Biomedical Physics & Engineering Express最新文献

Gastrointestinal (GI) endoscopy serves as a vital tool for assessing the GI tract and diagnosing related disorders. Recent progress in deep learning has shown significant improvements in identifying anomalies using sophisticated models and data augmentation strategies. This study introduces an enhanced approach to improve classification accuracy using 8,000 labeled endoscopic images from the Kvasir dataset, categorized into eight distinct classes. Leveraging EfficientNetB3 as the backbone, our proposed architecture eliminates the reliance on data augmentation while maintaining moderate model complexity. Our model achieves a test accuracy of 94.25%, alongside precision and recall of 94.29% and 94.24%, respectively. Furthermore, Local Interpretable Model-agnostic Explanation (LIME) saliency maps are employed to enhance interpretability by highlighting critical regions in the images that influence model predictions. To facilitate real-world usability, a user-friendly interface was developed using Gradio, enabling users to upload images, generate predictions, view confidence levels, and maintain a history of past results. This work underscores the importance of integrating high classification accuracy, interpretability, and accessibility in advancing medical imaging applications.

Electroencephalogram (EEG)-based emotion recognition holds great potential in affective computing, mental health assessment, and human-computer interaction. However, EEG signals are non-stationary, noisy, and composed of multiple frequency bands, making direct feature learning from raw data particularly challenging. While end-to-end models alleviate the need for manual feature engineering, advancing the performance frontier of lightweight architectures remains a crucial and complex challenge for practical deployment. To address these issues, we propose LMSA-Net (Lightweight Multi-Scale Attention Network), a lightweight, interpretable, and end-to-end model that directly learns spatio-temporal features from raw EEG signals. The architecture integrates learnable channel weighting for adaptive spatial encoding, multi-scale temporal separable convolution for rhythm-specific feature extraction, and Sim Attention Module for parameter-free saliency enhancement. Our proposed LMSA-Net is evaluated on three benchmark datasets, SEED, SEED-IV, and DEAP, under subject-dependent protocols. It achieves top performance on SEED (65.53% accuracy), competitive results on SEED-IV (48.52% accuracy), and strong performance in arousal classification on DEAP, demonstrating good generalization. Ablation studies confirm the critical role of each proposed module. Frequency analysis reveals that our multi-scale temporal kernels inherently specialize in distinct EEG rhythms, validating their neurophysiological alignment. Lightweight design is evidenced by minimal parameters (7.64K) and low latency, ideal for edge deployment. Interpretability analysis further shows the model's focus on emotion-related brain regions. LMSA-Net thus delivers an efficient, interpretable, and high-performing solution. The code is available at https://github.com/rhr0411/LMSA-Net.git.

Objective: In photon beam radiotherapy, modern delivery techniques have allowed to substantially reduce the beam energy needed for the safe and efficient irradiation of deep-seated targets, with even Co-60 beams being now able to irradiate targets at any depth. The purpose of this work is to determine if for electron radiotherapy, advanced beam delivery techniques allow the use of beam energies currently available in the clinic to treat target sites usually reserved for photons or very high energy charged particles.

Methods: Segmented computed tomography images from three sites, brain, lung and prostate, were used to model radiotherapy treatments in two modalities: conformal 3D and converging small field. Monte Carlo simulation was used to calculate the absorbed dose distribution in each patient for conformal 3D very-high energy plans and converging small-field, low energy (< 50 MeV) electrons. For comparison, converging small field plans for 6 MV x-ray beams were also calculated.

Main results: It is shown that, for the three test cases simulated in this work, electrons with energies in the 20 -25 MeV range delivered via the converging small-field modality can produce treatment plans that rival those obtained via conformal very high energy electrons in terms of target dose homogeneity and sparing of the organs at risk. Furthermore, such electron plans also compare well to those obtained with the photon beams.

Significance: While the consensus has always been that to reach deeper tumors, higher electron energies, in the order of 150 -200 MeV are needed, this work shows that this is not the case and, when small, circular electron fields are delivered in a converging manner, energies below 30 MeV are enough to properly irradiate tumors located at relevant radiological depths for a variety of treatment sites.

