Biomedical Physics & Engineering Express最新文献

This research presents an integrated framework designed to automate the classification of pulmonary chest x-ray images. Leveraging convolutional neural networks (CNNs) with a focus on transformer architectures, the aim is to improve both the accuracy and efficiency of pulmonary chest x-ray image analysis. A central aspect of this approach involves utilizing pre-trained networks such as VGG16, ResNet50, and MobileNetV2 to create a feature ensemble. A notable innovation is the adoption of a stacked ensemble technique, which combines outputs from multiple pre-trained models to generate a comprehensive feature representation. In the feature ensemble approach, each image undergoes individual processing through the three pre-trained networks, and pooled images are extracted just before the flatten layer of each model. Consequently, three pooled images in 2D grayscale format are obtained for each original image. These pooled images serve as samples for creating 3D images resembling RGB images through stacking, intended for classifier input in subsequent analysis stages. By incorporating stacked pooling layers to facilitate feature ensemble, a broader range of features is utilized while effectively managing complexities associated with processing the augmented feature pool. Moreover, the study incorporates the Swin Transformer architecture, known for effectively capturing both local and global features. The Swin Transformer architecture is further optimized using the artificial hummingbird algorithm (AHA). By fine-tuning hyperparameters such as patch size, multi-layer perceptron (MLP) ratio, and channel numbers, the AHA optimization technique aims to maximize classification accuracy. The proposed integrated framework, featuring the AHA-optimized Swin Transformer classifier utilizing stacked features, is evaluated using three diverse chest x-ray datasets-VinDr-CXR, PediCXR, and MIMIC-CXR. The observed accuracies of 98.874%, 98.528%, and 98.958% respectively, underscore the robustness and generalizability of the developed model across various clinical scenarios and imaging conditions.

Background.Using radar for non-contact measuring human vital signs has garnered significant attention due to its undeniable benefits. However, achieving reasonably good accuracy in contactless measurement senarios is still a technical challenge.Materials and methods.The proposed method includes two stages. The first stage involves the process of datasegmentation and signal channel selection. In the next phase, the raw radar signal from the chosen channel is subjected to modified Pan-Tompkins.Results.The experimental findings from twelve individuals demonstrated a strong agreement between the contactless radar and contact electrocardiography (ECG) devices for heart rate measurement, with correlation coefficient of 98.74 percentage; and the 95% limits of agreement obtained by radar and those obtained by ECG were 2.4 beats per minute.Conclusion.The results showed high agreement between heart rate calculated by radar signals and heart rate by electrocardiograph. This research paves the way for future applications using non-contact sensors to support and potentially replace contact sensors in healthcare.

Prolonged sleeping postures or unusual postures can lead to the development of various ailments such as subacromial impingement syndrome, sleep paralysis in the elderly, nocturnal gastroesophageal reflux, sore development, etc Fibre Bragg Gratings (a variety of optical sensors) have gained huge popularity due to their small size, higher sensitivity and responsivity, and encapsulation flexibilities. However, in the present study, FBG Arrays (two FBGs with 10 mm space between them) are employed as they are advantageous in terms of data collection, mitigating sensor location effects, and multiplexing features. In this work, Liquid silicone encapsulated FBG arrays are placed in the head (E), shoulder (C, D), and lower half body (A, B) region for analyzing the strain patterns generated by different sleeping postures namely, Supine (P1), Left Fetus (P2), Right Fetus (P3), and Over stomach (P4). These strain patterns were analyzed in two ways, combined (averaging the data from each FBG of the array) and Individual (data from each FBG was analyzed separately). Both analyses suggested that the FBGs in the arrays responded swiftly to the strain changes that occurred due to changes in sleeping postures. 3D histograms were utilized to track the strain changes and analyze different sleeping postures. A discussion regarding closely related postures and long hour monitoring has also been included. Arrays in the lower half (A, B) and shoulder (C, D) regions proved to be pivotal in discriminating body postures. The average standard deviation of strain for the different arrays was in the range of 0.1 to 0.19 suggesting the reliable and appreciable strain-handling capabilities of the Liquid silicone encapsulated arrays.

