Kei Takase

Integrating spinal arterial anatomy with topographic MRI patterns and their underlying mechanisms facilitates accurate diagnosis and differentiation of spinal cord infarction from other myelopathies, ensuring prompt and appropriate clinical management.

Objective: Deep tumor invasion is a recognized pathologic risk factor for cervical lymph node metastasis in patients with hypopharyngeal carcinoma, but the relationship between imaging findings of the primary tumor and clinical lymph node metastasis has not been fully defined. In this study, we assessed whether imaging-defined deep invasion and tumor volume are associated with clinical lymph node metastasis. Methods: This retrospective study included 142 consecutive patients with newly diagnosed hypopharyngeal squamous cell carcinoma who underwent contrast-enhanced CT and MRI before treatment. Deep anterior wall (paraglottic space) invasion was graded according to tumor contact with the superior laryngeal neurovascular bundle. Deep posterior wall invasion was evaluated on MRI. Tumor diameter and volume were measured, and associations with lymph node status were assessed. Multivariable logistic regression was used to identify independent factors of clinical lymph node metastasis. Results: Clinical cervical lymph node metastasis was present in 78/142 patients (54.9%). The median tumor volume was greater in node-positive than in node-negative tumors (3.92 vs. 0.22 cm ³ , P P OR =2.42; 95% CI: 1.02-5.78; P =0.04) and tumor volume ( OR =1.32 per cm ³ ; 95% CI: 1.08-1.73; P =0.025) were independently associated with clinical cervical lymph node metastasis. Conclusion: Deep anterior wall invasion and larger tumor volume were associated with clinical cervical lymph node metastasis in hypopharyngeal carcinoma.

Purpose: To evaluate whether Radiology Information System (RIS) action log-guided continuous quality improvement (KAIZEN) improved CT operational efficiency. Materials and methods: This single-center retrospective before-and-after quality improvement study included 150,540 weekday CT examinations from 2018 to 2022. Five workflow interventions were introduced sequentially from January through February 2021. Daily median examination time, daily median patient entry/exit time, and the end time of the final weekday CT examination were analyzed using segmented interrupted time-series regression with interruptions in January 2020 and January 2021 and heteroskedasticity- and autocorrelation-consistent standard errors. Key performance indicator (KPI) models were adjusted for daily volume, day of the week, and inpatient proportion; unadjusted models were used as sensitivity analyses. Results: Daily median examination time decreased from 351 s at baseline to 328 s after KAIZEN. KAIZEN initiation was associated with an immediate reduction in adjusted examination time (− 17.37 s; 95% CI, − 31.35 to − 3.40 s; p = 0.015), whereas the unadjusted sensitivity analysis was directionally consistent but non-significant (− 11.07 s; p = 0.060). Entry/exit time decreased from 216 to 198 s and showed a significant adjusted immediate reduction (− 15.99 s; 95% CI, − 23.93 to − 8.04 s; p < 0.001), consistent with the unadjusted analysis. The final weekday CT examination ended 12.49 min earlier (95% CI, − 22.11 to − 2.87 min; p = 0.011). Conclusion: RIS action log-guided KAIZEN was temporally associated with shorter turnover, earlier completion of weekday operations, and reduced adjusted examination time. Replication in diverse settings is warranted. Secondary In this single-center study, RIS action log-guided KAIZEN was temporally associated with shorter CT patient entry/exit time, earlier weekday completion, and reduced adjusted examination time, supporting routine log-based workflow monitoring.

To assess differences and correlations between left ventricular myocardial strain measurements derived from cardiac CT (CCT) and cardiac MR (CMR) in patients with ischemic (IHD) and non-ischemic heart disease (NIHD). This single-center retrospective study included patients who underwent both full cardiac cycle CCT and CMR within 3 months. Global longitudinal (GLS), radial (GRS), and circumferential strain (GCS) were measured for both CCT and CMR. The mean difference with limits of agreement (LOA) and Pearson's correlation coefficient (r) with 95% confidence intervals (CI) were calculated. A prespecified margin of ± 1.5%, based on clinically acceptable differences, was used as a reference for assessing agreement. Sixty-three patients (mean age 68 ± 14 years; 70% male; 48% NIHD) were included. GLS and GCS from CCT showed minimal bias and good correlation with CMR (GLS, difference, -1.1% [LOA, -4.6%-2.4%]; r = 0.89 [95% CI 0.83-0.93]; GCS, difference, -1.2% [LOA, -7.4%-4.9%]; r = 0.78 [95% CI 0.66-0.86]). GRS from CCT showed slight overestimation compared to CMR (difference, 2.8% [LOA, -12.7%-18.3%]; r = 0.70 [95% CI 0.55-0.81]). The small bias in GLS was consistent between IHD and NIHD subgroups (-2.8% vs. 0.5%, p = 0.694), while GRS and GCS differences were more pronounced in the IHD subgroup (GRS, 5.2% vs. 0.1%, p = 0.014; GCS, -2.8% vs. 0.5%, p < 0.001). GLS and GCS from CCT showed good agreement with CMR, supporting their clinical feasibility. However, GRS and GCS measurements may be influenced by imaging protocols and disease backgrounds, necessitating standardized protocols for clinical application.

