Dilated Heterogeneous Convolution for Cell Detection and Segmentation Based on Mask R-CNN.

Owing to the variable shapes, large size difference, uneven grayscale, and dense distribution among biological cells in an image, it is very difficult to accurately detect and segment cells. Especially, it is a serious challenge for some microscope imaging devices with limited resources owing to a large number of learning parameters and computational burden when using the standard Mask R-CNN. In this work, we propose a mask R-DHCNN for cell detection and segmentation. More specifically, Dilation Heterogeneous Convolution (DHConv) is proposed by designing a novel convolutional kernel structure (i.e., DHConv), which integrates the strengths of the heterogeneous kernel structure and dilated convolution. Then, the traditional homogeneous convolution structure of the standard Mask R-CNN is replaced with the proposed DHConv module to it adapt to shape and size differences encountered in cell detection and segmentation tasks. Finally, a series of comparison and ablation experiments are conducted on various biological cell datasets (such as U373, GoTW1, SIM+, and T24) to verify the effectiveness of the proposed method. The results show that the proposed method can obtain better performance than some state-of-the-art methods in multiple metrics (including AP, Precision, Recall, Dice, and PQ) while maintaining competitive FLOPs and FPS.

Efficacy of deep brain stimulation for Tourette syndrome and its comorbidities: A meta-analysis.

Tourette Syndrome (TS) is a neurodevelopmental disorder characterized by multiple motor and vocal tics, often accompanied by comorbid disorders. Optional treatments for patients with TS include behavioral therapy, pharmacotherapy, and neurostimulation techniques. Deep brain stimulation (DBS) has been considered a therapeutic approach for refractory TS and its comorbid symptoms. However, systematic comparison is necessary to understand the therapeutic effect of DBS among patients with TS with various comorbid symptoms, demographic characteristics, or stimulation targets. Consequently, our research aimed to assess the clinical efficacy of DBS in alleviating the symptoms of TS and its comorbidities. A systematic literature search was conducted across five databases: PubMed, Web of Science, MEDLINE, Embase, and PsycINFO. The primary outcome was the mean change in the global score of the Yale Global Tic Severity Scale (YGTSS), which assesses the severity of tics. The secondary outcomes included mean improvement of comorbid symptoms, such as obsessive-compulsive behaviors (OCB), depression symptoms and anxiety symptoms. In total, 51 studies with 673 participants were included in this meta-analysis. Overall, the DBS led to a significant improvement in tic symptoms (p â€‹< â€‹0.001), as well as the comorbid obsessive-compulsive, depression, and anxiety symptoms with effect sizes of 1.88, 0.88, 1.04, and 0.76 accordingly. In the subgroup analysis, we found that striatum stimulation led to a more significant improvement in OCB in patients with TS compared to that observed with thalamic stimulation (p â€‹= â€‹0.017). The relationship between sex, age, and target with the improvement of tics, depression, and anxiety was not statistically significant (p â€‹= â€‹0.923, 0.438, 0.591 for different male proportions; p â€‹= â€‹0.463, 0.425, 0.105 for different age groups; p â€‹= â€‹0.619, 0.113, 0.053 for different targets). In conclusion, DBS is an efficient treatment option for TS, as well as the comorbid OCB, depression symptoms, and anxiety symptoms. It is important to highlight that stimulating the striatum is more effective in managing obsessive-compulsive symptoms compared to stimulating the thalamus.

Clinical characteristics and surgical outcomes of low-grade epilepsy-associated brain tumors.

