Yuko Nakamura
Research Interests
Make your next move.
Prepare your application
Review drafts before you send them.
Explore related searches
Articles (13)
Year:
2026
Purpose To investigate the effect of multiscale sampling artificial intelligence (msAI) software adapted to small hepatic lesions on the diagnostic performance of readers interpreting gadoxetic acid-enhanced hepatobiliary-phase (HBP) images. Methods HBP images of 30 patients harboring 186 hepatic lesions were included. Three board-certified radiologists, 9 radiology residents, and 2 general physicians interpreted HBP image data sets twice, once with and once without the msAI software at 2-week intervals. Jackknife free-response receiver-operating characteristic analysis was performed to calculate the figure of merit (FOM) for detecting hepatic lesions. The negative consultation ratio (NCR), percentage of correct diagnoses turning into incorrect by the AI software, was calculated. We defined readers whose NCR was lower than 10% as those correctly diagnosed the false findings presented by the software. Results The msAI software significantly improved the lesion localization fraction (LLF) for all readers (0.74 vs 0.82, p < 0.01); the FOM did not (0.76 vs 0.78, p = 0.45). In lesion-size-based subgroup analysis, the LLF (0.40 vs 0.53, p < 0.01) improved significantly with the AI software even for lesions smaller than 6 mm, whereas the FOM (0.63 vs 0.66, p = 0.51) showed no significant difference. Among 10 readers with an NCR lower than 10%, not only the LLF but also the FOM were significantly better with the software (LLF 0.77 vs 0.82, FOM 0.79 vs 0.84, both p < 0.01). Conclusion The detectability of small hepatic lesions on HBP images was improved with msAI software especially when its results were properly evaluated.
Year:
2026
Purpose We developed an interventional radiology (IR) simulator using a virtual reality system (the VR–IR simulator) to teach IR procedures to medical students. In this study, we investigated the effectiveness of this teaching method. Materials and methods All ninety-nine fifth-year medical students attended a conventional classroom lecture. To teach students the actual procedure, they were randomly divided into two groups: One received conventional verbal explanations and educator demonstrations (the conventional group [ n = 44]), and the other received VR–IR simulator training (the VR–IR simulator group [ n = 55]). Afterward, they underwent a test using an augmented reality- (AR-) IR simulator (the VIST ® G5 image-guided AR–IR simulator, Mentice, Gothenburg, Sweden). The total procedure time, amount of contrast media used, fluoroscopic time, and patient peak skin dose in the simulated patients were compared between groups. A board-certified radiologist evaluated ten aspects of the procedure technique using a 5-point Likert scale (total: 50 points). Results Two students in the VR–IR simulator group were excluded due to VR sickness and simulator malfunction. There were no significant differences between the VR–IR simulator group and the conventional group regarding total procedure time (median [25–75% interquartile range]: 13.5 [11.8–14.5] vs. 14.3 [12.3–16.8] minutes, p = 0.11), fluoroscopic time (10.1 [8.5–13.0] vs. 11.0 [8.6–13.7] minutes, p = 0.31), and patient peak skin dose (276 [243–373] vs. 303 [239–395] mGy, p = 0.57), respectively. However, the amount of contrast media used was significantly lower (28.0 [21.0–36.2] vs. 40.0 [32.3–50.9] mL, p < 0.01) and the technical achievement scores by the radiologist (36 [34–44] vs. 31 [29–32], p < 0.01) were significantly higher in the VR–IR simulator group. Conclusion The VR–IR simulator helped reduce the amount of contrast media in interventional procedures and improved technical achievement scores.