Background: The clinical use of coronary artery calcium (CAC) scoring for risk stratification in Japan is limited by the absence of population‐specific reference data, as applying Western‐derived thresholds is inappropriate because of known ethnic variations. This study aimed to establish the first comprehensive, age‐ and sex‐specific CAC reference values for a healthy Japanese population. Methods: In this single‐center retrospective study, we analyzed data from 4891 asymptomatic Japanese adults (63.2% men; median age, 58 years) without a history of atherosclerotic cardiovascular disease or diabetes. Age‐ and sex‐specific CAC percentile curves were generated using nonparametric regression modeling. Results: Men exhibited a higher CAC burden than women, with scores increasing with age in both sexes. The relationship between detectable CAC (CAC >0) and age was nonlinear: concave down for men and concave up for women, indicating different progression patterns. Compared with MESA (Multi‐Ethnic Study of Atherosclerosis) data, the Japanese cohort had a markedly lower CAC burden than White participants and systematically lower scores than Chinese American women. Conclusions: This study provides the first large‐scale, age‐ and sex‐specific CAC reference values for a healthy Japanese population. The generated percentile curves offer a practical tool for clinicians to immediately assess a patient’s CAC burden relative to their peers, underscoring that using foreign‐derived thresholds is inappropriate for risk stratification in Japan.

Background: Accurate esophageal cancer staging relies on 18F fluorodeoxyglucose positron emission tomography (18F FDG-PET), but its interpretation is complex and time-intensive. This diagnostic burden is exacerbated by significant workforce shortages in both radiology and surgery, thus necessitating automated support systems. The emergence of advanced large language models (LLMs) has raised expectations for their potential to fulfill this role in complex medical tasks. Objective: We evaluated the diagnostic accuracy of LLMs for staging esophageal cancer using 18F FDG-PET images, with a focus on their ability to assess lymph nodes (LNs; clinical N [cN]) and distant metastases (clinical M [cM]) for automated radiology reporting. Methods: This retrospective study included 120 consecutive adult patients who were diagnosed with esophageal squamous cell carcinoma and underwent 18F FDG-PET/computed tomography at Tohoku University Hospital between January 2019 and December 2021. Patients with prior treatment, nonsquamous cell carcinoma histology, or blood glucose levels ≥200 mg/dL were excluded. Frontal maximum-intensity projection positron emission tomography images were extracted, standardized, and analyzed along with information regarding the tumor location. Six LLMs (GPT-5, GPT-4.5, GPT-4.1, OpenAI-o3, -o1, and GPT-4 Turbo) and 4 blinded human evaluators (a nuclear medicine specialist, a gastrointestinal surgeon, and 2 radiology residents) assessed the presence of thoracic and abdominal LN metastases on a region-level basis and determined cN and cM staging on a patient-level basis. The model analyses were performed using the application programming interface in a zero-shot setting. Radiology reports served as the reference standard. Diagnostic agreement and accuracy were evaluated using Cohen κ and the Cochran Q test. Additionally, to account for the class imbalance in the dataset, the Matthews Correlation Coefficient was calculated as a robust metric for binary classification performance. Post hoc McNemar tests were performed with Bonferroni correction; statistical significance for pairwise comparisons was set at P<.0083 (adjusted from P<.05) using JMP Pro (version 18.0; SAS Institute Inc). Results: The average accuracy was 41/120 (34%) to 94/120 (78%) for LLMs and 72/120 (60%) to 102/120 (85%) for physicians, with significantly higher accuracy for physicians (P<.05) in the thoracic LN, abdominal LN, and cN stages. Interrater reliability was slight to fair for LLMs (κ: –0.07 to 0.25) and fair to substantial for physicians (κ: 0.27 to 0.74). Matthews Correlation Coefficient scores were consistently higher for physicians (0.28 to 0.75) than for LLMs (–0.07 to 0.32). Among the LLMs, GPT-5 demonstrated the highest overall accuracy, with newer LLMs showing improved diagnostic accuracy when compared with previous models in identifying abdominal LN metastases and cM staging, though they showed weaker consistency for cN staging. For example, in thoracic LN detection, GPT-5 achieved 76/120 (63%) accuracy, whereas other LLMs achieved 72/120 (60%) or lower accuracy. Conclusions: Although current LLMs have not yet reached physician-level accuracy in comprehensive staging, recent models show promise in assisting with specific diagnostic tasks.

