Physics in medicine and biology最新文献

Objective Previous MR guided radiation therapy (MRgRT) radiofrequency coil arrays have been limited to one to two rows of coils in the head-foot direction because of the desire to place radiation-opaque coil circuitry outside the window through which the radiation beam travels. However, such layouts limit parallel imaging undersampling in the head-foot direction. We recently demonstrated a three-row array with a remote coil circuit that improved parallel imaging performance, while preserving the signal-to-noise ratio (SNR) and the radiolucent window. Here we evaluate a four-row prototype design to determine if further parallel imaging advantages could be realized. Approach We built remote circuits that allowed radio-opaque components to be placed outside the field of view through which the radiation beam is expected to travel. The circuit consisted of a phase shifter to cancel the phase introduced by the radiolucent coaxial link between the circuit and coil, followed by standard components for tuning, matching, detuning, and preamplifier decoupling. Measurements were performed on an abdominal phantom to compare single-channel coils with remote or local circuits, followed by tests on a 16-channel four-row array. Main results The four-row array maintained SNR comparable to two-and three-row designs while supporting 3× head-foot acceleration (minimum reciprocal g-factor = 0.74) and 2×3 multi-directional acceleration (minimum reciprocal g-factor = 0.72), capabilities which were not achievable with previous designs. Significance These results demonstrate the technical feasibility of four-row designs, which may benefit MRgRT applications that require high SNR and temporal-resolution.

Objective: Radiotherapy (RT) requires accurate and consistent patient positioning to ensure precise radiation delivery and minimize unnecessary exposure to healthy tissues. Conventional workflows rely heavily on clinicians' experience and repeated CT-based registration, leading to inefficiency, patient discomfort, and potential alignment inconsistencies. This work aims to develop an automatic, robust, and low-cost posture offset detection method that overcomes these limitations.

Approach: We propose a prompt-guided incremental fine-tuning model built upon a large-scale image segmentation backbone. The system captures real-time 2D images from a single RGB camera and automatically generates adaptive prompt points based on individual body shapes and postures, improving segmentation robustness and reducing environmental interference. An incremental fine-tuning strategy enables continuous adaptation to newly collected patient images throughout the treatment cycle. Furthermore, a multi-level offset analysis framework is introduced, integrating contour-level, keypoint-level, and pixel-level estimations to identify, localize, and quantify posture deviations across multiple granularities. The system is deployed clinically to collect real RT data and construct a dedicated validation dataset.

Main results: Extensive experiments on real clinical data show that the proposed method achieves accurate, fast, and stable posture offset detection. It substantially improves positioning consistency and efficiency compared with conventional workflows. Ablation studies further demonstrate the effectiveness and necessity of each module within the framework.

Significance: This study provides a practical and low-cost solution for RT positioning, reducing clinician workload and patient burden while improving treatment accuracy. It demonstrates the potential of prompt-guided incremental adaptation and multi-level offset analysis in real RT environments, offering a promising direction for intelligent, automated radiotherapy positioning systems.

Objective.As part of treatment planning for radiotherapy, the organs at risk (OARs) are delineated on the patient's CT scan. This work aims to develop a method to measure variability in OAR delineations and detect errors.Approach.A normative modelling approach was implemented by training a variational autoencoder (VAE) on a dataset of images and delineations to model the "acceptable" variability distribution. The trained VAE was then used to reconstruct unseen cases. Disagreements between input and reconstructed delineations highlighted regions where the input deviated from the training distribution. This approach was validated by evaluating the reconstructions of spinal cord and brainstem delineations where common clinical errors had been introduced.Main results.Results showed that the model successfully detected errors, even when only a few voxels or slices were added or removed. Distance to agreement maps were generated to quantify the magnitude of the disagreements in misclassified regions. These results were further validated by manually evaluating some of the test cases.Significance.This tool has the potential of assisting clinicians in reviewing and validating OAR delineations.

