Medical physics最新文献

Background: Radiation therapy delivers precise doses to tumors, but accurately measuring internal tissue doses remains a challenge. Current methods, such as ionization chambers and radiographic films, rely on external measurements, which cannot provide direct, in vivo dose feedback. X-ray acoustic computed tomography (XACT) was developed to generate thermoacoustic signals when x-rays deposit energy into water or tissue, enabling the reconstruction of dose distribution patterns through acoustic signals. However, the longer pulse width of x-rays from linear accelerators reduces the efficiency of thermoacoustic signal conversion, lowering the signal-to-noise ratio (SNR) of radiofrequency (RF) signals. This noise significantly affects the quality of reconstructed XACT images. Overcoming the impact of noise is essential for advancing XACT toward accurate dose detection.

Purpose: This study aims to develop a frequency-aware denoising (FAD) method for overcoming the impact of band-limited and white noise in RF signals for XACT.

Methods: Real RF signals were acquired from an XACT system using radiotherapy megavoltage (MV) x-rays, a water tank, and an ultrasound transducer. To capture the frequency characteristics of these RF signals, we first estimated the probability density function (PDF) of their frequency spectrum. To generate synthetic RF data that closely approximates realistic noisy signals for model training, noise was sampled from this PDF, incorporating both magnitude and random phase components, and combined with simulated signals and white noise. A conditional diffusion model was trained on these synthetic signals to obtain the FAD model. A total of 3150 frequency-aware RF data samples were used to train the FAD model. For testing, acoustic RF signal data excited by five different x-ray shapes were measured, denoised by the FAD model, and finally reconstructed into XACT. The performance of the method was evaluated based on XACT image quality using SNR analysis and γ passing rate, and compared with results from Raw-RF and background noise-removed (BNR-RF) methods.

Results: The FAD-RF method produced XACT images with clearer structural details and fewer artifacts. It achieved the highest SNR among the tested methods, with a mean SNR of 27.6 ± 5.0, outperforming both Raw-RF (22.9 ± 2.2, p < 0.05) and BNR-RF (22.0 ± 3.0, p < 0.05). In terms of spatial accuracy, the FAD-RF method also outperformed in γ analysis, achieving a mean γ passing rate of 79.0% ± 2.4%, significantly higher than Raw-RF (50.2% ± 20.0%, p < 0.05) and BNR-RF (73.5% ± 3.6%, p = 0.078).

Conclusion: The FAD-RF method relies solely on the RF signals for denoising, making it practical and efficient for real-world applications. It demonstrates effective noise suppression and enhanced spatial accuracy in XACT image reconstruction.

Background: Generative adversarial networks (GANs) have recently demonstrated significant potential for producing virtual images with the same characteristics as real-life landscapes, thereby enhancing various medical tasks.

Purpose: To design a region-guided focal GAN (Focal-GAN) for translating images between CT and MRI and test its clinical applicability in patients with hepatocellular carcinoma (HCC).

Methods: Between January 2012 and October 2021, two cohorts of patients with HCC who underwent contrast-enhanced CT (Center 1, n = 685) and MRI (Center 1, n = 516; Center 2, n = 318) were retrospectively enrolled. We trained the Focal-GAN model by adding tumor regions to a baseline Cycle-GAN framework to steer the model toward focal attention learning. The quality of the images generated was assessed using an open-source MRQy tool. The clinical applicability of the Focal-GAN was evaluated by applying the nnUNet and ResNet-50 model for tumor segmentation and microvascular invasion (MVI) prediction in HCC on the generated images.

Results: In the ablation tests, Focal-GAN achieved a higher fidelity than the conventional Cycle-GAN in the generated image quality assessment with MRQy. Regarding applicability, regardless of tumor size, nnUNet trained with focal-GAN-generated images achieved higher Dice scores than nnUNet trained using Cycle-GAN-generated images for HCC segmentation in both internal (0.607 vs. 0.341, p < 0.01) and external (0.796 vs. 0.753, p < 0.001) validation. Additionally, ResNet-50 trained with Focal-GAN-generated images produced higher areas-under-curve (AUCs) than ResNet-50 trained with real images for MVI prediction in both internal (0.754 vs. 0.665, p = 0.048) and external (0.670 vs. 0.579, p < 0.001) validation.