Year:
2025
Background A deep learning-based image reconstruction (DLR) algorithm that can reduce the statistical noise has been developed for PET/CT imaging. It may reduce the administered dose of 18 F-FDG and minimize radiation exposure while maintaining diagnostic quality. This retrospective study evaluated whether the injected 18 F-FDG dose could be reduced by applying DLR to PET images. To this aim, we compared the quantitative image quality metrics and the false-positive rate between DLR with a reduced 18 F-FDG dose and Ordered Subsets Expectation Maximization (OSEM) with a standard dose. Results This study included 90 oncology patients who underwent 18 F-FDG PET/CT. They were divided into 3 groups (30 patients each): group A ( 18 F-FDG dose per body weight [BW]: 2.00—2.99 MBq/kg; PET image reconstruction: DLR), group B (3.00–3.99 MBq/kg; DLR), and group C (standard dose group; 4.00—4.99 MBq/kg; OSEM). The evaluation was performed using the signal-to-noise ratio (SNR), target-to-background ratio (TBR), and false-positive rate. DLR yielded significantly higher SNRs in groups A and B than group C ( p < 0.001). There was no significant difference in the TBR between groups A and C, and between groups B and C ( p = 0.983 and 0.605, respectively). In group B, more than 80% of patients weighing less than 75 kg had at most one false positive result. In contrast, in group B patients weighing 75 kg or more, as well as in group A, less than 80% of patients had at most one false-positives. Conclusions Our findings suggest that the injected 18 F-FDG dose can be reduced to 3.0 MBq/kg in patients weighing less than 75 kg by applying DLR. Compared to the recommended dose in the European Association of Nuclear Medicine (EANM) guidelines for 90 s per bed position (4.7 MBq/kg), this represents a dose reduction of 36%. Further optimization of DLR algorithms is required to maintain comparable diagnostic accuracy in patients weighing 75 kg or more.
Year:
2025
Objective: To evaluate whether our new protocol that uses a saline test injection and a leak detection sensor (LDS) reduces the frequency and amount of contrast media (CM) extravasation during the intravenous CM administration for CT. Methods: This retrospective study included 20,342 patients who underwent CECT at our hospital from March 2021 to November 2021 (old protocol, direct patient observation, and CM injection pressure monitoring, n=10,529) and from March 2024 to November 2024 (new protocol, old protocol plus saline test injection, and the LDS attachment, n=9813). We compared the frequency and the volume of extravasation between the 2 protocols using the Fisher exact test and the Mann-Whitney U test. We also evaluated the accuracy of the LDS. Results: Extravasation occurred in 51 patients (age 72.1±12.2 y, 33 men) under the old protocol and in 26 patients (age 73.6±9.0 y, 17 men) with the new protocol. The overall frequency of extravasation and the number of patients with an extravasation volume of at least 20 mL were significantly lower with the new protocol than the old protocol (0.48% vs. 0.26%; 0.16% vs. 0.03% all, P <0.01). The extravasation volume was significantly reduced with the new protocol (14 vs. 6 mL, P <0.01). The sensitivity of the LDS to detect extravasation of 3, 5, 10, and 15 mL was 50%, 88%, 93%, and 100%, respectively; specificity was 99% for all. Conclusions: Our new protocol reduced the frequency and dose of CM extravasation.
Year:
2025
Purpose Artificial intelligence (AI) algorithms for lung nodule detection assist radiologists. As their performance using ultra-high-resolution CT (U-HRCT) images has not been evaluated, we investigated the usefulness of 0.25-mm slices at U-HRCT using the commercially available deep-learning-based lung nodule detection (DL-LND) system. Materials and methods We enrolled 63 patients who underwent U-HRCT for lung cancer and suspected lung cancer. Two board-certified radiologists identified nodules more than 4 mm in diameter on 1-mm HRCT slices and set the reference standard consensually. They recorded all lesions detected on 5-, 1-, and 0.25-mm slices by the DL-LND system. Unidentified nodules were included in the reference standard. To examine the performance of the DL-LND system, the sensitivity, and positive predictive value (PPV) and the number of false positive (FP) nodules were recorded. Results The mean number of lesions detected on 5-, 1-, and 0.25-mm slices was 5.1, 7.8 and 7.2 per CT scan. On 5-mm slices the sensitivity and PPV were 79.8% and 46.4%; on 1-mm slices they were 91.5% and 34.8%, and on 0.25-mm slices they were 86.7% and 36.1%. The sensitivity was significantly higher on 1- than 5-mm slices ( p < 0.01) while the PPV was significantly lower on 1- than 5-mm slices ( p < 0.01). A slice thickness of 0.25 mm failed to improve its performance. The mean number of FP nodules on 5-, 1-, and 0.25-mm slices was 2.8, 5.2, and 4.7 per CT scan. Conclusion We found that 1 mm was the best slice thickness for U-HRCT images using the commercially available DL-LND system.