Objective: Particle therapy relies on up-to-date knowledge of the stopping power of the patient tissues to deliver the prescribed dose distribution. The stopping power describes the average particle motion, which is encoded in the distribution of prompt-gamma photon emissions in time and space. We reconstruct the spatiotemporal emission distribution from multi-detector Prompt Gamma Timing (PGT) data. Solving this inverse problem relies on an accurate model of the prompt-gamma transport and detection including explicitly the dependencies on the times of emission and detection.

Approach: Our previous work relied on Monte-Carlo (MC) based system models. The tradeoff between computational resources and statistical noise in the system model prohibits studies of new detector arrangements and beam scanning scenarios. Therefore, we propose here an analytical system model to speed up recalculations for new beam positions and to avoid statistical noise in the model.

Main results: We evaluated the model for the MERLINO multi-detector-PGT prototype. Comparisons between the analytical model and a MC-based reference showed excellent agreement for single-detector setups. When several detectors were placed close together and partially obstructed each other, intercrystal scatter led to differences of up to 10 % between the analytical and MC-based model. Nevertheless, when evaluating the performance in reconstructing the spatiotemporal distribution and estimating the stopping power, no significant difference between the models was observed. Hence, the procedure proved robust against the small inaccuracies of the model for the tested scenarios.

Significance: The model calculation time was reduced by factor of 1500, now enabling many new studies for PGT-based systems.

Objective: Boron Neutron Capture Therapy is a cancer radiotherapy that uses the selective uptake of boron compounds by tumor cells, followed by neutron irradiation. Conventional dosimetry generally assumes a homogeneous boron distribution within tissues, yet evidence indicates intracellular heterogeneity. This work aims to improve the Photon Isoeffective Dose model (PID) for Glioblastoma Multiforme (GBM) by incorporating subcellular-scale effects: (i) a correction factor for the stochastic nature of energy deposition due to intracellular boron localization, and (ii) the treatment of the nucleus-to-cytoplasm boron concentration ratio as a stochastic variable.

Approach: The boron-10 microdistribution in U-87 glioblastoma cells was quantified for the first time through neutron autoradiography, revealing preferential accumulation in the nucleus. Following these experimental data, the nucleus-to-cytoplasm boron concentration ratio was described by a lognormal random variable, consistent with biological uptake processes. The correction factor was applied to the dosimetry of U-87 radiobiological data. Then, updated radiobiological parameters and subcellular-scale effects were integrated into the PID formalism and applied to a clinical case of GBM.

Main results: The outcome was a Microdosimetric Photon Isoeffective Dose Model, which extends conventional PID by explicitly including intracellular boron heterogeneity. Applied to U-87 data, proposed corrections revealed a 47% reduction in the Compound Biological Effectiveness factor compared to conventional calculations, showing that neglecting subcellular distribution substantially overestimates the boron dose. For the clinical case, the total dose and 1-year Progression-Free Survival (PFS) differed only by 4% and 3%, respectively, compared to conventional dosimetry. However, perturbation analyses indicated that under higher intracellular heterogeneity, plausible in vivo, the deviations could become substantial (up to 22% in dose and 68% in PFS).

Significance: These findings highlight the relevance of subcellular-scale modeling. The proposed Microdosimetric Model, grounded on experimentally derived microdosimetric corrections, provides a robust framework to improve both the accuracy and the personalization of BNCT treatment planning.

