Radiological Physics and Technology最新文献

This phantom study compared breast-level surface dose and image noise between single-energy CT (SECT) and fast kilovoltage-switching dual-energy CT (DECT) under matched volume CT dose index (CTDIvol) conditions. A chest phantom was scanned using a 256-row CT scanner. Surface dose was measured with real-time dosimeters placed on the bilateral anterior chest wall at the breast level and corresponding bilateral posterior positions. Tube currents of 195 mA for SECT and 250 mA for DECT provided matched console-displayed CTDIvol values. Each protocol was repeated 20 times as technical replicates. Image noise was measured in five circular regions of interest at the bilateral anterior and posterior chest walls and heart. Mean surface dose was lower with DECT than SECT at the breast level (15.86 vs. 17.02 mGy; mean difference [DECT minus SECT], -1.16 mGy; 95% confidence interval [CI], -1.71 to -0.61; P < 0.001) and posterior surface (15.77 vs. 17.26 mGy; mean difference, -1.49 mGy; 95% CI, -2.21 to -0.77; P < 0.001). Mean image noise was 7.08 HU with DECT and 6.66 HU with SECT, with no statistically significant difference (mean difference, 0.42 HU; 95% CI, -0.17 to 1.01; P = 0.160). Under the tested conditions, DECT showed lower anterior and posterior surface doses than SECT performed without CT automatic exposure control. No statistically significant difference in image noise was observed, but this does not establish equivalent image quality. These findings should not be generalized to clinically optimized SECT protocols.

Current quality assurance phantoms often lack MR signal, deformability, or real-time dosimetry capability, limiting comprehensive end-to-end evaluation for MR-guided online adaptive radiotherapy (MRgOART), including verification of deformable image registration (DIR) accuracy and the adaptive workflow within a single phantom. This study aimed to develop a CT/MRI-compatible anthropomorphic, deformable male pelvic phantom for a comprehensive end-to-end test of MRgOART. Reference CT value and MR signal intensity of the prostate, bone, and urine were obtained from a cohort of 30 prostate cancer patients who underwent treatment with a 0.35 T MR-Linac. MR images were acquired using a TRUFI (balanced steady-state free precession) sequence. Candidate materials were optimized to reproduce patient-like CT and MR signal characteristics, and their temporal stability was evaluated over 6 months. Relative electron density (rED) was validated using elemental analysis to ensure dose-calculation reliability. The phantom incorporated deformable prostate, rectum, and bladder models and allowed insertion of an ionization chamber for real-time internal dose measurement. The optimized materials reproduced CT value and MR signal intensity within the mean ± 1 standard deviation of patient reference data, providing clear image contrast without artifacts. CT values remained stable over time, while minor fluctuations in MR signal intensity remained within the range of physiological variability. Composition-based rED values showed small deviations from those derived by CT, which were deemed acceptable for the end-to-end test. The developed phantom enables realistic, multimodal, and deformable end-to-end evaluation of MRgOART, supporting accurate assessment of DIR, adaptive planning, and dose delivery with real-time dosimetry.

This study evaluated the effects of acquisition parameters on relaxation times measured using magnetic resonance fingerprinting (MRF). MRF enables the simultaneous acquisition of multiple quantitative maps by varying the flip angle (FA) and repetition time (TR) in a pseudo-random manner and by pattern-matching signal evolutions to a pre-calculated dictionary. To assess parameter dependency, multiple acquisition protocols were generated with varying numbers of echoes (500, 1000, 1500, and 3000), TR widths (12-14 ms, 12-17 ms, and 12-20 ms), and maximum FA ranges using the MRF Development Kit. The International Society for Magnetic Resonance in Medicine/National Institute of Standards and Technology system phantom was scanned ten times per condition on a 3-Tesla scanner. Coefficients of variation (CVs) and agreement were calculated for T1 and T2 values. Results indicated that 1500 and 3000 echoes produced comparable agreement and reproducibility (agreement within ± 7.77% and CVs ≤ 1.13% for both T1 and T2). At 500 echoes, CVs increased substantially and agreement decreased, with T2 values being more affected than T1 values. At 1000 echoes with an expanded TR width (12-20 ms), CVs were 1.15% or less and agreement was within ± 5.20%, comparable to higher echo counts while reducing scan time by approximately 20%. This study demonstrated that the number of echoes, TR width, and maximum FA significantly influence T1 and T2 agreement and reproducibility, suggesting that expanding the TR range can stabilize relaxation times even with fewer echoes.