Background: Low-grade epilepsy-associated brain tumors (LEATs) are found to be the second most common lesion-related epilepsy. Malignant potential of LEATs is very low and the overall survival is good, so the focus of treatment is focused more on seizure outcome rather than oncological prognosis. Objectives: This study was conducted to evaluate the risk factors of seizure outcomes after resection in patients with LEATs. Design: A retrospective study. Methods: A retrospective analysis of patients with LEATs who underwent resective surgery in our three epilepsy centers between October 2010 and April 2023 with a minimum follow-up of 1 year. Demography, clinical characters, neurophysiology, and molecular neuropathology were assessed for association with postoperative seizure outcomes at 1-, 2-, and 5-year follow-up. Synthetic minority oversampling technique (SMOTE) algorithm model was performed to handle the imbalance of data distribution. Gaussian Naïve Bayes (GNB) algorithms were created as a basis for classifying outcomes according to observation indicators. Results: A total of 111 patients were enrolled in the cohort. The most common pathology was ganglioglioma (n = 37, 33.3%). The percentage of patients with seizure freedom was 91.0% (101/111) at 1-year follow-up, 87.5% (77/88) at 2-year follow-up, and 79.1% (53/67) at 5-year follow-up. Partial resection had a significantly poor seizure outcome compared to total resection and supratotal resection (p < 0.05). The epileptiform discharge on post-resective intraoperative electrocorticography (ECoG) or postoperative scalp electroencephalography (EEG) were negative factors on postoperative seizure freedom at 1-, 2-, or 5-year follow-ups (p < 0.05). The area under the receiver-operating characteristic curve value of the GNB-SMOTE model was 0.95 (95% CI, 0.876-1.000), 0.892 (95% CI, 0.656-0.934), and 0.786 (95% CI, 0.491-0.937) at 1-, 2-, and 5-year follow-up, respectively. Conclusion: The partial resection, post-resective intraoperative ECoG, and postoperative scalp EEG were valuable indicators of poor seizure outcomes. The utilization of post-resective intraoperative ECoG is beneficial to improve seizure outcomes. Based on the data diversity and completeness of three medical centers, a multivariate correlation analysis model was established based on GNB algorithm.

A dataset of the quality of soybean harvested by mechanization for deep-learning-based monitoring and analysis.

Deep learning and machine vision technology are widely applied to detect the quality of mechanized soybean harvesting. A clean dataset is the foundation for constructing an online detection learning model for the quality of mechanized harvested soybeans. In pursuit of this objective, we established an image dataset for mechanized harvesting of soybeans. The photos were taken on October 9, 2018, at a soybean experimental field of Liangfeng Grain and Cotton Planting Professional Cooperative in Guanyi District, Liangshan, Shandong, China. The dataset contains 40 soybean images of different qualities. By scaling, rotating, flipping, filtering, and adding noise to enhance the data, we expanded the dataset to 800 frames. The dataset consists of three folders, which store images, label maps, and record files for partitioning the dataset into training, validation, and testing sets. In the initial stages, the author devised an online detection model for soybean crushing rate and impurity rate based on machine vision, and research outcomes affirm the efficacy of this dataset. The dataset can help researchers construct a quality prediction model for mechanized harvested soybeans using deep learning techniques.

Precise Phase Measurement for Fringe Reflection Technique through Optimized Camera Response.

The Fringe Reflection is a robust and non-contact technique for optical measurement and specular surface characterization. The periodic alternation between dark and light cycles of the fringe patterns encodes the geometric information and provides a non-contact method of spatial measurement through phase extraction. Precisely expressing the positions of the points of the fringe pattern is a fundamental requirement for an accurate fringe reflection measurement. However, the nonlinear processes, both in generating the fringe pattern on a screen and capturing it using pixel values, cause inevitable errors in the phase measurement and eventually reduce the system's precision. Aiming at reducing these nonlinear errors, we focus on constructing a new quantity from the pixel values of the photos of the fringe patterns that could linearly respond to the ideal fringe pattern. To this end, we hypothesize that the process of displaying the fringe pattern on a screen using a control function is similar to the process of capturing the pattern and converting the illuminating information into pixel values, which can be described using the camera's response function. This similarity allows us to build a scaled energy quantity that could have a better linear relation with the control function. We optimize the extracted camera response function using an objective to increase the precision and reduce the quoted error. Experiments designed to determine the positions of points along the quartile lines verify the effectiveness of the proposed method in improving fringe reflection measurement precision.