Objective: This study aims at evaluating the gadolinium (Gd) beam filter in generating high quality dual-energy cone beam CT (CBCT) images with a dual-layer flat panel detector (FPD). Approach: First of all, the Cramer-Rao lower bounds (CRLBs) were estimated on dual-energy X-ray projections to determine the optimal settings such as beam filter thickness, CsI:TI scintillator thickness, and copper (Cu) filter thickness between the two detector layers. Afterwards, dual-energy CT imaging were numerically simulated with varied Gd/Cu beam filter thickness and Cu filter between the two detector layers. Finally, physical experiments were validated on our dual-layer FPD based dual-energy CBCT imaging benchtop. Main results: For a specific dual-energy imaging task, in general, optimizations are required in order to achieve satisfactory imaging performance. Compared to the Cu beam filter, the CRLB analyses found that Gd beam filter would generate material basis with lower noise levels, i.e., higher signal-to-noise ratio (SNR). In addition, physical experiments show that the Gd beam filter might be a better choice than the Cu beam filter in ensuring accurate basis decomposition. Specifically, the Gd beam filter yielded higher SNR than the Cu beam filter in all cases: by 41.3 (water) and 52.9 (bone) in numerical simulations; by 31.0 (water) and 0.9 (iodine) for the cylindrical phantom; and by 15.6 (water) and 13.3 (iodine) for the head phantom. However, the virtual mono-chromatic images (VMIs) generated with Cu beam filter always have higher contrast-to-noise ratio (CNR) than with the Gd beam filter. Both numerical simulations and physical experiments show that the inter-layer Cu filter may enhance the basis decomposition performance, but at the expense of reducing the total radiation efficiency. Significance: For 160 mm diameter phantom, this study found that the Gd filter is a promising alternative to the conventional Cu filter in improving the dual-energy basis imaging performance of a dual-layer FPD based CBCT system.

Objective. Targeted radionuclide therapy using alpha-particle-emitting radiopharmaceuticals (alpha-RPT) is increasingly recognized as an effective, safe and economically viable clinical treatment. However, it is restricted to few cancer types, and to metastatic or unresectable tumors as a palliative treatment. Broader implementation of alpha-RPT across cancer types and earlier disease stages is hampered by limitations of current clinical dosimetry. Alpha-RPT administration regimens rely on fixed protocols for intermittent radioactivity (RT) administration, without dynamic adjustments. This study provides a computational proof-of-concept of continuous dynamic-discretized RT administration strategy for alpha-RPT inspired by black hole (BH)-like dynamics.Approach.BHs can exhibit impressive forms of convergence, stability and robustness, ensuring a trapped region, in which matter cannot escape from it. When extrapolated to cancer therapy, the tumor is analogically considered as a mass inside a BH, in which the BH center represents the cancer remission, and the alpha-RPT administration acts as the gravitational attraction pulling the tumor mass towards the center (where a complete remission is reached). Using a recently validated mathematical model of Actinium-225 alpha-RPT in a Murine breast cancer model, we were able to predict geometro-radiopharmacokinetics and tumor dynamics for different number of tumor cells, discretization intervals, and a wide variation range of tumor parameters.Main results.Our results show that BH-like RT administration can significantly reduce total administered RT and treatment duration compared with current clinical practice based on intermittent administration, while maintaining therapeutic efficacy, even under highly uncertain tumor dynamics. Reductions in treatment duration up to 48.8% were obtained, as well as reductions in maximum/average RT administration up to 54.3%/81.1%.Significance. These findings suggest that adaptive control strategies may overcome key limitations of current alpha-RPT protocols, allowing dynamically adjusted RT administration according to tumor state data obtained from biomarker data and/or theranostic imaging. This strategy holds the potential to refine clinical protocols and expand alpha-RPT beyond its current limitations, establishing the 'biological BH' as a new high-impact foundation for spreading alpha emitting RPT to primary cancers and multiple cancer types.

Objective: Current automatic segmentation models in radiotherapy, which are predominantly unimodal and image-based, have limited generalizability due to boundary ambiguity and the lack of guideline integration. This study proposes a text-guided segmentation network, termed TG-SegNet, for the automatic delineation of clinical target volumes (CTVs) in rectal cancer radiotherapy.

Approach: Data from 567 preoperative patients with rectal cancer were retrospectively collected. Text prompts contained (i) patient case information (age, sex, tumor stage, tumor location, position) and (ii) guideline-derived descriptions indicating which CTV subsites should be included. TG-SegNet integrates computed tomography (CT)-derived visual features with structured clinical text prompts encoded by PubMedBERT, fused via cross-attention and fine-grained fusion. The model was trained on 452 patients and tested on 115. Its performance was compared with that of nnU-Net and two ablated variants (TG-SegNet without text prompts and TG-SegNet with simplified fusion). The evaluation comprised quantitative metrics, including Dice similarity coefficient (DSC), 95% Hausdorff distance (HD95), mean surface distance (MSD), surface DSC (S-DSC), and average path length (APL), along with blinded expert scoring and an efficiency analysis. In additional analyses, we conducted text-prompt and module ablations.