The objective of this study is to perform an end-to-end (E2E) technical validation of a total body irradiation (TBI) workflow using an anthropomorphic phantom, integrating simulation, treatment planning, patient-specific quality assurance, and phantom-based dosimetry to ensure accurate and reproducible dose delivery. A full E2E test was performed using the Rando Phantom and Varian Eclipse TPS. CT scans were acquired in head-first supine (HFS) and feet-first supine (FFS) orientations and registered to simulate full-body imaging. Treatment plans were optimized for both orientations, with specific gradient dose regions defined for overlap zones. ArcCHECK and OSLDs were used for pre-treatment and phantom-embedded dosimetric verification, respectively. Cone Beam CT (CBCT) assessed positional accuracy across eight treatment fractions. Dose discrepancies and alignment errors were evaluated across anatomical regions. HU comparison between the CT console and TPS revealed only minor deviations, all within ± 6 HU, supporting the consistency of HU representation in the TPS. Gamma passing rates exceeded 97% across most regions. OSLD dose differences remained within ± 2.5% for most sites, except thorax and overlap zones, which reached ± 4.8%. CBCT-based setup verification demonstrated sub-millimeter alignment deviations across fractions, indicating consistent positional accuracy. E2E validation using the Rando Phantom supports technical accuracy and consistency in the TBI workflow. Notable dose variations in overlap and heterogeneous regions highlight the need for enhanced QA in complex geometries. The findings support the technical feasibility and reproducibility of the workflow, while further patient-specific and motion-inclusive studies are required to establish broader clinical applicability.

This study aimed to develop and validate an automated magnetic resonance imaging (MRI)-based pipeline for temporal classification of intracerebral hemorrhage (ICH) using clinically feasible multi-sequence MRI, 2.5D U-Net segmentation, and machine learning. This retrospective study included a development cohort of 56 patients and an independent validation cohort of 115 patients with ICH. Five MRI sequences-diffusion-weighted imaging, apparent diffusion coefficient map, fluid-attenuated inversion recovery (FLAIR), time-of-flight magnetic resonance angiography, and susceptibility-weighted angiography-were analyzed. A pontine reference region of interest was automatically placed using FLAIR-based brain segmentation. Hematoma regions were automatically segmented using a 2.5D U-Net trained in the development cohort, and the largest three-dimensional connected component was defined as the representative hematoma region of interest. Pontine ROI-referenced signal features and shape features were extracted, and Random Forest, XGBoost, and multinomial logistic regression classifiers were compared for five-stage temporal classification. The selected final model was applied to the validation cohort. In the validation cohort, automated hematoma segmentation achieved a Dice similarity coefficient of 0.794 ± 0.208, sensitivity of 0.799 ± 0.235, and precision of 0.842 ± 0.151. One case with automated hematoma ROI extraction failure was excluded from temporal classification analysis. In the remaining 114 cases, the final Random Forest model achieved an accuracy of 0.807, balanced accuracy of 0.763, macro F1-score of 0.749, and mean Average Precision of 0.851. The proposed automated MRI-based pipeline showed feasible performance for ICH temporal classification and may support objective MRI-based assessment of hemorrhage stage.

This study evaluates organ doses from ¹⁸F-FDG whole-body PET-CT using the phantom-based software PARaDIM and patient image-based software RT-PHITS and compares with reference data from ICRP Publication 128. Absorbed dose coefficients (mGy/MBq) were calculated using ICRP Publication 145 mesh-type phantoms in PARaDIM and 20 patient datasets (10 male, 10 female) from 'The Cancer Imaging Archives' website in RT-PHITS. Organs were segmented in 3D Slicer, and dose coefficients were derived after physical decay correction of the resultant dose rates (mGy/s). The patient cohort (n = 20; 10 males, 10 females) had an average age of 52.1 years, height of 172.1 cm, weight of 76.65 kg, and total administered activity of 306.55 MBq. Absorbed dose coefficients from PARaDIM and RT-PHITS ranged from 0.01 to 0.07 mGy/MBq. RT-PHITS showed acceptable agreement with ICRP 128 and PARaDIM for most organs, except for the urinary bladder and heart wall. Standard deviations (SD) for RT-PHITS urinary bladder doses were 0.04 relative to ICRP 128 and 0.12 relative to PARaDIM; heart wall SDs were 0.016 and 0.014, respectively. Overall, RT-PHITS organ doses yielded an average SD of 0.0073, with the highest deviation in the heart wall. The results indicate that anatomical differences lead to measurable variations between the generalized and individualized dosimetry methodologies. Because a consistent physical decay model of ¹⁸F was used, these variations primarily reflect structural and geometric factors. This study provides a framework for advancing individualized dosimetry approaches and supports future developments in cumulative exposure assessments and potential theranostic applications.

This study investigated whether irradiation-induced chemical changes in sucrose solutions could be utilized to control magnetic resonance imaging (MRI) relaxation properties and to propose a novel design strategy for quantitative MRI phantoms. Aqueous sucrose solutions at concentrations of 5-40% (w/w) containing 3.6 g/L NaCl were irradiated with 6 MV photons at doses of 0-20,000 monitor units. T₁ and T₂ relaxation times were measured at 1.5 T. Hydrogen peroxide (H₂O₂) and glucose concentrations were quantified to assess the radiolytic products. Linear regression and Type III ANOVA were used to evaluate the effects of sucrose concentration and irradiation dose. The T₂ relaxation times increased with increasing irradiation dose (p < 0.001), with a significant interaction between sucrose concentration and dose (p < 0.001). The slope of T₂ monotonically decreased with increasing sucrose concentration. R₂ (defined as 1/T₂) showed a significant linear dose-dependence at all concentrations, with the absolute magnitude of the negative regression coefficients increasing at higher sucrose concentrations, which was consistent with the reciprocal relationship between T₂ and R₂. In contrast, both T₁ and R₁ (defined as 1/T₁) exhibited minimal dose-dependence. The H₂O₂ concentration increased stepwise with the dose and demonstrated a strong positive correlation with T₂ (Spearman's ρ = 1.00), whereas glucose showed no consistent dose-dependence. Irradiation enabled concentration-dependent modulation of T₂ relaxation in sucrose solutions without altering the initial composition. This externally controllable approach provides a simple and flexible framework for designing quantitative MRI phantoms with tunable relaxation properties and offers a new paradigm beyond conventional composition-based phantom fabrication.

