Biomedical Physics & Engineering Express最新文献

Gadolinium-based contrast agents (GBCAs) are commonly employed with T1-weighted (T1w) MRI to enhance lesion visualization but are restricted in patients at risk of nephrogenic systemic fibrosis. In addition, variations in GBCA administration can introduce imaging inconsistencies. This study develops an efficient 3D deep-learning framework to generate T1-contrast enhanced images (T1C) from pre-contrast multiparametric MRI. We propose the 3D latent rectified flow (T1C-RFlow) model for generating high-quality T1C images. First, T1w and T2-FLAIR images are input into a pretrained autoencoder to acquire an efficient latent space representation. A rectified flow diffusion model is then trained in this latent space representation. The T1C-RFlow model was trained on a curated dataset comprised of the Brain Tumor Segmentation (BraTS) 2024 glioma (GLI; 1480 patients), meningioma (MEN; 1141 patients), and metastases (MET; 1475 patients) datasets. Selected patients were split into training (N = 2860), test (N = 614), and validation (N = 612) sets. Model performance was evaluated with the normalized mean squared error (NMSE) and structural similarity index measure (SSIM). Both qualitative and quantitative results demonstrate that the T1C-RFlow model outperforms benchmark 3D models (pix2pix, denoising diffusion probability models (DDPM), Diffusion Transformers (DiT-3D) trained in the same latent space. T1C-RFlow achieved the following metrics - GLI: NMSE 0.044 ± 0.047, SSIM 0.935 ± 0.025; MEN: NMSE 0.046 ± 0.029, SSIM 0.937 ± 0.021; MET: NMSE 0.098 ± 0.088, SSIM 0.905 ± 0.082. In a blinded reader study of 15 patients (5 GLI, 5 MEN, 5 MET), T1C-RFlow received the highest diagnostic-quality scores across all tumor types (3.80 ± 0.45, 3.20 ± 0.45, and 2.60 ± 0.89 on a 5-point Likert scale), significantly outperforming all baseline methods (p < .05). Further studies showed T1C-RFlow to have the best tumor reconstruction performance and significantly faster denoising times (6.9 s volume^-1, 200 steps) than conventional DDPM models in both latent space (37.7 s, 1000 steps) and patch-based in image space (4.3 h volume^-1). Our proposed method generates synthetic T1C images that closely resemble radiological features of ground truth T1C in much less time than previous diffusion models. Further development may permit a practical method for contrast-agent-free MRI for brain tumors. Code is made available at https://github.com/zacheidex/An-Efficient-3D-Latent-Diffusion-Model-for-T1-contrast-Enhanced-MRI-Generation.

Purpose. This study introduces an Adaptive Optimization-based Multi-Material Decomposition with Total Nuclear Variation (AO-MMD-TNV) for dual-energy CT (DECT), designed to achieve accurate and noise-robust material fraction estimation.Methods. Three experimental datasets-a digital phantom, a tissue characterization phantom, and a human-shaped phantom-were used to evaluate the proposed method. The algorithm combines a robust Huber data term, iteratively reweighted L1 sparsity, and Total Nuclear Variation (TNV) regularization under simplex constraints. Adaptive weighting was applied to balance physical fidelity, sparsity, and boundary coherence. The performance was quantitatively compared with a conventional triplet-based MMD approach using metrics of volume fraction accuracy (VFA) and standard deviation (STD).Results. The AO-MMD-TNV achieved 100% VFA for all five materials in the digital phantom, with complete noise suppression (STD = 0). In the tissue characterization phantom, the method demonstrated improved VFA in most ROIs, though minor accuracy reductions were observed for breast and solid water (ROIs 3 and 4) due to their similar attenuation to background. In the human-shaped phantom, the AO-MMD-TNV maintained stable performance across six organ-related ROIs, outperforming MMD in quantitative consistency and noise robustness.Conclusions. The proposed AO-MMD-TNV framework effectively enhances quantitative reliability, reduces noise and crosstalk, and preserves anatomical boundaries. While further optimization is needed for low-contrast materials, AO-MMD-TNV demonstrates strong potential for precise, physically consistent DECT-based material quantification and clinical application.

