Physics in medicine and biology最新文献

Objective. The biomechanical properties of tissue are of interest in preclinical cancer research where their changes can be related to treatment response. Preclinical shear wave elastography (SWE) may be used to measure the viscoelastic properties of tumours although their small size presents challenges. Here we study the repeatability of 3D shear wave speed (SWS) measurements using continuous harmonic vibrations under different conditions in a preclinicalin-vivotumour model.Approach. Subcutaneous tumours (MDA-MB-231) grown on the flank of athymic nude mice (n= 4) were imaged using a system comprising a research ultrasound scanner and a mechanically translated 18 MHz linear imaging probe. Shear waves were induced in the tumours by external contactors driven at three different vibration frequencies (500, 700 and 1000 Hz), in two orientations (top and side), sedated in separate sessions using injectables or breathable anaesthesia. Measurements were repeated over three consecutive days. 3D tumour volume outlines were used to determine the spatial transformation required to register sets of 3D SWS data, allowing measurement of repeatability of the 3D pattern of SWS using normalised cross correlation.Main results. Analysis of variance of mean SWS measurements (2-5 m s^-1) revealed significant differences between the tumours (p< 0.001), and vibration frequencies (p< 0.001). Mean SWS was not significantly affected by the choice of anaesthetic or tumour orientation. Intratumoural SWS spatial distributions showed improved day-to-day repeatability when obtained from the same tumour (+76% increase in normalised cross correlation compared to different tumours), the same orientation (+39% compared to different orientations), and when using a side orientation at 500 Hz (+18% compared to top orientations). Breathing motion with gaseous anaesthesia was found to be slower (∼1.5 s vs ∼0.5 s period) but with greater amplitude (<0.6 vs <0.3 mm) than with injectable. Side orientation was found to reduce respiratory motion amplitude. SWS measurements and their repeatability however were not significantly affected by the choice of anaesthesia, and therefore variation in breathing motion.Significance. SWE with continuous vibration is a repeatable and feasible technique forin-vivopreclinical use.

Objective. Ultra-high dose rate irradiation (FLASH) is a promising way to reduce the adverse effects of radiotherapy treatment. However, the impact of dosimetric parameters on the FLASH effect is still not well understood and few devices can deliver irradiation in FLASH conditions. The aim of this study is to demonstrate the capability of a 25 MeV proton beamline to deliver the prescribed dose to a biological sample in conventional and FLASH conditions.Approach. We characterized the different parameters influencing dose delivery (beam energy, irradiation time, beam intensity) to confirm the accuracy of our irradiation protocol. In parallel, we developed a Monte-Carlo simulation of the beamline for irradiation planning. Then dose uniformity and reproducibility with respect to dose rate were then evaluated using radiochromic films.Main results. The uncertainty on dose delivery is around 1% for both conventional and FLASH dose rates. Dose monitoring shows that the prescribed dose is delivered within a margin of 3%, with a uniformity better than 5% for a dose rate up to 150 Gy s^-1. The results of the Monte-Carlo simulation of the beamline are in strong agreement with our experimental measurements, thus validating the model.Significance. This work demonstrates the validation of a passive irradiation beamline delivering uniform irradiation with dose rates up to 150 Gy s^-1for centimetric irradiation fields. By characterizing beam parameters independently, we propose a dose validation method for beam energies where standard dose-calibrated detectors cannot be used.