Multi-channel electrohysterography enabled uterine contraction characterization and its effect in delivery assessment.

Uterine contractions are routinely monitored by tocodynamometer (TOCO) at late stage of pregnancy to predict the onset of labor. However, TOCO reveals no information on the synchrony and coherence of contractions, which are important contributors to a successful delivery. The electrohysterography (EHG) is a recording of the electrical activities that trigger the local muscles to contract. The spatial-temporal information embedded in multiple channel EHG signals make them ideal for characterizing the synchrony and coherence of uterine contraction. To proceed, contractile time-windows are identified from TOCO signals and are then used to segment out the simultaneously recorded EHG signals of different channels. We construct sample entropy SamEn and Concordance Correlation based feature ψ from these EHG segments to quantify the synchrony and coherence of contraction. To test the effectiveness of the proposed method, 122 EHG recordings in the Icelandic EHG database were divided into two groups according to the time difference between the gestational ages at recording and at delivery (TTD). Both SamEn and ψ show clear difference in the two groups (p<10-5) even when measurements were made 120 h before delivery. Receiver operating characteristic curve analysis of these two features gave AUC values of 0.834 and 0.726 for discriminating imminent labor defined with TTD ≤ 24 h. The SamEn was significantly smaller in women (0.1433) of imminent labor group than in women (0.3774) of the pregnancy group. Using an optimal cutoff value of SamEn to identify imminent labor gives sensitivity, specificity, and accuracy as high as 0.909, 0.712 and 0.743, respectively. These results demonstrate superiority in comparing to the existing SOTA methods. This study is the first research work focusing on characterizing the synchrony property of contractions from the electrohysterography signals. Despite the very limited dataset used in the validation process, the promising results open a new direction to the use of electrohysterography in obstetrics.

The natural course of bacillus Calmette-Guérin induced bladder lesions: A long-term follow-up study and systematic review.

Objective: Bacillus Calmette-Guérin (BCG) instillation is the standard adjuvant treatment for intermediate- and high-risk non-muscle-invasive bladder cancer after transurethral resection. Nevertheless, its toxicity often causes bladder complications. On follow-up cystoscopy, post-BCG bladder lesions can be pathologically benign, urothelial carcinoma recurrence, or other types of bladder malignancy. Only a small number of case reports have been published on post-BCG bladder lesions. Their clinical features, natural course, and management remain unknown. Methods: We retrospectively studied cystoscopic videos and medical records of BCG-treated bladder cancer patients at our center. During a long-term follow-up, we took biopsies on tumor-like lesions and described their changes. In addition, we summarized previous studies on post-BCG bladder lesions by systematic literature searching and review. Results: We described a series of three cases with post-BCG bladder lesions mimicking tumor recurrence from a total of 38 cases with follow-up data for more than 5 years. Those lesions could last, grow, or disappear spontaneously, and remain pathological benign for years. In systematic review, we identified and analyzed a total of 15 cases with post-BCG bladder lesions with detailed clinical information. Eleven of the 15 were benign and have a good prognosis with nephrogenic adenoma being the most common pathological type. Conclusion: Based on previous studies and our experience, benign lesions after BCG instillation cannot distinguish with cancer recurrence by cystoscopy alone, even under narrow band imaging mode. Nonetheless, given most of them have a good prognosis, random biopsy or transurethral resection might be spared in the patients with long-term negative biopsy and urine cytology.

Survey Results of Honey Bee Colony Losses in Winter in China (2009-2021).