This study aims to investigate the impact of position-dependent CT number variations and image bit-depth, which determines the range of encodable CT numbers, on relative electron density (RED) conversion, a process essential for accurate radiation dose calculation, for patients with high-Z materials or metal implants. A commercially available phantom containing tissue-equivalent rods and metal implant inserts was scanned using a clinical CT simulator at nine positional offsets (isocenter, ±6.5 cm, and ±10.5 cm in lateral and anterior-posterior directions). CT images were acquired with both 12-bit and 16-bit scanning modes under identical scanning conditions. HU values for each insert were measured across positions and converted to RED using the clinical HU-to-RED calibration curve. HU deviations increased with positional displacement, particularly in high-density materials such as titanium and stainless steel. In 12-bit datasets, HU saturation at 3071 HU led to errors in the RED underestimation, while 16-bit imaging preserved accurate HU values even at off-center positions. Off-center positioning and bit-depth limitations significantly affect HU representation and RED conversion, especially with high-Z materials. Adoption of 16-bit acquisition or manual assignment of electron density values is recommended to prevent HU saturation and further ensure robust dose calculation. These findings highlight the importance of imaging bit depth and patient centering as part of routine radiotherapy quality assurance, particularly for cases involving metallic implants or off-isocenter targets.

On a global scale, cancer ranks high in mortality rate. There is a need for better technology since the current treatments are insufficient. There have been several new cancer treatments that have gone straight from the lab to the clinic, but the manufacturing of nanomedicine products, made possible by the fast expansion of nanotechnology, has enormous potential for enhancing cancer treatment approaches. The advent of nanotechnology has opened the door to the possibility of multi-functionality and very precise targeting strategies. They have the potential to enhance the pharmacodynamic and pharmacokinetic profiles of conventional treatment approaches, which might lead to a reevaluation of the effectiveness of current anti-cancer drugs. A novel technique to enhance traditional onco-immunotherapies, recruiting nanoparticle-based delivery systems, which are adaptable carriers for a broad range of molecular payloads. The delivery of molecular payloads to the target site and their release may be well-regulated. We summarize the latest developments in nanobiotechnology for improving immunotherapies and reshaping tumour microenvironments (TMEs) in this review. The current clinical challenges that impede the real-time implementation of cancer nanomedicine are discussed, and this review study consolidates existing knowledge and recent advancements in the use of nanoparticles for cancer therapy. This provides researchers, clinicians, and students with a comprehensive understanding of the current state of the field. Finally, potential future directions are highlighted to enhance the therapeutic efficacy and facilitate the clinical translation of cancer nanomedicine.

In the process of treating bacterial infections, the excessive use of antibiotics can easily lead to the development of antibiotic resistance, which results in serious public health problems. In this context, lightmediated antibacterial strategies have attracted considerable attention due to their rapid bactericidal effects and low likelihood of inducing resistance.Carbon quantum dots (CQDs) are primarily composed of carbon, hydrogen, and oxygen, and are highly suitable as photosensitizers in photodynamic therapy due to their low toxicity and good biocompatibility. In this work, we used gallic acid as a precursor and employs a microwave confined heating strategy to achieve large-scale preparation of high-quality nitrogen-doped CQDs (GA-CQDs) within 5 min (6 g per batch). System analysis indicates that the GA-CQDs exhibit a uniform particle size distribution, stable energy level structure, and excellent photostability. It is noteworthy that low-concentration GA-CQDs can efficiently generate reactive oxygen species under the illumination of white light LEDs, thereby exhibiting excellent photodynamic antibacterial effects against Streptococcus mutans, which are significantly higher than using gallic acid alone. In summary, this work presents a simple method for converting antibiotics into CQDs, endowing them with broader application prospects in the field of antimicrobial agents.

This paper presents a lightweight hybrid framework that integrates a Haar-initialized Parametric Wavelet Transform (PWT) with a Convolutional Neural Network (CNN) enhanced by a multi-head Self-Attention mechanism for efficient and interpretable tumor identification from compressed Magnetic Resonance Imaging (MRI) brain image data. In this study, a Parametric Wavelet Transform (PWT) layer was proposed for efficient compression and adaptive feature extraction from brain MRI images. Initialized with Haar wavelet filters, the PWT layer is trainable, enabling the model to learn optimal frequency decompositions directly from the data while preserving critical diagnostic features. MRI images are preprocessed through this PWT layer to selectively extract and stack the approximation (cA) and diagonal detail (cD) subbands, effectively reducing spatial redundancy and enhancing the representation of diagnostically salient structures for downstream classification. A custom lightweight CNN backbone extracts local features from the frequency-domain representations. At the same time, the integrated self-attention module learns relevant patterns and improves the discriminative power across wavelet-transformed inputs. Model interpretability is addressed using Grad-CAM visualizations, which highlight tumor-relevant regions, thereby improving transparency and clinical trust. Based on the experimental findings, the proposed framework achieves a classification accuracy of 96.0%, outperforming benchmark architectures such as MobileNetV2 (93.0%) and MobileNetV3Small (95.2%) while maintaining fewer trainable parameters (~2.8 million) and achieving faster training time. An ablation study was conducted to evaluate the individual contributions of the PWT compression, CNN backbone, and self-attention module, confirming the additive benefit of each component in achieving optimal performance. The model is successfully deployed on a Raspberry Pi 5, confirming its suitability for real-time, point-of-care, edge-based medical imaging applications. Overall, this work introduces a novel combination of adaptive frequency-domain compression, attention-driven refinement, and efficient embedded deployment for robust and interpretable brain tumor classification.