Background. Uncertainty quantification (UQ) has emerged as a crucial component in deep learning-based medical image analysis, particularly in radiotherapy (RT). Addressing uncertainty is essential for improving the reliability, interpretability, and clinical applicability of AI-driven models in key RT tasks, including segmentation, image registration, synthetic image generation, dose prediction and dose accumulation. Despite significant advancements, challenges remain in integrating UQ techniques into RT clinical workflows.Purpose. This review synthesizes recent developments in UQ methods applied to RT. It introduces a structured classification of UQ techniques, evaluates their impact on clinical workflows, and highlights emerging trends from studies published from 2020 to 2025.Methods. A systematic search was conducted on PubMed and Google Scholar for articles published from January 2020 to June 2025. Keywords included 'uncertainty', 'radiotherapy', and task-specific terms such as 'segmentation', 'registration', 'synthetic image generation', 'image-to-image translation', 'dose prediction', or 'dose accumulation'. Studies were classified based on the type of uncertainty estimation technique, imaging modality, and associated RT task.Results. Segmentation emerged as the most common RT task addressed by UQ methods, followed by image registration, synthetic image generation and dose prediction. Probabilistic techniques such as Bayesian neural networks, Monte Carlo dropout, and ensemble learning, dominate the field, particularly for modeling epistemic uncertainty. Studies demonstrated that uncertainty maps enhance model interpretability, guide clinical review of auto-segmentations, and support quality assurance processes.Conclusion. UQ has the potential to enhance the robustness of AI-driven RT workflows. While substantial progress has been made, further efforts are needed to standardize evaluation protocols, improve computational efficiency, and develop user-friendly interfaces for clinical integration. Future research should aim to close the gap between technical advances and their clinical deployment to ensure uncertainty-aware models contribute effectively to personalized RT.

Objective.Dedicated positron emission tomography (PET) scanners designed for specific organs or clinical applications require compact detector modules with high depth-of-interaction (DOI) and time-of-flight (TOF) capabilities. In this study, we present the design and evaluation of a compact, ready-to-use PET detector panel optimized for such scanners.Approach.The panel, measuring 98.4 × 104.2 mm², comprises a 4 × 3 array of four-layer, dual-readout detector towers. Detector towers operate in side-irradiation configuration, thereby enabling DOI measurement across the layers, while axial positioning is derived from the dual-ended readout. Each tower is built from a 8 × 4 × 1 array of 2.05 × 4.4 × 30 mm³Lutetium Fine Silicate (LFS) crystals, axially coupled to strip-shaped multi-pixel photon counters, with both ends of each strip read out through independent electronic channels. A high-speed electronic readout system based on the picoTDC application-specific integrated circuit was developed to enable precise timing and amplitude measurements. Calibration and performance evaluations were conducted under realistic and scaled conditions. A full-range energy calibration was performed at crystal-level using multiple gamma-emitting isotopes to linearize the detector's response and extract energy resolution. Calibration for axial-positioning along the length of the crystals (between two readout ends) was achieved through a simple flood irradiation-based method, eliminating the need for point-specific irradiations.Main results.Average energy resolutions of 14.2%, 14.3%, 15.3%, and 15.4% were achieved for crystals in layers 1 through 4, respectively. DOI and transaxially positioning steps of 4.4 mm, and 2.05 mm, respectively are obtainable based on layer and crystal pitch. The measured axial spatial resolutions were 3.78 mm, 3.84 mm, 4.01 mm, and 4.78 mm full-width-half-maximum for layers 1 through 4, respectively. TOF resolution averaged 196 ± 7 ps for layer 1-1 pair, gradually degrading to 220 ± 17 ps for layer 4-4 pairs.Significance.Balancing performance, scalability, and manufacturability, this detector panel offers a practical and easily calibratable solution for next-generation organ-dedicated PET systems with DOI-TOF capability.

This study introduces a novel adaptive reconstruction algorithm based on the time reversal filtering factor (TRFF) method, aimed at improving the quality and accuracy of magnetic thermoacoustic (MTA) imaging. The TRFF method overcomes the limitations of traditional time-reversal (TR) techniques by incorporating a dynamically adjustable filtering factor and a weighting function that adaptively processes acquired signals. This adaptive approach assigns higher weights to high-quality signals, while effectively suppressing noise and artifacts. We applied the TRFF algorithm to both numerical simulations and experimental setups, demonstrating its effectiveness in improving signal-to-noise ratio (SNR), reducing artifacts, and enhancing the contrast of target signals. In numerical simulations, we compared the TRFF method to conventional TR methods, using metrics such as root mean square error, peak SNR, and structural similarity index. The results highlighted the superior performance of the TRFF method in reconstructing high-quality images. For experimental validation, we utilized the TRFF algorithm for multi-layer processing of three-dimensional MTA data, significantly improving imaging quality for deep tissues. We optimized a 16-channel array ultrasonic transducer (AUT-16) for efficient three-dimensional imaging. Separately, a 128-channel arc-shaped AUT (AAUT-128) was developed to achieve real-time imaging. The AUT-16 enabled faster scanning and better spatial information reconstruction, while the AAUT-128 facilitated high-frame-rate real-time imaging of dynamic magnetic nanoparticles, showcasing its potential for dynamic biomedical monitoring. This study marks significant advancements in both signal processing and hardware design for MTA imaging. The integration of the TRFF method enhances both pseudo-3D and real-time imaging capabilities, presenting a promising approach for future applications in biomedical diagnostics and complex tissue imaging.

