Radiological Physics and Technology最新文献

The gamma-ray dose rate distribution at the Kindai University Reactor (UTR-KINKI) was measured using the thermoluminescent (TL) properties of beryllium oxide (BeO) ceramic plates. The reactor, operating at an extremely low thermal power of 1 W, is widely used for nuclear research, including radiation biology and detector development. In neutron-gamma mixed fields, determining the gamma-ray dose rate accurately is technically challenging due to the neutron sensitivity of conventional dosimeters. In this study, low-Na BeO ceramic thermoluminescence dosimeters (TLDs) were employed to selectively measure gamma-ray dose rates in the irradiation hole of UTR-KINKI, without the need for neutron correction. A comparative assessment was conducted using Na-doped BeO powder TLDs, and thermal neutron flux measurements were performed using a Li-glass scintillator. The results demonstrated that the height-dependent trend of the gamma-ray dose rate distribution was consistent with previous measurements obtained via paired ionization chambers. However, the absolute values of the gamma-ray dose rates measured with the BeO ceramic TLDs were approximately 10-30% higher than those determined by the paired ionization chamber. This discrepancy is likely due to neutron sensitivity considerations in previous studies. The gamma-ray dose rate at the reactor center was evaluated as approximately 24 cGy h^-1. This study highlights the applicability of BeO ceramic TLDs for gamma-ray dosimetry in mixed radiation fields, offering a neutron-insensitive alternative for precise dose measurements in reactor environments.

Accurate estimation of absorbed doses in organs/tissues is essential for effective internal dosimetry. This is especially the case for positron-emission tomography-utilized radiopharmaceuticals that contain positron-emitting radionuclides. To achieve this, it is essential to calculate S-coefficients (S), basic coefficients representing the absorbed dose in the target organ per unit of nuclear transformation in the source organ. In addition, as the evolution of computational phantoms from stylized, voxelized, to mesh-type models continues, updating the S-coefficients to correspond with the new phantom generation becomes required. We employed the DoseCalcs Monte Carlo platform to estimate S-coefficients for four positron-emitting radionuclides, namely, C-11, N-13, O-15, and F-18. Based on decay and energy data for emitted positrons that were obtained from ICRP Publication 107, the simulations involved 24 regions as internal radiation sources in the male and female mesh-type phantoms of the International Commission on Radiological Protection (ICRP). We calculated the S-coefficients for 25 radiosensitive target regions. The graphs of S-coefficients for all target $\leftarrow$ source pairs exhibit similar trends for the four radionuclides. We compared the results with the OpenDose database, which calculated S-coefficients for voxelized phantoms. The comparison showed that the S-coefficients and the OpenDose voxelized values were very close for most target regions in the mesh-type phantoms. However, discrepancies were observed in specific cases, such as thyroid $\leftarrow$ UBCs and liver $\leftarrow$ HeW. These discrepancies arise primarily from the differences in organs/tissues locations and shapes, as well as the differences in material composition, which is distributed across the large inter-distance between the source and target, contributing to significant variations.

This study evaluated the dose calculation accuracy of a Monte Carlo (MC)-based independent dose calculation system (IDCS) for CyberKnife brain stereotactic treatment plans and compared it with ray-tracing (RT) and MC algorithms within the MultiPlan treatment planning system (TPS). Beam modeling accuracy was validated for 11 circular fields using measured output factors (OPF), percentage depth dose (PDD), and off-center ratio (OCR). A total of 200 retrospective brain stereotactic treatment plans were analyzed (50 prescribed 23 Gy in 1 fraction, 50 prescribed 35 Gy in 3 fractions, and 100 prescribed 41.5 Gy in 5 fractions). Among these, 24 quality assurance (QA) plans were evaluated using homogeneous cylindrical phantoms and ionization chambers. Dose-volume histogram (DVH) was calculated, and gamma analysis (3%/1 mm, 10% threshold) was performed. IDCS aligned with measured data, with OPF and PDD/OCR errors within 3% and 4%, respectively, except for small-field underestimations in the build-up region. For QA plans, TPS overestimated the measured dose (RT: 0.5% ± 2.6%, p = 0.58, MC: 1.7% ± 3.1%, p = 0.07), while IDCS underestimated it (- 1.3% ± 2.3%, p = 0.07). Gamma passing rates were 98.9% ± 1.5% (TPS-RT vs. IDCS) and 99.9% ± 0.3% (TPS-MC vs. IDCS). DVH metrics (planning target volume [PTV]: D_98%, D_95%, and D_2%) showed clinically acceptable differences. IDCS showed greater dose calculation accuracy than the TPS-RT algorithm and could identify dose discrepancies in specific cases, thereby confirming its reliability for CyberKnife QA.