Segmenting glioblastoma in medical imaging remains challenging due to the tumor's irregular shape, heterogeneous texture, and poorly defined boundaries, which often lead to inaccurate delineation by conventional methods. To address these challenges, we propose a novel segmentation framework that leverages Topological Data Analysis (TDA) to capture intrinsic topological features of gliomas, reducing reliance on large annotated datasets while improving structural fidelity. Our approach exploits TDA's interpretable filtrations and persistent homology alongside explicit spatial adjacency information to enhance segmentation accuracy. The method proceeds sequentially: (1) Whole Tumor (WT) segmentation via an Automated Thresholding algorithm on reverse filtration (R-filtration), (2) Enhancing Tumor (ET) segmentation using carefully selected persistent homology features, and (3) non-enhancing Tumor Core (TC) and Edema (ED) segmentation based on their spatial adjacency with ET through a criterion-driven iterative process. To quantify segmentation performance, we introduce a novel fuzzy Edge-Dice score applicable across all steps. Evaluated on the public BRATS2021 and BRATS2022-Reg datasets, our TDA-based framework achieves robust and accurate segmentation, highlighting the potential of TDA methods to complement or surpass conventional deep learning approaches in real-world medical imaging applications.

Objective. Diffusing alpha-emitters radiation therapy (Alpha-DaRT) involves implanting stainless-steel seeds containing 70-200 kBq of²²⁴Ra into solid tumors. The decay of²²⁴Ra generates alpha particles and short-lived alpha-emitting atoms, which diffuse and deposit absorbed dose up to millimeters from the seed. As magnetic res- onance imaging (MRI)-guided brachytherapy is commonly used in colorectal cancer, it has also been adopted in pre-clinical orthotopic intra-rectal studies to investigate the therapeutic potential of Alpha-DaRT. However, the metallic seeds cause metal artifacts in MRI, distorting signals and degrading images. Overcoming these artifacts is essential for accurate tumor assessment and treatment planning. This study aimed to reduce metal artifacts in images acquired with 7 T pre-clinical MRI in the presence of Alpha-DaRT seeds using an orthotopic colorectal adenocarcinoma animal model.Approach. Inert seeds were imaged in gelatin phantoms and the dorsal cavity of a mouse carcass. For thein-vivoanimal model, NOD scid gamma mice with HT-29 colorectal adenocarcinoma tumors (∼5-7 mm) were injected with inert seeds. MRI sequence parameters such as echo and repetition times, slice thickness, and readout bandwidth were tested using the phantom and refined with carcass imaging, leading toin-vivoimaging.Main results. Multiple sequence protocols were tested on gelatin phantoms and compared to the original T₂-weighted turbo spin echo (TSE) sequence protocol. An improved T₂-weighted TSE sequence protocol reduced metal artifact volumes in the gelatin phantom from 147.8 ± 31.0 mm³to 87.0 ± 22.4 mm³in the axial slices, resulting in a 41% reduction. Validation in the mouse carcass confirmed high-quality soft-tissue imaging.In-vivoimages with live mice showed a statistically significant reduction (p<0.001) in metal artifact volume with the improved sequence.Significance. A T₂-weighted turbo spin echo protocol effectively mitigated metal artifacts from Alpha-DaRT seeds in a colorectal adenocarcinoma animal model, allowing for clearer visualization of seed placement within the tumor and improving the accuracy of pre-clinical studies.

Transcription factor (TF) and target gene interactions are pivotal in gene regulatory networks, influencing molecular biology and disease mechanisms, particularly cancer. Dysregulated TFs contribute to aberrant gene expression, driving tumor progression. While experimental methods like ChIP-seq and RNA-seq provide valuable insights, their high cost and scalability constraints necessitate computational alternatives. Machine learning offers promising solutions, yet data imbalance remains a major challenge affecting predictive accuracy. This study introduces a novel approach integrating K-means++ clustering with a data balancing strategy to enhance TF-target interaction prediction. By selecting low-frequency TFs within clusters based on an inverse information principle, our method mitigates data bias and improves model generalization. Additionally, we incorporate deep learning with random walk sampling and skip-gram embeddings to extract informative representations of heterogeneous biological networks. Experimental results using five-fold cross-validation demonstrate superior performance, achieving an average AUC of 0.9452 ± 0.0047. Our framework enhances predictive accuracy while addressing data imbalance, offering significant applications in molecular biology and biomedical research for TF-target gene discovery and therapeutic development. .

