Nihon Hoshasen Gijutsu Gakkai zasshi最新文献

Purpose: This study aimed to propose a mathematical model to describe motion blur in hepatobiliary-phase magnetic resonance images.

Methods: Simulated images were constructed based on the mathematical model of motion blur, and this model was then validated by comparing the simulated images with the magnetic resonance images obtained using a dynamic phantom. The mathematical model employed a convolution integral that combined the stationary-state signal distribution and the probability density distribution of the positional shifts during acquisition. The stationary signal distribution was obtained from the magnetic resonance images of a bottle phantom, and the probability density distribution was derived from sinusoidal motion mimicking respiratory movement. Both visual and quantitative evaluations were performed between the simulated images based on the mathematical model and the magnetic resonance images acquired using the dynamic phantom.

Results: The simulated and acquired magnetic resonance images exhibited similar shapes. The difference in the lengths of the respective high-intensity regions was 0.7 mm.

Conclusion: The proposed mathematical model effectively represents motion blur in magnetic resonance images.

Purpose: Accurate quantification of body fat distribution is essential in preventive medicine due to its public health outcomes. Specifically, visceral fat area measured from abdominal computed tomography (CT) serves as a key diagnostic criterion for obesity-related metabolic disorders. This study evaluated the efficacy of zero-shot segmentation capabilities of the Segment Anything Model (SAM) and its medical variant (MedSAM) for adipose tissue delineation on abdominal CT images.

Methods: Segmentation of the subcutaneous and visceral fat compartments was performed using point prompts on abdominal images. For a comprehensive evaluation, we compared the foundation models with a supervised U-Net architecture trained specifically for this task. The analysis utilized the publicly available annotated abdominal adipose tissue CT image dataset, which contained expert-annotated ground truth labels for adipose tissue compartments.

Results: The average Dice scores for visceral fat segmentation were 0.59 for SAM, 0.29 for MedSAM, and 0.93 for the U-Net. These findings indicate that while SAM achieved moderate performance, MedSAM exhibited limited effectiveness for this task. Both foundation models showed substantial performance gaps compared to the task-specific supervised approach.

Conclusion: The findings revealed limitations in the zero-shot segmentation capabilities of SAM and MedSAM for specialized medical imaging tasks, particularly for adipose tissue quantification in abdominal CT. Although foundation models offer theoretical advantages in generalizability and deployment efficiency, their performance in specialized medical applications requires significant enhancement through task-specific fine-tuning, and optimized prompt engineering strategies. Further studies are required to enhance the clinical utility of these models for reliable and accessible fat segmentation.

Purpose: We aimed to develop an AI-based system to score the positioning in mammography (MG), with the goal of establishing a foundation for future technical support.

Methods: Using 800 mediolateral oblique (MLO) images, we developed an AI model (Mask Generation Model) for automatic extraction of three regions: the pectoralis major muscle, the mammary gland region, and the nipple. Using this model, we extracted three regions from 1544 MLO images and generated mask images. The mask of the breast silhouette was automatically created using image processing techniques, resulting in a total of four mask images per image for 1544 MG. Afterward, we annotated those mask images and developed an AI model (Score Evaluation Model) for evaluating positioning with scores. The evaluation included six items: the insertion and shape of the pectoralis major muscle, the gap between the pectoralis major and the mammary gland, the extent of the mammary region, the breast silhouette, and the direction of stretching. We developed a system that quantifies MG positioning using AI-based scoring.

Results: For the mask generation model, the DICE coefficient, which indicates the concordance of region extraction, was 0.968. For the score evaluation model, the accuracy of the six models was 0.825 or higher, and the AUC values from the ROC analysis were 0.873 or higher, indicating high performance.

Conclusion: A system that evaluates positioning with a score has been developed. The scores assigned to the specific items can be utilized as quantitative indicators for future technical support.