This study aimed to propose a deep learning-based segmentation framework to delineate prostate lesions across multiple MRI acquisitions and derived parametric maps, including apparent diffusion coefficient (ADC) map, diffusion kurtosis imaging (DKI)-derived parameter maps (D map and K map), T2-weighted imaging (T2WI), and T2*-weighted imaging-derived parameter maps (T2* map and R2* map). Then, a comparison was conducted among the model's segmentation performance across MRI-derived images to identify those that provide the most discriminative information for accurate lesion identification. 51 patients underwent multiparametric MRI sequences, which included T2-weighted imaging (T2WI), diffusion-weighted imaging (DWI), and T2*-weighted images. Three expert radiologists conducted manual lesion annotations. All images were preprocessed, labeled, and augmented before training the U-Net++ model. The segmentation model's performance was evaluated using Dice similarity coefficient, Intersection over Union (IoU), sensitivity, and specificity metrics. The IoU values for the ADC map, D map, K map, T2WI, T2* map, and R2* map were 0.8907, 0.8559, 0.9504, 0.9250, 0.9441, and 0.8781, respectively. The corresponding Dice coefficient scores were 0.9416, 0.9211, 0.9744, 0.9604, 0.9709, and 0.9342. These results indicate a significant degree of overlap between the predicted and ground truth segmentation masks. These findings emphasize the complementary value of combining optimized deep learning architectures with advanced MRI-derived images, which could enhance diagnostic precision and facilitate more informed clinical decision-making.

This study quantitatively evaluated the impact of differences in computed tomography (CT) numbers and elemental compositions between commercially available brain-tissue-equivalent density plugs (BDPs) and actual brain tissue on dose calculations in a radiation therapy treatment planning system (RTPS). The mass density and elemental composition of BDP were analyzed using elemental analysis and X-ray fluorescence spectroscopy. The CT numbers of the BDP and actual brain tissue were measured and compared, with effective atomic numbers (EANs) calculated based on compositional analysis and the International Commission on Radiological Protection Publication 110 data for brain tissues. The theoretical CT numbers were derived using the stoichiometric CT number calibration (SCC) method. The dose calculations were performed using the modified CT number-to-relative electron density (RED) and mass density (MD) conversion tables in Eclipse v16.1, employing AAA and Acuros XB algorithms, employing the physical material table in AcurosXB_13.5. The dose metrics D_2%, D_50%, and D_98% were evaluated. Significant differences in elemental composition were found, particularly in carbon (73.26% in BDP vs. 14.3% in brain tissue) and oxygen (12.52% in BDP vs. 71.3% in brain tissue). The EANs were 6.6 for BDP and 7.4 for brain tissue. The mean CT numbers were 23.30 HU for the BDP and 37.30 HU for brain tissue, a 14 HU discrepancy. Nevertheless, dose calculation deviations were minimal, typically within ± 0.2%, with a maximum discrepancy of 0.6% for D_98%. Although CT numbers and elemental compositions exhibited notable differences, their impact on dose calculations in the evaluated RTPS algorithms was negligible.

Clinical measurement of the maximum slope (MS) using ultrafast dynamic contrast-enhanced (UF-DCE) breast magnetic resonance imaging (MRI) is typically performed by placing a region of interest (ROI) in the most enhanced area within a lesion. However, previous studies have not clarified whether visually identified enhanced areas consistently exhibit the highest MS values. These ROI-based MS measurements require MS maps to ensure appropriate ROI placement. However, generating MS maps requires specialized software capable of pixel-by-pixel MS calculations, which are available only at a few facilities. Therefore, this study proposed a simplified method for generating MS maps. This method involves subtracting consecutive UF-DCE images, applying temporal maximum intensity projection, normalizing the resulting image by dividing it by the pre-contrast image signal intensity, and converting it to a slope by dividing it by the temporal resolution. The MS maps generated using the proposed method were compared with those obtained using a robust pixel-by-pixel curve-fitting method, in addition to the final-phase UF-DCE images. In all cases with breast lesions (n = 13), the signal intensity distributions on the proposed MS maps closely resembled those on the curve-fitting maps, with a significantly higher similarity than those on the final-phase UF-DCE images (p < 0.001). The derived mean absolute error of MS values after regression-based modification was 0.78 ± 0.72 (%/s). The proposed method improves the reliability of ROI placement in conventional ROI-based MS measurements and supports the direct quantification of MS values from map pixel data.