【目的】CT検査における実効線量の推定には，dose length product（DLP）に変換係数を乗じる方法が広く用いられている．本研究では，患者の体型を考慮した新しい線量指標であるsize-specific dose length product（SS-DLP）を提案し，その有用性を従来のDLPと比較して検討することを目的とした．【方法】SS-DLPは，体型を考慮した線量指標である平均size-specific dose estimate（SSDE）にスキャン長を乗じて算出する指標として定義した．成人の胸部95例および腹部–骨盤部89例のCTスキャンデータを用い，線量計算ソフトウェアWAZA-ARI v2で実効線量をシミュレーションした．DLPおよびSS-DLPについて実効線量との回帰式を作成し，決定係数（R2），平均絶対誤差（MAE），Bland-Altmanプロットを用いて推定精度を評価した．また，BMIと推定誤差の関係も検証した．【結果】SS-DLPを用いることで，従来のDLPに比べてR2が向上し，MAEが減少した．Bland-Altmanプロットでは誤差範囲の縮小と平均誤差の改善が確認された．更に，SS-DLPを使用した場合，推定誤差のBMI依存性が軽減され，体型による誤差が抑えられることが示された．また，SS-DLPを用いた回帰式は，異なるスキャン部位間でも変換係数に高い一貫性を示し，スキャン部位や範囲の不確実性にも適応性が高いことが明らかとなった．【結語】SS-DLPは，CT検査における実効線量推定の精度を向上させる有用な指標である．この指標の活用により，体型を考慮した適切な線量管理が可能となり，放射線被ばく管理の更なる改善が期待される．

BACKGROUND Accurate esophageal cancer staging relies on 18F fluorodeoxyglucose positron emission tomography (18F FDG-PET), but its interpretation is complex and time-intensive. This diagnostic burden is exacerbated by significant workforce shortages in both radiology and surgery, thus necessitating automated support systems. The emergence of advanced large language models (LLMs) has raised expectations for their potential to fulfill this role in complex medical tasks. OBJECTIVE We evaluated the diagnostic accuracy of LLMs for staging esophageal cancer using 18F FDG-PET images, with a focus on their ability to assess lymph nodes (LNs; clinical N [cN]) and distant metastases (clinical M [cM]) for automated radiology reporting. METHODS This retrospective study included 120 consecutive adult patients who were diagnosed with esophageal squamous cell carcinoma and underwent 18F FDG-PET/computed tomography at Tohoku University Hospital between January 2019 and December 2021. Patients with prior treatment, nonsquamous cell carcinoma histology, or blood glucose levels ≥200 mg/dL were excluded. Frontal maximum-intensity projection positron emission tomography images were extracted, standardized, and analyzed along with information regarding the tumor location. Six LLMs (GPT-5, GPT-4.5, GPT-4.1, OpenAI-o3, -o1, and GPT-4 Turbo) and 4 blinded human evaluators (a nuclear medicine specialist, a gastrointestinal surgeon, and 2 radiology residents) assessed the presence of thoracic and abdominal LN metastases on a region-level basis and determined cN and cM staging on a patient-level basis. The model analyses were performed using the application programming interface in a zero-shot setting. Radiology reports served as the reference standard. Diagnostic agreement and accuracy were evaluated using Cohen κ and the Cochran Q test. Additionally, to account for the class imbalance in the dataset, the Matthews Correlation Coefficient was calculated as a robust metric for binary classification performance. Post hoc McNemar tests were performed with Bonferroni correction; statistical significance for pairwise comparisons was set at P<.0083 (adjusted from P<.05) using JMP Pro (version 18.0; SAS Institute Inc). RESULTS The average accuracy was 41/120 (34%) to 94/120 (78%) for LLMs and 72/120 (60%) to 102/120 (85%) for physicians, with significantly higher accuracy for physicians (P<.05) in the thoracic LN, abdominal LN, and cN stages. Interrater reliability was slight to fair for LLMs (κ: –0.07 to 0.25) and fair to substantial for physicians (κ: 0.27 to 0.74). Matthews Correlation Coefficient scores were consistently higher for physicians (0.28 to 0.75) than for LLMs (–0.07 to 0.32). Among the LLMs, GPT-5 demonstrated the highest overall accuracy, with newer LLMs showing improved diagnostic accuracy when compared with previous models in identifying abdominal LN metastases and cM staging, though they showed weaker consistency for cN staging. For example, in thoracic LN detection, GPT-5 achieved 76/120 (63%) accuracy, whereas other LLMs achieved 72/120 (60%) or lower accuracy. CONCLUSIONS Although current LLMs have not yet reached physician-level accuracy in comprehensive staging, recent models show promise in assisting with specific diagnostic tasks. CLINICALTRIAL

