Journal of Medical and Biological Engineering最新文献

Purpose

This study investigated hamstring activation in the frequency domain and jump-landing performance in a specialized-training athletic population and a healthy control group.

Methods

Thirty male athletes engaged in power training, both with and without jumping sports, or endurance training, together with ten healthy participants were recruited. Surface EMG electrodes were attached to the bellies of the lateral hamstring (LH) and medial hamstring (MH). The median EMG frequency was analyzed during takeoff, flight, before ground contact, after ground contact, and landing in countermovement jumps (CMJ) and drop-vertical jumps (DJ). Kinetic outcomes were also investigated.

Results

The power-trained athletes (with and without jumping sports) exhibited a lower median EMG frequency in the MH during takeoff (p = 0.001 for DJ) and in the LH (p = 0.008 for DJ) and MH during landing (p = 0.004 for CMJ and 0.001 for DJ) compared with the endurance-trained or control groups. Furthermore, the power-trained group demonstrated greater jump heights (p = 0.009 for CMJ and p = 0.003 for DJ). All the athletic groups showed a lower landing force (p = 0.022) and loading rate (p = 0.043) in CMJ than the control group.

Conclusion

Training background differences influenced hamstring recruitment during jumping. Power-trained athletes exhibited a lower median EMG frequency and better jumping performance. All the athletes demonstrated a more effective landing strategy than the control group. These findings suggest the potential for enhancing athletic performance and aiding in landing strategy by exploiting different training styles.

Purpose

Biopolymeric materials, especially composites, are extensively used as wound healing scaffolds in tissue engineering due to their ability to mimic the essential properties of the native tissue. This research aims to investigate the usability of honeybee silk (HS), which could be an alternative silk source to silkworm silk, in tissue engineering (TE) applications. HS, which has not been used in scaffold fabrication, and chitosan (CH), frequently used in TE, were combined to produce a novel and cost-effective biocompatible CH–HS scaffold.

Methods

HS, CH and CH–HS were characterized using XRD, FTIR and SEM to determine structure and functional groups. SEM analysis was performed for different CH concentrations (0.5%, 1% and 2%) and different ratios of CH:HS (1:2, 1:1 and 2:1, respectively). The antioxidant properties, antibacterial activity and as well as biofilm formation and ability to destroy mature biofilm activity of HS and CH–HS were shown. The human breast cancer MDA-MB231 cells were used to investigate possible effects on cell viability proliferation.

Results

The smallest pore size was determined to be 70.7 µm on average at a ratio of 1:1 at 1% CH concentration. The antioxidant properties of HS and CH–HS were shown. The CH–HS showed antibacterial activity against Escherichia coli and Pseudomonas aeruginosa, as well as inhibition of biofilm formation and destruction of mature biofilm. Additionally, the MDA-MB-231 cells appeared significantly elongated and denser when seeded on the CH–HS over 24 h and 48 h.

Conclusion

This study demonstrated the usability of honeybee silk, a promising but underutilized material, tissue engineering and its potential for future studies. Considering the materials used and our promising results, the synthesized CH–HS scaffold was observed to have microbiological and cellular effects that may be useful in future biomedical applications for wound healing.

Purpose

To improve the performance of deep-learning-based image segmentation, a sufficient amount of training data is required. However, it is more difficult to obtain training images and segmentation masks for medical images than for general images. In deep-learning-based colon polyp detection and segmentation, research has recently been conducted to improve performance by generating polyp images using a generative model, and then adding them to training data.

Methods

We propose SemanticPolypGAN for generating colonoscopic polyp images. The proposed model can generate images using only the polyp and corresponding mask images without additional preparation of input condition. In addition, the semantic generation of the shape and texture of polyps and non-polyp parts is possible. We experimentally compare the performance of various polyp-segmentation models by integrating the generated images and masks into the training data.

Results

The experimental results show improved overall performance for all models and previous work.

Conclusion

This study demonstrates that using polyp images generated by SemanticPolypGAN as additional training data can improve polyp segmentation performance. Unlike existing methods, SemanticPolypGAN can independently control polyp and non-polyp parts in a generation.

Purpose

To evaluate the diagnostic performance of low-keV virtual monoenergetic imaging (VMI) using dual-energy CT (DECT) with deep learning image reconstruction (DLIR) in patients with hepatocellular carcinoma (HCC).

Methods

This retrospective study included patients with HCC undergoing DECT scans between February 2019 and March 2022. VMI was reconstructed with hybrid iterative reconstruction (HIR) at 70-keV (HIR70keV) and 40-keV (HIR40keV) and DLIR at 40-keV (DLIR40keV). Two radiologists calculated the contrast-to-noise ratio (CNR) of the HCC. The possible presence of HCC was assessed by two additional radiologists. CNR was compared using Friedman’s test. Diagnostic performance was compared between three groups using Cochran’s Q test and jackknife alternative free-response receiver operating characteristic analysis.