A measurable fraction of examinations in retrospective imaging archives lack a recorded body weight, limiting metadata completeness for tasks such as cohort auditing, weight-stratified analysis, and the assembly of machine-learning training datasets. We developed and compared three approaches for body-weight prediction from a single middle-slice CT image: a LightGBM model on 19 CT-derived tabular features (water equivalent diameter, effective diameter, body composition, scanner parameters), an ImageNet-pretrained ResNet-18 on images alone, and a ResNet-18 with late fusion of the same 19 tabular features into the 512-dimensional image embedding. Models were trained with 5-fold patient-disjoint cross-validation on a combined train+validation development pool of 30,857 examinations from 16,169 patients; the best-performing fold of each model family was selected by within-fold validation mean absolute error (MAE) and evaluated, without further tuning, on a held-out test partition of 3,479 examinations from 1,796 patients (patient-disjoint from training and validation). On the test set, the fusion ResNet-18 achieved MAE 4.06 kg (95% CI 3.90-4.23; R² = 0.92), with 69.5% of predictions within ± 5 kg and 93.8% within ± 10 kg. A weighted ensemble of the fusion ResNet and LightGBM produced a small further reduction in MAE to 3.99 kg (paired bootstrap p < .001), but the 0.07 kg difference is below practically meaningful thresholds. The simpler single-model ResNet-18 with metadata fusion may be useful for retrospective missing-weight imputation, with split-conformal 90% and 95% prediction intervals (empirical test coverage 89% and 94%) available for case-level uncertainty quantification.

Conventional univariate feature-selection pipelines for radiation pneumonitis (RP) prediction are prone to overfitting in small radiotherapy cohorts, while alternative dimensionality-reduction approaches remain insufficiently evaluated. This study compared a Principal Component Analysis (PCA)-based ensemble pipeline comprising variance pre-filtering, PCA, and LASSO with a conventional pipeline comprising Mann-Whitney U testing, Spearman correlation filtering, and ElasticNet. The analysis included 73 retrospectively analyzed patients with stage III non-small cell lung cancer (training, n = 52; testing, n = 21). A total of 214 features, including 107 radiomic and 107 dosiomic features, were extracted from planning CT images and three-dimensional dose distributions. Both pipelines used five-model ensemble logistic regression with inverse-frequency weighting. Performance was evaluated using testing AUC, the training-to-testing AUC gap, 10-fold nested cross-validation, paired bootstrap comparison of AUC differences with 2,000 resamples, Brier score, and bootstrap feature-selection stability. The PCA-based model achieved a testing AUC of 0.765 with a training-to-testing AUC gap of 0.020, compared with an AUC of 0.684 and a gap of 0.181 for the conventional model (ΔAUC = + 0.081, p = 0.265). The PCA-based model also achieved a nested cross-validation AUC of 0.742. PCA-based dimensionality reduction produced a smaller training-to-testing AUC gap and greater feature-selection stability than conventional univariate filtering, indicating improved internal robustness and reduced overfitting in this small-cohort setting.

To evaluate whether carbon dioxide (CO₂) insufflation modifies the relationship between scan delay and mesenteric venous attenuation during fixed-delay portal venous phase scans in contrast-enhanced CT colonography (CTC). This retrospective study included 170 patients (mean age, 66.6 ± 13.0 years; 72 men) who underwent contrast-enhanced abdominal CT between April 2017 and October 2021 with portal venous phase scans acquired at fixed delays of 25, 30, or 35 s after the start of arterial-phase scanning. Examinations were classified as the CO₂ insufflation group (n = 86) or the non-insufflation group (n = 84). Attenuation of the inferior and superior mesenteric veins (IMV, SMV) was measured at each delay. Associations among insufflation status, scan delay, and venous attenuation were evaluated using multivariable linear regression including a CO₂-by-delay interaction term. In the CO₂ insufflation group, IMV attenuation was highest at 25 s (283.0 ± 40.4 HU) and decreased at 30-35 s (p < 0.001), whereas in the non-insufflation group IMV attenuation was higher at 30-35 s than at 25 s (p = 0.002). SMV attenuation showed a similar directional pattern with smaller magnitude. Significant CO₂-by-delay interactions were observed for both veins (p < 0.05), indicating different delay-attenuation relationships by insufflation status. In routine fixed-delay CTC, CO₂ insufflation was associated with a modified relationship between scan delay and mesenteric venous attenuation, most prominently in the IMV. Mesenteric venous enhancement on portal venous phase scans should be interpreted in the context of insufflation status.