Polycaprolactone (PCL), chitosan (CS), and hydroxyapatite (HA) have emerged as complementary biomaterials for the design of advanced scaffolds in tissue engineering (TE). Individually, PCL offers excellent mechanical strength and formability but suffers from hydrophobicity and slow degradation. CS provides biocompatibility, antibacterial properties, and favorable cell-material interactions, yet its insufficient mechanical stability limits standalone use. HA, a bioactive ceramic, enhances osteoconductivity; nevertheless, it is brittle in pure form. Recent advances focus on integrating these three components into hybrid composites to harness their desired characteristics. Novel fabrication approaches, including electrospinning and 3D printing have been optimized to tailor scaffold architecture, porosity, and mechanical integrity. Studies highlight enhanced cellular adhesion and differentiation, as well as improved angiogenic and antibacterial performance when functionalized with bioactive agents or nanoparticles. For instance, the incorporation of nano-HA into the PCL/CS scaffolds markedly boosted skin fibroblast cells (HSF 1184) proliferation, yielding a 23% increase compared to PCL/CS scaffolds by day 3. Besides, HA-PCL/CS nanofibrous composite scaffolds demonstrated a marked improvement in mechanical stiffness, showing an increase of greater than 15% in modulus of elasticity compared to the PCL/CS scaffold. Despite these advances, challenges remain in achieving controlled degradation, uniform dispersion of components, and scalable, reproducible fabrication for clinical translation. This current review fills a critical gap by providing the first comprehensive analysis of advancements in PCL-CS-HA ternary TE systems, an area that remains unexplored despite existing reviews on individual materials and their binary combinations. It analyzes latest developments in PCL-CS-HA composites, highlighting their structure, characteristics, processing strategies, biological outcomes, and future directions.

Low-Dose Computed Tomography (LDCT) reduces radiation risk but introduces high noise levels that compromises diagnostic quality. To address this, we propose a Hybrid Generalized Efficient Layer Aggregation Network-UNet (GELAN-UNet) model, which incorporates medical priors into a progressive modular architecture. This design uses medically enhanced modules in shallower layers to capture fine details and computationally efficient blocks in deeper layers to reduce cost. Key innovations include a novel low-frequency retention path and an edge-aware attention mechanism, both crucial for preserving critical diagnostic structures. Evaluated on the public Mayo Clinic dataset, the proposed method achieves a superior peak signal-to-noise ratio (PSNR) of 45.28 dB - a 12.45% improvement over the original LDCT - while maintaining an optimal balance between denoising performance and computational efficiency. The critical importance of the low-frequency path, as revealed by ablation studies, validates the rationality of the hybrid strategy, which is further supported by comparisons with full medical and frequency-aware variants. This work delivers a high-performance denoising model alongside a practical, efficient architectural paradigm - rigorously validated through systematic exploration - for integrating domain-specific medical knowledge into deep learning frameworks.