Results

Thirty-two patients (mean age 73.19 ± 11.86, 23 males) with 36 HCCs were enrolled. The CNR of DLIR40keV was significantly higher than HIR70keV and HIR40keV (p < 0.001 and 0.001). The sensitivities for the detection of HCC were HIR70keV, 63.9%; HIR40keV, 72.2%; DLIR40 keV, 83.3%, and HIR70keV, 52.8%; HIR40keV, 61.1%; DLIR40 keV, 77.8% for observers 1 and 2, respectively. DLIR40keV sensitivity was significantly higher than HIR70keV on both readers (p = 0.020 and 0.013). The figures of merit (FOM) were HIR70keV, 0.86; HIR40keV, 0.92; DLIR40 keV, 0.96, and HIR70keV, 0.84; HIR40keV, 0.90; and DLIR40 keV, 0.94 for observers 1 and 2, respectively. For both observers, DLIR40keV FOM was significantly higher than HIR70keV (p = 0.013 and 0.012).

Conclusion

DLIR40keV achieved the best CNR among the three groups. HCC detectability was significantly improved at DLIR40keV compared to HIR70keV.

Abstract

Purpose

Radial extracorporeal shock wave therapy (rESWT) has been used in clinical and rehabilitation fields. However, the formulation of related clinical treatment protocols and the full potential of its therapeutic efficacy are constrained due to limited understanding of shock wave sources. This study aimed to further clarify the characteristics of shock wave sources generated at different medium interfaces.

Methods

Shock wave generated by rESWT device at the interface of different media (soft tissue-mimicking-phantom, water and air) was measured based on flexible polyvinylidene fluoride (PVDF) sensors. The temporal and spectral characteristics of the shock wave source were analyzed.

Results

The wave generated at the phantom interface was similar to that at the water interface under the same impact pressure, both being largely different from that at the air interface, where the absolute value of the peak pressure was significantly reduced. The spectral properties of the shock wave generated in different media were similar, with distinct peak frequencies, varying modulation frequencies in phantom (12.2 kHz), water (8.5 kHz), and air (7.2 kHz), and a relatively constant carrier frequency (between 82 and 83 kHz). Under the different impact pressures, there were no variations in the peak frequency at the same medium interface, indicating that the impact pressure mainly impacts the shock wave amplitude, but not the peak frequency.

Conclusion

The shock waves generated at different medium interfaces exhibited temporal and spectral differences. Therefore, measurement results in biological soft tissues cannot be simply replaced by the measurement results in air or water. The results of this study are expected to provide important information for evaluating rESWT devices and optimizing clinical shock wave treatment protocols.

Abstract

Purpose

Non-invasive fetal electrocardiography (fECG) offers crucial information for assessing early diagnosis of fetal distress and morbidity. However, the non-invasive fECG signals probably contain various non-stationary noises, which may generate a bad influence on signal processing. Signal quality assessment plays a crucial role in accurate feature estimation for obtaining high-quality signals.

Methods

This study develops a comprehensive framework for the assessment of signal quality for non-invasive fECG signals. Firstly, the ECG collection equipment is employed to gather abdominal ECG signal data from eight pregnant women in the hospital. Secondly, signal preprocessing is operated including signal segmentation and data normalization process. Subsequently, a total of thirty-seven signal quality indexes (SQIs) are extracted which consist of the amplitude-based SQI, R-wave-based SQI, statistical-based SQI, fractal dimension SQI, power spectrum distribution-based SQI, and entropy domain-based SQI. Then, in order to reduce the dimensionality of features and improve the experimental performance, information gain is carried out to identify the subset of the optimal features. At last, the classifier combines different feature numbers to classify the quality of the non-invasive fECG signal.

Results

Ten classifiers are selected to perform a classification task between good-quality and bad-quality abdominal signals. The experimental results show that the combination of twenty-four effective features and random forest achieved the highest classification outcome, which in terms of the ACC, and F1 scores are 0.9508, and 0.9510, respectively.

Conclusion

The experimental results indicate that our work can reliably assess the signal quality for non-invasive fECG signals and filter out good-quality signals. This proposed algorithm can help to improve the accuracy of fetal signal extraction and fetal heart rate estimation for further analysis, which is beneficial to promoting fetal health monitoring.

Purpose

The purpose of this study is to confirm whether it is possible to acquire a certain degree of diagnostic ability even with a small dataset using domain-specific transfer learning. In this study, we constructed a simulated caries detection model on panoramic tomography using transfer learning.

Methods

A simulated caries model was trained and validated using 1094 trimmed intraoral images. A convolutional neural network (CNN) with three convolution and three max pooling layers was developed. We applied this caries detection model to 50 panoramic images and evaluated its diagnostic performance.

Results

The diagnostic performance of the CNN model on the intraoral film was as follows: C0 84.6%; C1 90.6%; C2 88.6%. Finally, we tested 50 panoramic images with simulated caries insertion. The diagnostic performance of the CNN model on the panoramic image was as follows: C0 75.0%, C1 80.0%, C2 80.0%, and overall diagnostic accuracy was 78.0%. The diagnostic performance of the caries detection model constructed only with panoramic images was much lower than that of the intraoral film.

Conclusion

Domain-specific transfer learning methods may be useful for saving datasets and training time (179/250).

Abstract