Conclusions: </strong>No association was observed between FDG accumulation in interstitial lesions and AE in patients with lung cancer complicated with IP. We may have to remain cautious about the risk of AE in lung cancer complicated with IP, even when FDG accumulation in interstitial lesions is high or low.</p>","PeriodicalId":14691,"journal":{"name":"Japanese Journal of Radiology","volume":" ","pages":""},"PeriodicalIF":2.1,"publicationDate":"2025-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145040165","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":"Comparison of DLIR and ASIR-V algorithms for virtual monoenergetic imaging in carotid CTA under a triple-low protocol.","authors":"Juan Long, Chenzi Wang, Meng Yu, Xiaohan Liu, Wenbei Xu, Zhongxiao Liu, Chong Wang, Yang Wu, Aiyun Sun, Shuai Zhang, Chunfeng Hu, Kai Xu, Yankai Meng","doi":"10.1007/s11604-025-01866-7","DOIUrl":"https://doi.org/10.1007/s11604-025-01866-7","url":null,"abstract":"<p><strong>Background: </strong>Stroke, frequently associated with carotid artery disease, is evaluated using carotid computed tomography angiography (CTA). Dual-energy CTA (DE-CTA) enhances imaging quality but presents challenges in maintaining high image clarity with low-dose scans.</p><p><strong>Objectives: </strong>To compare the image quality of 50 keV virtual monoenergetic images (VMI) generated using Deep Learning Image Reconstruction (DLIR) and Adaptive Statistical Iterative Reconstruction-V (ASIR-V) algorithms under a triple-low scanning protocol in carotid CTA.</p><p><strong>Methods: </strong>A prospective study was conducted with 120 patients undergoing DE-CTA. The control group (Group 1), with a noise index (NI) of 4.0 and a contrast agent dose of 0.5 mL/kg, used the ASIR-V algorithm. The experimental group was divided into four subgroups: Group 2 (ASIR-V 50%), Group 3 (DLIR-L), Group 4 (DLIR-M), and Group 5 (DLIR-H), with a higher NI of 13.0 and a reduced contrast agent dose of 0.4 mL/kg. Objective image quality was assessed through signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and standard deviation (SD), while subjective quality was evaluated using a 5-point Likert scale. Radiation dose and contrast agent volume were also measured.</p><p><strong>Results: </strong>The triple-low scanning protocol reduced radiation exposure by 53.2%, contrast agent volume by 19.7%, and injection rate by 19.8%. The DLIR-H setting outperformed ASIR-V, demonstrating superior image quality, better noise suppression, and improved contrast in small vessels. VMI at 50 keV showed enhanced diagnostic clarity with minimal radiation and contrast agent usage.</p><p><strong>Conclusion: </strong>The DLIR algorithm, particularly at high settings, significantly enhances image quality in DE-CTA VMI under a triple-low scanning protocol, offering a better balance between radiation dose reduction and image clarity.</p>","PeriodicalId":14691,"journal":{"name":"Japanese Journal of Radiology","volume":" ","pages":""},"PeriodicalIF":2.1,"publicationDate":"2025-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145023357","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":"Responce to Coronary computed tomography angiography using the diluted contrast material protocol: a technique for achieving uniform coronary artery enhancement.","authors":"Kardos Marek","doi":"10.1007/s11604-025-01864-9","DOIUrl":"https://doi.org/10.1007/s11604-025-01864-9","url":null,"abstract":"","PeriodicalId":14691,"journal":{"name":"Japanese Journal of Radiology","volume":" ","pages":""},"PeriodicalIF":2.1,"publicationDate":"2025-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144954751","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":"Improvement of image quality of diffusion-weighted imaging (DWI) with deep learning reconstruction of the pancreas: comparison with respiratory-gated conventional DWI.","authors":"Kazuki Oyama, Fumihito Ichinohe, Yasuo Adachi, Yoshihiro Kito, Katsuya Maruyama, Minoru Mitsuda, Thomas Benkert, Omar Darwish, Yasunari Fujinaga","doi":"10.1007/s11604-025-01790-w","DOIUrl":"10.1007/s11604-025-01790-w","url":null,"abstract":"<p><strong>Purpose: </strong>This study aimed to evaluate the efficacy of deep learning-based reconstruction (DLR) in improving pancreatic diffusion-weighted imaging (DWI) quality.</p><p><strong>Materials and methods: </strong>In total, 117 patients (mean age of 68.0 ± 12.9 years) suspected of pancreatic diseases underwent magnetic resonance imaging (MRI) between July and December 2023. MRI sequences included respiratory-gated conventional diffusion-weighted images (RGC-DWIs), respiratory-gated diffusion-weighted images with deep learning-based reconstruction (DLR) (RGDLR-DWIs), and breath-hold diffusion-weighted images with DLR (BHDLR-DWIs) (short TE and long TE equal to other DWIs) at a 3 T MR system. Among these patients, 27 had solid lesions. Two radiologists qualitatively assessed pancreatic shape, main pancreatic duct (MPD) visualization, and solid lesion conspicuity using a 5-point scale. Quantitative analysis included apparent diffusion coefficient (ADC) values for pancreatic parenchyma and solid lesions, signal-to-noise ratio (SNR), pancreas-to-muscle signal-intensity ratio (PM-SIR) and lesion-to-pancreas signal-intensity ratio (LP-SIR). Differences among DWI sequences were analyzed using Friedman's and Bonferroni's tests.</p><p><strong>Results: </strong>Qualitatively, BHDLR-DWIs (short TE) had the highest scores for pancreatic shape and MPD but lowest for solid lesions visibility, whereas RGDLR-DWIs had the highest score for solid lesions. Quantitatively, BHDLR-DWIs (short TE) had the lowest ADC values for pancreatic parenchyma and solid lesions, with the highest PM-SIR. There was no significant difference between BHDLR-DWIs (short TE) and RGDLR-DWIs for solid lesion ADC values. RGC-DWIs had the highest SNR, though differences from RGDLR-DWIs and BHDLR-DWIs (short TE) were not significant. Although LP-SIR in RGDLR-DWIs were the lowest, the difference was not significant.