Objective. To estimate dose rates delivered by using radioactive¹⁹⁸Au nanoparticles for prostate cancer nanobrachytherapy, identifying contribution by photons and electrons emmited from the source.Approach. Utilizingin silicomodels, two different anatomical representations were compared: a mathematical model and a unstructured mesh model based on the International Commission on Radiological Protection (ICRP) Publication 145 phantom. Dose rates by activity were calculated to the tumor and nearby healthy tissues, including healthy prostate tissue, urinary bladder wall and rectum, using Monte Carlo code MCNP6.2.Main results. Results indicate that both models provide dose rate estimates within the same order of magnitude, with the mathematical model overestimating doses to the prostate and bladder by approximately 20% compared to the unstructured mesh model. The discrepancies for the tumor and rectum were below 4%. Photons emmited from the source were defined as the primary contributors to dose to other organs, while 97.9% of the dose to the tumor was due to electrons emmited from the source.Significance. Our findings emphasize the importance of model selection in dosimetry, particularly the advantages of using realistic anatomical phantoms for accurate dose calculations. The study demonstrates the feasibility and effectiveness of¹⁹⁸Au nanoparticles in achieving high dose concentrations in tumor regions while minimizing exposure to surrounding healthy tissues. Beta emissions were found to be predominantly responsible for tumor dose delivery, reinforcing the potential of¹⁹⁸Au nanoparticles in localized radiation therapy. We advocate for using realistic body phantoms in further research to enhance reliability in dosimetry for nanobrachytherapy, as the field still lacks dedicated protocols.

Backgrounds. Virtual reality (VR) simulates real-life events and scenarios and is widely utilized in education, entertainment, and medicine. VR can be presented in two dimensions (2D) or three dimensions (3D), with 3D VR offering a more realistic and immersive experience. Previous research has shown that electroencephalogram (EEG) profiles induced by 3D VR differ from those of 2D VR in various aspects, including brain rhythm power, activation, and functional connectivity. However, studies focused on classifying EEG in 2D and 3D VR contexts remain limited.Methods. A 56-channel EEG was recorded while visual stimuli were presented in 2D and 3D VR. The recorded EEG signals were classified using two machine learning approaches: traditional machine learning and deep learning. In the traditional approach, features such as power spectral density (PSD) and common spatial patterns (CSP) were extracted, and three classifiers-support vector machines (SVM), K-nearest neighbors (KNN), and random forests (RF)-were used. For the deep learning approach, a specialized convolutional neural network, EEGNet, was employed. The classification performance of these methods was then compared.Results. In terms of accuracy, precision, recall, and F1-score, the deep learning method outperformed traditional machine learning approaches. Specifically, the classification accuracy using the EEGNet deep learning model reached up to 97.86%.Conclusions. EEGNet-based deep learning significantly outperforms conventional machine learning methods in classifying EEG signals induced by 2D and 3D VR. Given EEGNet's design for EEG-based brain-computer interfaces (BCI), this superior classification performance suggests that it can enhance the application of 3D VR in BCI systems.

Synovial sarcoma (SS) is a rare cancer that forms in soft tissues around joints, and early detection is crucial for improving patient survival rates. This study introduces a convolutional neural network (CNN) using an improved AlexNet deep learning classifier to improve SS diagnosis from digital pathological images. Key preprocessing steps, such as dataset augmentation and noise reduction techniques, such as adaptive median filtering (AMF) and histogram equalization were employed to improve image quality. Feature extraction was conducted using the Gray-Level Co-occurrence Matrix (GLCM) and Improved Linear Discriminant Analysis (ILDA), while image segmentation targeted spindle-shaped cells using repetitive phase-level set segmentation (RPLSS). The improved AlexNet architecture features additional convolutional layers and resized input images, leading to superior performance. The model demonstrated significant improvements in accuracy, sensitivity, specificity, and AUC, outperforming existing methods by 3%, 1.70%, 6.08%, and 8.86%, respectively, in predicting SS.

With the development of deep learning in the field of medical image segmentation, various network segmentation models have been developed. Currently, the most common network models in medical image segmentation can be roughly categorized into pure convolutional networks, Transformer-based networks, and networks combining convolution and Transformer architectures. However, when dealing with complex variations and irregular shapes in medical images, existing networks face issues such as incomplete information extraction, large model parameter sizes, high computational complexity, and long processing times. In contrast, models with lower parameter counts and complexity can efficiently, quickly, and accurately identify lesion areas, significantly reducing diagnosis time and providing valuable time for subsequent treatments. Therefore, this paper proposes a lightweight network named MCI-Net, with only 5.48 M parameters, a computational complexity of 4.41, and a time complexity of just 0.263. By performing linear modeling on sequences, MCI-Net permanently marks effective features and filters out irrelevant information. It efficiently captures local-global information with a small number of channels, reduces the number of parameters, and utilizes attention calculations with exchange value mapping. This achieves model lightweighting and enables thorough interaction of local-global information within the computation, establishing an overall semantic relationship of local-global information. To verify the effectiveness of the MCI-Net network, we conducted comparative experiments with other advanced representative networks on five public datasets: X-ray, Lung, ISIC-2016, ISIC-2018, and capsule endoscopy and gastrointestinal segmentation. We also performed ablation experiments on the first four datasets. The experimental results outperformed the other compared networks, confirming the effectiveness of MCI-Net. This research provides a valuable reference for achieving lightweight, accurate, and high-performance medical image segmentation network models.