Purpose: In patients with esophageal squamous cell carcinoma (ESCC), it can be challenging to distinguish whether mediastinal lymph nodes (LNs) with an increased ¹⁸F-fluorodeoxyglucose (FDG) uptake are metastatic LNs or inflammatory ones. The aim of this study was to verify the superiority of dynamic ¹⁸F-fluorodeoxyglucose-positron emission tomography (FDG-PET) over conventional static positron emission tomography (PET) for detecting LN metastases in patients with ESCC using a silicon photomultiplier (SiPM) PET scanner. Patients and Methods: This prospective single-center study included 18 patients with ESCC who underwent dynamic PET/CT using SiPM PET and subsequently underwent radical esophagectomy and lymphadenectomy. FDG kinetics in the LNs were analyzed through spectral analysis and Patlak plots to calculate influx rate constant (Ki) values, followed by a conventional static PET scan and maximum standardized uptake value (SUVmax) measurement. Pathological analysis was used to define LN metastasis, and a receiver operating characteristic curve analysis was used to determine the diagnostic cutoffs for Ki and SUVmax. Results: Among 47 LNs (11 metastatic and 36 nonmetastatic) analyzed, metastatic LNs exhibited significantly higher Ki and SUVmax values than those of nonmetastatic LNs. The sensitivity for LN metastasis was comparable between SUVmax and Ki (both 72.7%). Meanwhile, the specificity in LN metastasis of Ki was higher than that of SUVmax (94.4% vs. 80.1%), as was the overall diagnostic accuracy of Ki (89.4% vs. 78.8%). Conclusions: Dynamic FDG PET/CT using SiPM PET may enhance the diagnostic accuracy for LN metastases in ESCC, enabling appropriate treatment intervention and improved outcomes for patients with ESCC.

Background: Artificial intelligence and large language models (LLMs)—particularly GPT-4 and GPT-4o—have demonstrated high correct-answer rates in medical examinations. GPT-4o has enhanced diagnostic capabilities, advanced image processing, and updated knowledge. Japanese surgeons face critical challenges, including a declining workforce, regional health care disparities, and work-hour-related challenges. Nonetheless, although LLMs could be beneficial in surgical education, no studies have yet assessed GPT-4o’s surgical knowledge or its performance in the field of surgery. Objective: This study aims to evaluate the potential of GPT-4 and GPT-4o in surgical education by using them to take the Japan Surgical Board Examination (JSBE), which includes both textual questions and medical images—such as surgical and computed tomography scans—to comprehensively assess their surgical knowledge. Methods: We used 297 multiple-choice questions from the 2021‐2023 JSBEs. The questions were in Japanese, and 104 of them included images. First, the GPT-4 and GPT-4o responses to only the textual questions were collected via OpenAI’s application programming interface to evaluate their correct-answer rate. Subsequently, the correct-answer rate of their responses to questions that included images was assessed by inputting both text and images. Results: The overall correct-answer rates of GPT-4o and GPT-4 for the text-only questions were 78% (231/297) and 55% (163/297), respectively, with GPT-4o outperforming GPT-4 by 23% (P=<.01). By contrast, there was no significant improvement in the correct-answer rate for questions that included images compared with the results for the text-only questions. Conclusions: GPT-4o outperformed GPT-4 on the JSBE. However, the results of the LLMs were lower than those of the examinees. Despite the capabilities of LLMs, image recognition remains a challenge for them, and their clinical application requires caution owing to the potential inaccuracy of their results.