Purpose. This study presents a novel method for fabricating a heterogeneous, tissue-equivalent mouse phantom model using additive manufacturing, together with dosimetric verification for applications in dosimetry for pre-clinical radiation research.Methods. Local Artificial Body for Radiation Analysis and Testing (LABRAT^®) mouse phantoms were developed based on the Digimouse model. After 3D rendering, a mold-and-assemble method of additive manufacturing was done using 1:1.3 polyurethane-resin material for lung tissue, 1:1 resin-hardener mixture for soft tissue, and resin with 30% hydroxyapatite for bone. Three types of phantoms were developed: LABRAT A (full mouse), LABRAT B (with ionization chamber provision), and LABRAT C (with axial slices along the head, upper lung, lower lung, abdomen, and spine for film dosimetry). Ionization chamber measurements were performed on LABRAT B under total-body irradiation (TBI) (0.5-2.0 Gy) using 130 kVp, 5.0 mA x-rays at a 23 cm source-to-phantom distance on top of a 5 cm PMMA slab. Film calibration and 2.5 Gy TBI were also conducted on LABRAT C to obtain axial dose maps. Computed tomography (CT) images were obtained, and CT numbers of the phantoms were extracted using Slicer 5.4.0.Results. The fabrication method produced identical LABRAT^®phantoms suitable for pre-clinical dosimetry. In the open field plan, the measured dose for the LABRAT B phantom inside the acrylic mouse restrainer was observed to agree by up to ±2.6% of the prescribed dose. Film images revealed the corresponding dose maps in each axial slice, which show gradients corresponding to doses of 0 to 3 Gy. Mean CT numbers were -621 ± 119 HU (lung), 70 ± 40 HU (soft tissue), and 430 ± 138 HU (bone).Conclusion. A heterogeneous mouse phantom was successfully developed and validated for dose verification in pre-clinical irradiation. LABRAT^®materials demonstrated appropriate anatomical and radiological equivalence, with accurate dosimetric performance and good geometric agreement with the Digimouse model.

Idiopathic pulmonary fibrosis significantly threatens patient survival and remains a condition with limited effective treatment options. There is an urgent need to expedite the exploration of idiopathic pulmonary fibrosis mechanisms and identify suitable therapeutic approaches. Non-invasive and rapid segmentation of lung tissue, coupled with fibrosis quantification, is essential for drug development and efficacy monitoring. In this study, 59 mice were divided into training, validation and test sets according to the ratio of 70%:15%:15%. Based on this ratio, we performed a six-fold cross-validation to ensure the reliability of our results and calculated the average performance across all test sets. At first, a 2.5D UNet was utilized to segment the lung tissue of mice, followed by the calculation of a fibrosis score based on the segmented output, which can be used to evaluate the degree of pulmonary fibrosis in mice. Dice score, precision and recall are used to evaluated the performance of 2.5D UNet. In the test set, the 2.5D UNet achieved an average Dice score of 0.938, precision of 0.941, and recall of 0.936 across the six-fold cross-validation. The fibrosis score effectively demonstrated the varying impacts of different modeling or treatment methods. The 2.5D UNet can effectively segment mice lung tissue and evaluate fibrosis scores, which lays a solid foundation for further research.

Introduction.Focal cortical dysplasia type II (FCD II) is a significant cause of drug-resistant epilepsy, and the full surgical resection of the lesion is linked with excellent disease-free outcomes. Its imaging hallmark is the white matter hyperintense funnel-shaped transmantle sign on T2-FLAIR magnetic resonance imaging (MRI). Manual delineation of this abnormality is challenging and inconsistent. Most current artificial intelligence (AI) segmentation tools focus on cortical features and do not fully evaluate the white matter component. We tested whether integrating an algorithm trained on white matter lesions may improve FCD II segmentation.Methods.We evaluated the combination of two AI algorithms, MELD Graph (surface-based FCD segmentation) and MindGlide (whole-brain/white-matter lesion segmentation tool) in 49 FCD cases with a radiologically confirmed transmantle sign. Segmentation accuracy was assessed against expert manual annotations using the Dice similarity coefficient and segmentation volumes.Results.MELD Graph detected the lesion in 31 cases, 22 of which had the transmantle sign included in the expert lesion mask. Among these, MindGlide detected the transmantle sign in eight cases (36%). The mean added Dice score was 0.033 (95% CI, 0.013-0.056). Overall Dice values of MELD Graph were 0.321 and increased to 0.354 with the addition of MindGlide. It also contributed additional lesion volume in these eight cases, ranging from 0.028 to 4.18 cm³, with a mean added volume of 0.77 cm³.Discussion.Despite not being trained on FCD data, MindGlide, when combined with MELD Graph, provided a modest improvement in FCD II segmentation, including the deep white matter component of the lesion that is not captured by MELD Graph.Conclusion.These findings provide preliminary evidence supporting the consideration of a sequential cortical and white matter segmentation approach in FCD II, which may guide further epilepsy-specific AI model development.