The PET Imaging Site Qualification Program for amyloid positron emission tomography (PET) in Japan includes visual evaluation of the cylinder phantom. This visual evaluation requires observation of the entire image of the phantom and confirmation of the absence of apparent artifacts. Because the evaluation is visually performed, inter-observer differences might exist among evaluators for difficult cases. Therefore, the workload of the staff who perform approval tasks must be reduced, and objective evaluation methods are needed. Thus, we attempted to develop an artificial-intelligence-based objective method for anomaly detection. Three artificial intelligence methods for anomaly detection were developed, and their accuracy was evaluated using AutoEncoder, AnoGAN, and a method combining feature extraction using AlexNet and a one-class support vector machine. In total, 10,207 normal images from 128 facilities and 594 abnormal images from eight facilities, all of which were submitted as part of application for amyloid PET certification, were used. Group five-fold cross-validation was employed for artificial intelligence training and evaluation. In addition, the performance of each artificial intelligence method was assessed using receiver operating characteristic analysis. The areas under the curve for anomaly detection using AutoEncoder, AnoGAN, and the method combining feature extraction using AlexNet and a one-class support vector machine were 0.80 ± 0.04, 0.77 ± 0.03, and 0.99 ± 0.01, respectively. Artificial intelligence effectively distinguished between normal and abnormal images with high accuracy. In the future, its practical implementation is anticipated to reduce the workload in the approval work for the Japanese site qualification program for amyloid PET.

This study aims to compare and evaluate the potential benefits of using single DV-based, multiple DV-based physical cost function, and biological-based cost functions for organs at risk (OARs) sparing in IMRT as well as VMAT plans of head and neck cancer. Forty head and neck cancer patients treated with inverse plan optimization techniques were retrospectively enrolled for this study. Three different treatment plans were optimized by single DV-based, multiple DV-based physical cost functions, and biological-based cost functions on MONACO 6.1® TPS. All three optimized plans were normalized to deliver the same prescribed target dose. All 120 optimized plans were analyzed using dose evaluation parameters. For IMRT plans, the biological cost functions (BCF) were superior to both DV-based optimizations when it came to the mean dose of parallel organs. For VMAT plans, multiple DV-based physical cost function optimization resulted in a lower mean dose of parallel organs when compared with other two optimization. The biological cost function significantly reduced the mean dose of parallel organs, for which multiple DV-based cost functions were not used. In both IMRT and VMAT plans, the DV-based physical cost function significantly reduced the maximum dose of serial organs, with the exception of the mandible. Biological-based optimization made it more likely that the parallel OARs would be spared in IMRT plans, while multiple DV-based optimization made it more likely that the parallel OARs would be spared in VMAT plans. Both DV-based optimization in IMRT and VMAT plans effectively spared the maximum dose of the serial organ.

Magnetic resonance imaging (MRI) is time-consuming, posing challenges in capturing clear images of moving organs, such as cardiac structures, including complex structures such as the Valsalva sinus. This study evaluates a computed tomography (CT)-guided refinement approach for cardiac segmentation from MRI volumes, focused on preserving the detailed shape of the Valsalva sinus. Owing to the low spatial contrast around the Valsalva sinus in MRI, labels from separate computed tomography (CT) volumes are used to refine the segmentation. Deep learning techniques are employed to obtain initial segmentation from MRI volumes, followed by the detection of the ascending aorta's proximal point. This detected proximal point is then used to select the most similar label from CT volumes of other patients. Non-rigid registration is further applied to refine the segmentation. Experiments conducted on 20 MRI volumes with labels from 20 CT volumes exhibited a slight decrease in quantitative segmentation accuracy. The CT-guided method demonstrated the precision (0.908), recall (0.746), and Dice score (0.804) for the ascending aorta compared with those obtained by nnU-Net alone (0.903, 0.770, and 0.816, respectively). Although some outputs showed bulge-like structures near the Valsalva sinus, an improvement in quantitative segmentation accuracy could not be validated.

This study aims to develop a flexible Geant4 application capable of modeling all IEC 61267 defined radiation qualities for the HOPEWELL Designs 225 kV X-ray generator, while systematically analyze the impact of various environmental and systematic factors. Using Geant4, we replicated the experimental setup of the LEGEX laboratory and simulated all IEC 61267 radiation qualities by adjusting relevant beam parameters. The model was validated by comparing simulated HVLs and spectra, measured with a CdTe X-123 spectrometer against experimental data, SRS78 software results, and IEC reference values. The simulation demonstrated strong agreement with experimental measurements and published data, confirming the validity of our Geant4 application. We derived the function that characterizes the behavior of Kinetic Energy Released per unit Mass (KERMA) in response to variations in each influencing factor. Geometrical misalignment is the primary contributor to deviations, followed by aluminum purity and diaphragm movement, while environmental factors induced minor fluctuations. Additionally, we quantified backscattered radiation and applied corrective measures to eliminate its impact on measurements. The developed Geant4 application provides a reliable tool for simulating IEC 61267 radiation qualities and optimizing dosimetric accuracy. Our framework offers a cost-effective alternative to replicate different scenarios multiple times to identify and minimizes uncertainties.