This study aimed to experimentally investigate the cell survival responses of tumor and normal cell lines to spatially heterogeneous carbon ion dose distributions with varying peak-to-valley dose ratios (PVDRs) and linear energy transfer (LET) conditions, and to assess the utility of equivalent uniform dose (EUD) as a quantitative metric for analyzing these responses. HSGc-C5 (tumor) and Nuli-1 (normal tissue) cell lines were irradiated using carbon ion beams with different spatial dose patterns (Grid, Frame, Half) and two PVDR levels under low LET conditions (~10 keV/μm). Additionally, high LET (~50 keV/μm) Spread Out Bragg Peak (SOBP) Grid patterns were used for the HSGc-C5 cells. Clonogenic survival assays were performed to evaluate biological response. Survival data were analyzed both as a function of delivered physical dose and EUD, calculated using an LQ model-based formulation. Tumor cells exhibited enhanced cytotoxic effects under high LET and high PVDR conditions (the dose required to reach SF = 0.1 was approximately 40% lower at PVDR = 4.0 and 10% lower at PVDR = 1.64 compared with the simulation results), whereas normal cells showed a slight sparing effect under low LET irradiation. Even at the same total dose and PVDR, different spatial dose patterns produced measurable differences in survival, underscoring the impact of spatial heterogeneity. EUD-based analysis further enabled quantitative comparison between heterogeneous and uniform dose distributions. These findings indicate that spatial dose heterogeneity and LET can be leveraged to enhance tumor control while reducing normal tissue damage in carbon ion therapy. The EUD approach may offer a practical tool for treatment plan evaluation in spatially modulated particle therapy.

Higher levels of PEEP are suspected to induce right heart dysfunction due to increased pulmonary vascular resistance (PVR). A U-shaped correlation of PVR and lung volume has been shown in animal models, with PVR increasing with lower and higher lung volumes. This physiological study aims to investigate the relation of transpulmonary pressure and PVR. Recruited healthy subjects underwent mask continuous airway pressure (CPAP), while esophageal manometry and echocardiographic assessment of PVR were performed. Of 43 screened subjects, 20 were identified in whom echocardiographic estimation of PVR was possible. During CPAP, echocardiographic PVR was lowest when transpulmonary pressures were close to 0 mbar, and increased as transpulmonary pressures became more positive, with a positive monotonic correlation (ρ = 0.337, p = 0.012). PVR with a transpulmonary pressure of 0 mbar was similar to PVR without CPAP (1.4 WU (IQR 1.3-1.5) vs. 1.2 WU (IQR 1.2-1.5), p = 0.069). Our findings suggest that PVR could be lowest when airway pressure does not exceed intrathoracic pressure. Future studies should investigate this relationship in ventilated patients. Echocardiography might be suitable to monitor PVR in the presence of sufficiently traceable tricuspid regurgitation, however validation in ventilated patients is needed to determine clinical applicability.

This study aimed to assess a visual‑tactile breath‑hold (BH) workflow integrated with Elekta Unity's comprehensive motion management (CMM) system for gated MR‑guided radiotherapy in situations where verbal coaching is impractical. A visual guidance program and a 3D‑printed couch‑mounted tactile pointer were implemented to instruct patients and stabilize voluntary BH. Two patients, one with pancreatic cancer and one with lung cancer, were treated using this workflow. Treatment beam gating was driven by CMM BH criteria, and audit log files from CMM‑guided treatments were analyzed. Expected gating efficiencies were 40% for the pancreas case and 51.4% for the lung case, while measured efficiencies were 42.59 ± 2.56% and 54.95 ± 0.54%, respectively. The corresponding beam‑on times were 14.75 ± 0.96 and 16.25 ± 0.50 minutes. The workflow reduced reliance on motion prediction for gating and mitigated frequent beam holds typically observed with free‑breathing strategies, thereby decreasing dosimetric uncertainty. These findings indicate that a visual‑tactile BH workflow on a 1.5 T MR‑Linac is feasible and practical, supporting efficient gated delivery and reproducible breath‑holds when verbal coaching is limited.