Conclusions: The designed Focal-GAN model can generate virtual MR images from unpaired CT images, thereby extending the clinical applicability of CT in the liver tumor diagnostic pathway.

Purpose: As a portable and cost-effective imaging modality with better accessibility than Magnetic Resonance Imaging (MRI), transcranial sonography (TCS) has demonstrated its flexibility and potential utility in various clinical diagnostic applications, including Parkinson's disease and cerebrovascular conditions. To better understand the information in TCS for data analysis and acquisition, MRI can provide guidance for efficient imaging with neuronavigation systems and the confirmation of disease-related abnormality. In these cases, MRI-TCS co-registration is crucial, but relevant public databases are scarce to help develop the related algorithms and software systems.

Acquisition and validation methods: This dataset comprises manually registered MRI and transcranial ultrasound volumes from eight healthy subjects. Three raters manually registered each subject's scans, based on visual inspection of image feature correspondence. Average transformation matrices were computed from all raters' alignments for each subject. Inter- and intra-rater variability in the transformations conducted by raters are presented to validate the accuracy and consistency of manual registration. In addition, a population-averaged MRI brain vascular atlas is provided to facilitate the development of computer-assisted TCS acquisition software.

Data format and usage notes: The dataset is provided in both NIFTI and MINC formats and is publicly available on the OSF data repository: https://osf.io/zdcjb/.

Potential applications: This dataset provides the first public resource for the development and assessment of MRI-TCS registration with manual ground truths, as well as resources for establishing neuronavigation software in data acquisition and analysis of TCS. These technical advancements could greatly boost TCS as an imaging tool for clinical applications in the diagnosis of neurological conditions such as Parkinson's disease and cerebrovascular disorders.

Purpose: Computational phantoms have been widely used in radiation protection, radiotherapy, medical imaging, surgery navigation, and digital anatomy. However, current Chinese phantoms lack representation for all sensitive groups including adults, children, and pregnant women. This manuscript aims to address this gap by developing novel open-access computational phantoms representing the Chinese population.

Acquisition and validation methods: The Chinese reference population (CRP) developed in this study includes 30 phantoms, available in both voxel and nonuniform rational B-spline (NURBS) formats, with ages in 0, 1, 2, 3, 4, 5, 6, 8, 10, 12, 15, 18 years, and adult male and female, as well as four pregnant women in early pregnancy, first trimester, second trimester, and third trimester. The development process involved image segmentation, NURBS reconstruction, and voxelization based on whole-body computed tomography (CT) scans of 22 original individual patients. Reference organ masses were directly obtained from the Chinese Reference Human Anatomical Physiological and Metabolic Data, as well as international commission on radiological protection (ICRP) Publication 89.

Data format and usage notes: Voxelized phantoms are accessible in DAT format as raw data, which can be opened by medical imaging softwares such as a medical image data analysis tool (AMIDE). Excel files contain descriptive information (ages, genders, phantom sizes, voxel sizes, organ masses, densities) and organ absorbed doses on ${}^{18}F-FDG$ application. All data in this study can be obtained from our official website (https://alldigitaltwins.com) and Zenodo (https://zenodo.org/records/14268606).

Potential applications: This work offers a collection of open-source age-dependent phantoms featuring anatomical data specific to the Chinese population. Researchers can utilize this dataset to modify and adapt the phantoms for specific applications, fostering innovation and progress, and enhancing accuracy and applicability in various fields.