</p><p><strong>Conclusion: </strong>BHDLR-DWIs (short TE) provided the best pancreatic morphology image quality, whereas RGDLR-DWIs were superior for solid lesion detection.</p>","PeriodicalId":14691,"journal":{"name":"Japanese Journal of Radiology","volume":" ","pages":"1509-1519"},"PeriodicalIF":2.1,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12396992/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144013692","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":"Altered connectivity among the triple brain networks in patients with mild cognitive impairment: a source-based morphometry study with a large elderly population.","authors":"Chihiro Yotsuya, Keita Watanabe, Sera Kasai, Yoshihito Umemura, Tomohiro Shintaku, Yuka Ishimoto, Miho Sasaki, Haruka Nagaya, Soichiro Tatsuo, Tatsuya Mikami, Yoshinori Tamada, Satoru Ide, Masahiko Tomiyama, Shingo Kakeda","doi":"10.1007/s11604-025-01791-9","DOIUrl":"10.1007/s11604-025-01791-9","url":null,"abstract":"<p><strong>Purpose: </strong>Previous research indicates brain network alterations in the default mode network (DMN), salience network (SN), and central-executive network (CEN) in individuals with mild cognitive impairment (MCI). However, replication has been inconsistent due to small, varied samples. We aimed to explore intra- and inter-networks alternations among DMN, SN, and CEN in individuals with MCI using multivariate source-based morphometry (SBM) with a larger, population-based sample.</p><p><strong>Materials and methods: </strong>This cross-sectional study included 1,997 participants (median age: 69 years; 61.9% female) who underwent three-dimensional (3D) T1-weighed imaging with 3 T magnetic resonance imaging (MRI). They were classified into 1,236 healthy controls (HC) and 761 individuals with MCI. SBM was used to extract triple brain networks as structural networks, and Z-scores were calculated. Intra-network comparisons of DMN, SN, and CEN between HC and MCI groups were conducted using logistic regression analysis. Inter-network comparisons among the triple brain networks were performed using structural equation modeling (SEM).</p><p><strong>Results: </strong>Connectivity (median Z score) of each network was lower in the MCI group than in the HC group: DMN (0.08 vs. - 0.12), SN (0.08 vs. - 0.23), and CEN (0.07 vs. - 0.06). Logistic regression showed significant association of SN connectivity with MCI (odds ratio 0.862, p < 0.05). SEM analysis revealed a significant group difference in the model where SN mediated input from CEN to DMN.</p><p><strong>Conclusion: </strong>We found altered network patterns in individuals with mild cognitive impairment, suggesting a transformation in network connectivity among DMN, SN, and CEN, particularly compensating for degraded SN connectivity.</p>","PeriodicalId":14691,"journal":{"name":"Japanese Journal of Radiology","volume":" ","pages":"1434-1444"},"PeriodicalIF":2.1,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12397200/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144325742","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":"Comparative performance of large language models in structuring head CT radiology reports: multi-institutional validation study in Japan.","authors":"Hirotaka Takita, Shannon L Walston, Yasuhito Mitsuyama, Ko Watanabe, Shoya Ishimaru, Daiju Ueda","doi":"10.1007/s11604-025-01799-1","DOIUrl":"10.1007/s11604-025-01799-1","url":null,"abstract":"<p><strong>Purpose: </strong>To compare the diagnostic performance of three proprietary large language models (LLMs)-Claude, GPT, and Gemini-in structuring free-text Japanese radiology reports for intracranial hemorrhage and skull fractures, and to assess the impact of three different prompting approaches on model accuracy.</p><p><strong>Materials and methods: </strong>In this retrospective study, head CT reports from the Japan Medical Imaging Database between 2018 and 2023 were collected. Two board-certified radiologists established the ground truth regarding intracranial hemorrhage and skull fractures through independent review and consensus. Each radiology report was analyzed by three LLMs using three prompting strategies-Standard, Chain of Thought, and Self Consistency prompting. Diagnostic performance (accuracy, precision, recall, and F1-score) was calculated for each LLM-prompt combination and compared using McNemar's tests with Bonferroni correction. Misclassified cases underwent qualitative error analysis.</p><p><strong>Results: </strong>A total of 3949 head CT reports from 3949 patients (mean age 59 ± 25 years, 56.2% male) were enrolled. Across all institutions, 856 patients (21.6%) had intracranial hemorrhage and 264 patients (6.6%) had skull fractures. All nine LLM-prompt combinations achieved very high accuracy. Claude demonstrated significantly higher accuracy for intracranial hemorrhage than GPT and Gemini, and also outperformed Gemini for skull fractures (p < 0.0001). Gemini's performance improved notably with Chain of Thought prompting. Error analysis revealed common challenges including ambiguous phrases and findings unrelated to intracranial hemorrhage or skull fractures, underscoring the importance of careful prompt design.</p><p><strong>Conclusion: </strong>All three proprietary LLMs exhibited strong performance in structuring free-text head CT reports for intracranial hemorrhage and skull fractures. While the choice of prompting method influenced accuracy, all models demonstrated robust potential for clinical and research applications. Future work should refine the prompts and validate these approaches in prospective, multilingual settings.</p>","PeriodicalId":14691,"journal":{"name":"Japanese Journal of Radiology","volume":" ","pages":"1445-1455"},"PeriodicalIF":2.1,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12396994/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144024519","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}