Objective.This study aims to comprehensively compare the PENHAN and FLUKA Monte Carlo codes for low-energy alpha particle transport and small-scale dosimetry using alpha-emitting radionuclides, and to assess their suitability for such applications.Approach.Two studies were performed through Monte Carlo simulations. First, monoenergetic alpha particles (3-10 MeV) were distributed in a micrometric water sphere and the dose deposition within it was calculated. Second, a simplified spherical cell model with uniformly distributed alpha-emitting radionuclides was used to computeS-values. PENHAN and FLUKA results were compared, and potential sources of discrepancy between them were analyzed. In addition, both codes were benchmarked against MIRDcell, an analytical tool widely used for dosimetric calculations in Targeted radionuclide therapy.Main results.In the monoenergetic study, the primary source of discrepancy between PENHAN and FLUKA was the stopping powers used for alpha particles. When the same stopping powers were employed, both codes yielded statistically compatible results, except at 3.0 and 3.5 MeV, where FLUKA showed an anomalous behavior. In the cell model, variations were below 3% but not negligible even when using identical stopping powers, suggesting an additional source of discrepancy: differences in the radionuclide emission spectra, particularly in the electron component. In both studies, PENHAN and FLUKA results were overall in good agreement with those from MIRDcell.Significance.This study demonstrates, for the first time, the suitability of PENHAN for low-energy alpha transport and small-scale dosimetry with alpha emitters, provided that accurate stopping powers are employed. It also supports the reliability of FLUKA in these scenarios and shows that both codes yield compatible results when using consistent stopping power datasets and radionuclide emission spectra. This work further highlights the importance of validating Monte Carlo codes in medical physics to ensure the reliability and reproducibility of their results.

Objective.Echo-planar imaging (EPI) can provide rapid quantitative susceptibility mapping (QSM) in single-shot acquisition but suffers from B₀inhomogeneity and susceptibility artifacts near air-tissue interfaces. To address these limitations, this work introduces single-shot echo planar time-resolved imaging (EPTI), which enables distortion-free and multi-contrast imaging for rapid QSM.Approach.A zig-zagk_y-tsampling trajectory was employed in two-dimensional single-shot EPTI and a locally low-rank subspace reconstruction with B₀updating was applied to generate distortion-free multi-echo images from highly undersampled data. The proposed EPTI-QSM method was systematically evaluated against the gold-standard three-dimensional (3D) gradient echo (GRE) and single-shot EPI with a uniform 4-step masking procedure. In addition, various echo selections from the EPTI images were investigated to assess their impact on QSM quality.Main Results.EPTI-QSM demonstrated better anatomical fidelity, reduced susceptibility artifacts, and a reliable brain coverage compared to single-shot EPI, particularly in regions affected by B₀inhomogeneity. Multi-echo EPTI data with echo times ranging from 10 to 30 ms further improved susceptibility quantification and mitigated signal dropouts near air-tissue interfaces. In addition, EPTI yielded distortion-free structural images compared to EPI, and improved image contrast compared to 3D GRE used for QSM, enabling clearer anatomical visualization.Significance.Single-shot EPTI enables distortion-free, multi-contrast images and rapid QSM reconstruction, offering a promising alternative to EPI-based QSM, particularly in applications requiring rapid and robust susceptibility quantification.