Background: Gamma-ray detection plays a crucial role in the fields of biomedicine, space exploration, national defense, and security. High-precision gamma photon detection relies on scintillation crystals, which attenuate gamma rays through mechanisms such as photoelectric effect and Compton scattering. These interactions generate light signals within the scintillation crystal, which are subsequently converted into electronic signals using photodetectors, enabling accurate readout, and analysis.

Purpose: Improving the readout efficiency of visible photons in crystal detectors can significantly improve the efficiency of gamma-ray detection. Scintillator crystals are usually hard and brittle materials, which are difficult to process. In this paper, we innovatively propose the method of using a femtosecond laser to process grooved microstructures on the light output surface of scintillator crystals to improve the detection efficiency, and thus enhance the comprehensive performance of gamma-ray detection.

Methods: Optical simulation software is first used to explore the enhancement of the light output performance by the grooved microstructures. Subsequently, a 5-dimension system for femtosecond laser processing of scintillator crystals was constructed, which can achieve accurate processing of grooved structures. Finally, the feasibility of the study was verified by applying grooved microstructure on crystal bars and crystal arrays.

Results: TracePro simulation results showed an average efficiency improvement in light output of 33.56% within the groove parameters: spacing from 20 to 140 µm, depth from 8 to 28 µm, and width from 10 to 30 µm. A custom-designed readout electronic system for gamma detection and a laser processing platform was then constructed to evaluate the feasibility of applying grooved structures to the lutetium-yttrium oxyorthosilicate (LYSO) crystal surface. According to the simulation results, 12 groups of crystal bars were fabricated with spacings from 60 to 140 µm, depths from 7 to 16 µm, and widths from 11 to 14 µm. Experimental results showed an average improvement of 20.4% in light output for the crystal bars, and that of the crystal arrays can be improved by 6.85% on average.

Conclusions: This study introduces a method of using femtosecond lasers to fabricate grooved microstructures on LYSO crystal surfaces, which has demonstrated a significant improvement in light output in both simulation and experimentation. This method can be applied to the production of crystal arrays at a low cost and on a large scale, showing promising potential for common gamma detection applications, such as medical imaging, industry, and astrophysics.

Background: Photon counting X-ray detectors (PCDs) provide better spatial resolution than energy integrating X-ray detectors, but even higher resolution is desired in some applications. Certain charge sharing compensation techniques such as coincidence counting preferentially detect photons that arrive at the boundary between pixels, and this could be used for subpixel localization of incident photons and enhanced spatial resolution.

Purpose: To estimate improvements to spatial resolution and detective quantum efficiency that are possible when using coincidence counting for high-resolution, non-spectral imaging.

Methods: The modulation transfer function (MTF), noise power spectrum (NPS), and detective quantum efficiency (DQE) were estimated using numerical simulations of a two-dimensional parallel-beam CT system. Coincidence counters were modeled either geometrically or with Monte Carlo simulations. The geometric model consisted of narrow coincidence counters that interlaced with wider ordinary pixels, and showed that using standard filtered backprojection decreases low-frequency DQE, but that a frequency weighting technique could be used to restore DQE. The Monte Carlo simulations were used to estimate the possible improvements that could be expected from real systems. The pixel pitch was 0.25 mm and the source apertures considered were 0, 0.125, and 0.5 mm. A numerical stent phantom was also used to illustrate possible improvements.

Results: Assuming a 0.125 mm source aperture and the Monte Carlo model, the limiting MTF (at 10%) increased from 20 to 40 lp/cm using coincidence counters. This can be explained by the increased sampling (and higher Nyquist limit) possible from coincidence counters. In the geometric model, coincidence counters are compared to conventional double sampling techniques such as in-plane flying focal spot, and the MTF at 38 lp/cm increased from 5% to 53%. Without the frequency weighting technique, low-frequency DQE was reduced by about 20%, but these losses are recovered with frequency weighting. Improvements are much more modest with the 0.25 mm source aperture because the system becomes source-limited.

Conclusions: Coincidence counting could be used to increase spatial resolution in PCDs. The increases in system resolution could be large if a high-resolution X-ray source were available.