Purpose: We developed software that can analyze central lung distance (CLD) and central skin distance (CSD) in deep inspiration breath hold (DIBH) irradiation and verified the reproducibility of the DIBH irradiation method in our hospital.

Methods: CLD and CSD were analyzed using In-house software using cine images acquired during treatment; intra- or inter-fractional motion was calculated from the obtained data and compared with other reports using systematic error (ΔΣ) and random error (Δσ).

Results: The ΔΣ and Δσ were 1.15 mm and 1.53 mm, respectively, which were the same as those reported in other reports.

Conclusion: The reproducibility of the irradiation method for left breast DIBH irradiation at our hospital is considered good. Due to its characteristics, regular and quantitative evaluation is required, and verification and evaluation are required at each facility.

Purpose: Typically, grades are compiled using spreadsheet software, and the file is then emailed with the password sent in a separate email. The government has pointed out that this method does not provide robust security. Introducing a new system is difficult due to the psychological factor of status quo bias. Therefore, we devised a new grade compilation system that uses spreadsheet software that can be easily implemented. The research aims to develop an automatic grade compilation system that implements RSA encryption.

Methods: As Excel's specifications do not allow for the display of more than 16 significant digits, we devised a method of applying modular exponentiation to modular arithmetic. The time required for encryption was measured five times and the average value was calculated.

Results: The time required for encryption was approximately 2.3 seconds, and RSA encryption is highly secure, it has the advantage of being able to distribute files with e-mail, and a paperless workflow could be established.

Conclusion: The developed system is highly secure and can be sent via email, and its automation is effective as a means of preventing disputes caused by human error.

Purpose: The objective of this study was to quantitatively evaluate and compare the image contrast and image quality of several recently developed CT-like images (oZTEo, LAVA, MERGE).

Methods: Using a self-made phantom created with a GE Healthcare (Milwaukee, WI, USA) MRI Pioneer 3.0T and pig leg bone, along with four types of tissue-mimicking phantoms, we compared cortical bone, contrast-to-noise ratio (CNR), signal change ratio of the tissue-simulating phantom, and visual evaluation.

Results: CR and CNR were lowest in the order of MERGE, LAVA, and oZTEo. The signal change ratio of background signal and visual evaluation were highest in the order of oZTEo, LAVA, and MERGE.

Conclusion: In comparing CT-like images using MRI, the bone cortex could be visualized in white using black-and-white inversion in all CT-like images, and fracture lines could also be visualized. By capturing CT-like images tailored to the characteristics of each imaging sequence, it is expected that the number of examinations that can be completed using MRI alone will increase, further enhancing the usefulness of MRI.

Purpose: The objective of this study was to determine the standard settings of the angiography system during EVT in Japan.

Methods: Survey requests were mailed to 493 facilities across Japan. The survey items included in the survey were system display dose setting of fluoroscopy dose (FL dose)/digital angiography dose (DA dose)/digital subtraction angiography dose (DSA dose), detector size, fluoroscopy pulse rate, DA/DSA frame rate, and standard additional filter settings. The number of valid responses for each item was system display dose setting: 187 systems (128 institutions), detector size: 187 systems (128 institutions), fluoroscopy pulse rate and DA/DSA frame rate: 185 systems (127 institutions), standard additional filter settings: 182 systems (125 institutions).

Results: Median (interquartile range) of each system display dose setting was FL: 5.9 (4.2-7.8) mGy/min, DA: 0.10 (0.06-0.16) mGy/frame, DSA: 0.82 (0.64-1.40) mGy/frame. The detector size for dose measurements was 103 systems (55%) with a diagonal of 35 cm or less and 84 systems (45%) with a diagonal of more than 35 cm. The most used FL pulse rate was 7.5 pulse/s (132 systems: 71%), DA frame rate was 7.5 frame/s (72 systems: 39%), and DSA frame rate was 3 frame/s (97 systems: 52%). More than 90% of the institutions used some kind of additional filter for imaging.

Conclusion: This survey showed the standard setting of the angiography system during EVT in Japan.