Objective.To investigate the feasibility of neutron capture enhanced particle therapy (NCEPT) using synchrotron-accelerated carbon ion beams by evaluating the production and characteristics of thermal neutrons (with energies below 0.5 eV), which are optimal for neutron capture reactions.Approach.The fluence of thermal neutrons was measured via gold detector activation in a PMMA phantom irradiated with scanning carbon ion beams. Monte Carlo simulations using MCNP 6.2 were concurrently conducted to verify the experimental results and assess the potential for NCEPT dose enhancement under spread-out Bragg peak (SOBP) beam conditions.Main results.The experimental measurement of thermal neutron fluence within the SOBP region was consistent with the Monte Carlo simulations. The simulations further revealed that the maximum neutron fluence appeared within the SOBP region. However, a quantitative comparison showed that the neutron fluence generated by the carbon ion beam is orders of magnitude lower than the minimum requirements for conventional BNCT. Consequently, the observed physical dose enhancement was not clinically significant.Significance.This study provides the first experimental evidence confirming the generation of thermal neutrons by synchrotron-accelerated scanning carbon ion beams. While the current neutron yield limits clinical utility, the spatial congruence between the maximum neutron fluence and the SOBP region remains a promising feature, serving as the basis for future research focusing on optimizing parameters of NCEPT.

Objective. Quality assurance of hyperthermia applicators can be a cumbersome task. Periodic validation of the fields generated by the applicator is crucial for ensuring proper device performance, but the required measurements are very time-consuming. While most clinics use heating rate as a parameter of interest, consensus exists that spatial variation of the electromagnetic field in three dimensions (3D) would be much more insightful. Unfortunately, such 3D coverage would require measurements at many locations.Approach. To address this challenge, we propose a compressed sensing based methodology that enables accurate E-field and specific absorption rate (SAR) reconstruction from significantly reduced sampling densities. Using a Lucite Cone Applicator (LCA) and a homogeneous tissue-mimicking phantom, E-field measurements were obtained via a robotic scanning system equipped with an isotropic EM field probe (EX3DV4, SPEAG). Field maps were reconstructed using a discrete cosine transform (DCT)-based compressed sensing algorithm and evaluated using peak signal-to-noise ratio (PSNR), structural similarity index (SSIM), and the area under 50%-iso-field contour overlap error. This error refers to the computational reconstruction accuracy of the compressed sensing algorithm when benchmarked against a densely sampled high-resolution reference scan.Main results. Results demonstrated that accurate field reconstruction can be achieved using only 8% of the full measurements, reducing acquisition time from 135 minutes to just 11 minutes, while maintaining clinically relevant precision (SSIM = 0.9, PSNR = 27 dB, 50%-iso-field contour overlap error is within ± 2.5%).Significance. In this way, the need for extensive measurements is reduced while validation reliability is maintained. This approach delivers a faster solution, enhancing information content while significantly reducing the time required for quality assurance in hyperthermia clinics.

Objective. Radiation biophysical modelling of the spatio-temporal events following energy deposition in a tissue-like medium is a useful tool for investigating mechanistic features of radiobiological processes. The present study focuses on the description of complex milieux and long time domains.Approach. Monte Carlo (MC) chemical track structure algorithms allow the formation, transport, and recombination of radical species under various irradiation conditions to be followed. This feature has been proposed to have outermost relevance, e.g. in the comprehension of the FLASH effect. Nevertheless, to extend the simulations predictability range in both temporal scales and realistic environments, while avoiding prohibitive running times, computationally lighter approaches have to be used in combination with the accurate step-by-step descriptions provided by MC. To this end, TRAX-CHEMxt has been implemented.Main results. We propose here an upgraded version of the code, capable now to investigate the chemical effects of radiation up to 1 s and in a more complex environment, featured not only by oxygenated water, but also by a representative biomolecule, RH, and an antioxidant component, XSH. The robustness of the code in this new configuration has been proven. Its predictions are compared with both full MC counterparts at the overlapping time scale, (1-10) µs, and available experimental data at longer temporal points, showing in all cases good agreements. The change in the chemical yields due to the presence of RH and XSH is then investigated, as a function of primary particle type, energy, LET, and target oxygenation.Significance. TRAX-CHEMxt can thus be effectively applied to study the impact of radiation-induced radicals at larger time scales on more complex systems, allowing for specific biological targets simulations.