Year:
2025
Advancements in computed tomography (CT) technology, particularly the emergence of dual-energy CT (DE-CT) and photon-counting detector CT (PCD-CT), can improve detection, characterization, and treatment monitoring of focal liver lesions. DE-CT, through its ability to differentiate tissues with similar densities and produce diverse datasets, has enhanced lesion visibility and diagnostic precision. PCD-CT further advances imaging with superior spatial resolution and material decomposition capabilities, offering potential for complex diagnostic scenarios. This review aimed to highlight the role of CT in hepatic imaging and its application to focal liver lesions. DE-CT improves lesion detectability using low-energy virtual monochromatic images, which enhance iodine contrast and reduce radiation and contrast agent doses. It also facilitates treatment response evaluation after locoregional therapies for hepatocellular carcinoma by quantifying biomarkers, such as the extracellular volume fraction. This review underscores the transformative impact of DE-CT and PCD-CT on liver imaging, emphasizing their complementary roles alongside magnetic resonance imaging. These innovations have paved the way for more precise diagnostics, improved treatment planning, and enhanced patient outcomes in the management of liver diseases.
Year:
2025
Background Low-keV virtual monoenergetic images (VMIs) of dual-energy computed tomography (CT) enhances iodine contrast for detecting small arteries like the Adamkiewicz artery (AKA), but image noise can be problematic. Deep-learning image reconstruction (DLIR) effectively reduces noise without sacrificing image quality. Purpose To evaluate whether DLIR on low-keV VMIs of dual-energy CT scans improves the visualization of the AKA. Material and Methods We enrolled 29 patients who underwent CT angiography before aortic repair. VMIs obtained at 70 and 40 keV were reconstructed using hybrid iterative reconstruction (HIR), and 40 keV VMIs were reconstructed using DLIR. The image noise of the spinal cord, the maximum CT values of the anterior spinal artery (ASA), and the contrast-to-noise ratio (CNR) of the ASA were compared. The overall image quality and the delineation of the AKA were evaluated on a 4-point score (1 = poor, 4 = excellent). Results The mean image noise of the spinal cord was significantly lower on 40-keV DLIR than on 40-keV HIR scans; they were significantly higher than on 70-keV HIR images. The CNR of the ASA was highest on the 40-keV DLIR images among the three reconstruction images. The mean image quality scores for 40-keV DLIR and 70-keV HIR scans were comparable, and higher than of 40-keV HIR images. The mean delineation scores for 40-keV HIR and 40-keV DLIR scans were significantly higher than for 70-keV HIR images. Conclusion Visualization of the AKA was significantly better on low-keV VMIs subjected to DLIR than conventional HIR images.
Year:
2024
Year:
2024
Background Recent guidelines recommend direct patient observation, pressure monitoring, and sensor devices to prevent extravasation during contrast media (CM) injection. However, it is impractical in terms of time and cost to install sensors for all patients. Purpose To identify risk factors for CM extravasations during contrast-enhanced computed tomography (CECT) in a large population and to establish criteria for placing the sensor device on patients. Material and Methods This retrospective study included 143,556 patients who underwent CECT at our hospital between April 2012 and July 2022. We performed multivariable logistic regression analysis between patients with (n = 350) and randomly selected patients without CM extravasation (n = 350). We investigated the percentage of patients with sensor devices and their sensitivity for detecting extravasation using receiver operating characteristic curve analysis. Results The extravasation rate was 0.27%. Multivariable logistic regression analysis showed that the injection rate (adjusted odds ratio [AOR] = 1.61, 95% confidence interval [CI] = 1.33–1.95: P <0.001), catheter gauge (AOR = 3.86, 95% CI = 1.92–7.76; P <0.001), the use of anticancer drugs (AOR = 1.81, 95% CI = 1.32–2.50; P <0.001), and existing catheters (AOR = 1.52, 95% CI = 1.10–2.11; P = 0.009) were significantly associated with extravasation. To achieve a sensitivity of 90%, 80%, 70%, 60%, and 50%, 80%, 65%, 50%, 40%, and 28% of all patients required the placement of a sensor device, respectively. Conclusion Sensitivity analysis established criteria for effective placing sensor devices.
Year:
2024
Year:
2024
Year:
2024
Year:
2024
Collaborators (4)
Yoshitaka Bito
Hokkaido University
Yukiko Honda
Hiroshima University Graduate School of Biomedical and Health Sciences
Hirotaka Inoue
Professor
National Institute of Technology, Kure College
Shintaro Ichikawa
Associate Professor
Hamamatsu University School of Medicine
Make your next move.
Prepare your application
Review drafts before you send them.