The escalating prevalence of diabetes mellitus underscores the critical need for non-invasive screening tools capable of early disease detection. Present diagnostic techniques depend on invasive procedures, which highlights the need for advancement of non-invasive alternatives for initial disease detection. Machine learning in integration with the optical sensing technology can effectively analyze the signal patterns associated with diabetes. The objective of this research is to develop and evaluate a non-invasive optical-based method combined with machine learning algorithms for the classification of individuals into normal, prediabetic, and diabetic categories. A novel device was engineered to capture real-time optical vascular signals from participants representing the three glycemic states. The signals were then subjected to quality assessment and preprocessing to ensure data reliability. Subsequently, feature extraction was performed using time-domain analysis and wavelet scattering techniques to derive meaningful characteristics from the optical signals. The extracted features were subsequently employed to train and validate a suite of machine learning algorithms. An ensemble bagged trees classifier with wavelet scattering features and random forest classifier with time-domain features demonstrated superior performance, achieving an overall accuracy of 86.6% and 80.0% in differentiating between normal, prediabetic, and diabetic individuals based on the optical vascular signals. The proposed non-invasive optical-based approach, coupled with advanced machine learning techniques, holds promise as a potential screening tool for diabetes mellitus. The classification accuracy achieved in this study warrants further investigation and validation in larger and more diverse populations.

Objective. Methods of detecting apnea based on acoustic features can be prone to misdiagnosed and missed diagnoses due to the influence of noise. The aim of this paper is to improve the performance of apnea detection algorithms in the Post Anesthesia Care Unit (PACU) using a denoising method that processes tracheal sounds without the need for separate background noise.Approach. Tracheal sound data from laboratory subjects was collected using a microphone. Record a segment of clinical background noise and clean tracheal sound data to synthesize the noisy tracheal sound data according to a specified signal-to-noise ratio. Extract the frequency-domain features of the tracheal sounds using the Short Time Fourier Transform (STFT) and input the Cascaded Redundant Convolutional Encoder-Decoder network (CR-CED) network for training. Patients' tracheal sound data collected in the PACU were then fed into the CR-CED network as test data and inversely transformed by STFT to obtain denoised tracheal sounds. The apnea detection algorithm was used to detect the tracheal sound after denoising.Results. Apnea events were correctly detected 207 times and normal respiratory events 11,305 times using tracheal sounds denoised by the CR-CED network. The sensitivity and specificity of apnea detection were 88% and 98.6%, respectively.Significance. The apnea detection results of tracheal sounds after CR-CED network denoising in the PACU are accurate and reliable. Tracheal sound can be denoised using this approach without separate background noise. It effectively improves the applicability of the tracheal sound denoising method in the medical environment while ensuring its correctness.

Postural balance is crucial for daily activities, relying on the coordination of sensory systems. Balance impairment, common in the elderly, is a leading cause of mortality in this population. To analyze balance, methods like postural adjustment analysis using electromyography (EMG) have been developed. With age, women tend to experience reduced mobility and greater muscle loss compared to men. However, few studies have focused on postural adjustments in women of different ages using EMG of the lower limbs during laterolateral and anteroposterior movements. This gap could reveal a decrease in muscle activation time with aging, as activation time is vital for postural adjustments. This study aimed to analyze muscle activation times in women of different ages during postural adjustments. Thirty women were divided into two groups: young and older women. A controlled biaxial force platform was used for static and dynamic balance tests while recording lower limb muscle activity using EMG. Data analysis focused on identifying muscle activation points and analyzing postural adjustment times. Results showed significant differences in muscle activation times between the two groups across various muscles and platform tilt conditions. Younger women had longer muscle activation times than older women, particularly during laterolateral platform inclinations. In anteroposterior movements, older women exhibited longer activation times compared to their laterolateral performance, with fewer differences between the groups. Overall, older women had shorter muscle activation times than younger women, suggesting a potential indicator of imbalance and increased fall risk.