Main results: TG-SegNet achieved the best performance across all quantitative metrics: DSC 0.927±0.022, HD95 7.01±6.05 mm, MSD 1.94±1.08 mm, S-DSC 0.799±0.074, and APL 7372±4452 (all p<0.01). In clinical evaluation, TG-SegNet significantly improved target coverage, guideline adherence, and overall clinical acceptability compared with nnU-Net and ablations (p<0.05), with boundary appropriateness comparable to nnU-Net. TG-SegNet had the shortest correction time (3.39±1.10 minutes), corresponding to 82.1% time savings versus manual delineation. Text-prompt ablations suggested that the CTV-subsite prompt component contributed more to performance. Module ablations showed that both cross-attention and fine-grained fusion were beneficial.

Significance: By integrating clinical semantics with imaging, TG-SegNet demonstrated superior accuracy, efficiency, and clinical acceptability over nnU-Net and ablated models, highlighting its potential for clinical Translation.

Objective: To develop and optimise a dedicated low-energy proton beamline at the Centre for Micro-Analysis of Materials (CMAM, Madrid, Spain) for radiobiological applications.

Approach: An automated irradiation system was implemented, integrating Gafchromic EBT3 radiochromic film (RCF) dosimetry corrected for linear energy transfer (LET) dependent quenching and post-irradiation darkening. Dosimetric calibration was performed using multichannel analysis, and beam performance was systematically evaluated as a function of distance, raster scanning area, beam intensity and reproducibility.

Main results: Optimal operating conditions were identified at moderate beam currents (≤ 1 nA) and scanning areas of 40 × 40 to 50 × 50 mm2, yielding homogeneous dose distributions with reproducibility better than 8%. The dosimetric protocol demonstrated linearity across clinically relevant dose ranges and allowed a reliable correlation between irradiation parameters and absorbed dose. Proof-of-concept experiments on U-87MG glioblastoma cells confirmed the system's ability to deliver controlled and biologically effective proton exposures, as demonstrated by clonogenic survival assays.

Significance: These results establish the CMAM beamline as a robust and versatile platform for preclinical proton radiobiology, providing accurate dosimetric control and supporting investigations of relative biological efficacy (RBE). The system facilitates translational advances in proton radiobiology, bridging physical and biological studies in low-energy proton irradiation.

Objective: This study aims to implement and evaluate a source modeling method for Monte Carlo (MC) simulation of computed tomography (CT) systems in the absence of manufacturer data. This work enables simulation of realistic CT x-ray sources by integrating experimental measurements, particularly x-ray spectrometry, and to assess its applicability to single-energy CT (SECT) scanners. Approach: An experimental method was implemented combining x-ray spectra using a CdTe spectrometer and a bowtie filter attenuation profile measured with an ionization chamber, and compared to manufacturer data. Two source models with different levels of complexity were created for GATE 10 simulations. First, a basic model (Single-spectrum) using one energy spectrum measured on the beam axis combined with the attenuation profile was considered. Multi-spectra model incorporated additional spectra off-axis. For comparison, the Single-spectrum model was used with manufacturer's data. These source models were evaluated by comparing simulations and measurements of half-value layers (HVL) of aluminum at three voltages (80 kV, 120 kV and 140 kV), as well as CT dose index (CTDI) values measured on both head and body phantoms. Main results: HVL of Single-spectrum and Multi-spectra models showed a better agreement with measurements, yielding a mean difference of 4%. CTDI values derived from simulations based on Multi-spectra source outperformed the other sources, with differences inferior to 7% with measurements. Single-spectrum results had good agreement with measurements (<10%), but underestimated dose evaluation in some cases by up to 16%. The Manufacturer source showed the largest discrepancies, especially at 140 kV. Significance: This work highlighted that modeling spectral variations across the x-ray beam significantly improves CT source model. The proposed framework offers a replacement for manufacturer data when not available and participates in the foundation for using Monte Carlo methods to support the integration of new CT systems.