There is growing concern that massive loss of honey bees can cause serious negative effects on biodiversity and ecosystems. Surveys of colony losses have been performed worldwide to monitor the dynamic changes and health status of honey bee colonies. Here, we present the results of surveys regarding winter colony losses from 21 provinces in China from 2009 to 2021, with a total of 1,744,324 colonies managed by 13,704 beekeepers. The total colony losses were low (9.84%; 95% Confidence Interval (CI): 9.60-10.08%) but varied among years, provinces, and scales of apiaries. As little is known about the overwintering mortality of Apis cerana, in this study, we surveyed and compared the loss rates between Apis mellifera and A. cerana in China. We found colonies of A. mellifera suffered significantly lower losses than A. cerana in China. Larger apiaries resulted in higher losses in A. mellifera, whereas the opposite was observed in A. cerana. Furthermore, we used generalized linear mixed-effects models (GLMMs) to evaluate the effects of potential risk factors on winter colony losses and found that the operation size, species, migration, migration×species interaction, and queen problems were significantly related to the loss rates. New queens can increase their colony overwintering survival. Migratory beekeepers and large operations reported lower loss rates.

Spermatophore development in drones indicates the metabolite support for sperm storage in honey bees (Apis cerana).

Developing effective long-term sperm storage strategies to maintain activity requires an understanding of the underlying spermatophore developmental phase in drones. Here we compared the developmental processes and metabolites about seminal vesicles of drones from different parentages (0-24 d)in honeybee colonies, including mated queens, virgin queens, and worker bees. The results showed a similar developmental trend of seminal vesicles in thethree groups of drones on the whole, although there were significant differences in developmental levels, as well as in other indicators. Correlation analysis showed significant positive correlations between seminal vesicle width and sperm viability. The metabolomics of the seminal vesicles in drones from mated queens showed differences of the metabolites in each stage. Particularly, squalene identified among them was validated a protective effect on sperm vitality in vitro experiments. Together the results of these assays support that there were significant differences in the developmental levels of seminal vesicles among the three groups of drones in honeybees, wherein a significant correlation between sperm viability and the developmental levels of seminal vesicles were dissected. The metabolomics analysis and semen storage experiments in vitro display signatures of squalene that may act as an effective protective agent in maintaining sperm viability. Collectively, our findings indicate that spermatophore development in drones provides metabolite support, which contributes to research on the differences of sperm viability among drones in the future.

Factors that Influence Subdural Hemorrhage Secondary to Intracranial Arachnoid Cysts in Children.

OBJECTIVE: This study aimed to investigate factors that influence subdural haemorrhage (SDH) secondary to intracranial arachnoid cysts (IACs) in children. METHODS: Data of children with unruptured IACs (IAC group) and those with SDH secondary to IACs (IAC-SDH group) were analyzed. Nine factors, sex, age, birth type (vaginal or caesarean), symptoms, side (left, right, or midline), location (temporal or nontemporal), image type (I, II, or III), volume, and maximal diameter, were selected. IACs were classified as types I, II, and III according to their morphological changes observed on computed tomography images. RESULTS: There were 117 boys (74.5%) and 40 girls (25.5%); 144 (91.7%) patients comprised the IAC group and 13 (8.3%) comprised the IAC-SDH group. There were 85 (53.8%) IACs on the left side, 53 (33.5%) on the right side, 20 (12.7%) in the midline region, and 91 (58.0%) in the temporal region. The univariate analysis showed significant differences in age, birth type, symptoms, cyst location, cyst volume, and cyst maximal diameter (P < 0.05) between the 2 groups. Logistic regression using the synthetic minority oversampling technique model showed that image type III and birth type were independent factors that influenced SDH secondary to IACs (ß0 = 4.143; ß for image type = -3.979; ß for birth type = -2.542) and that the representative area under the receiver-operating characteristic curve value was 0.948 (95% confidence interval, 0.898-0.997). CONCLUSIONS: IACs are more common in boys than in girls. They can be divided into 3 groups according to their morphological changes on computed tomography images. Image type III and caesarean delivery were independent factors that influenced SDH secondary to IACs.

Comparison of the mitochondrial genomes of three geographical strains of Apis laboriosa indicates high genetic diversity in the black giant honeybee (Hymenoptera: Apidae).

Apis laboriosa is the largest honeybee that lives mainly on cliff faces, with strong migratory ability. In this study, we firstly sequenced and assembled two complete mitochondrial genomes of A. laboriosa isolated from two distant locations in China (Chongqing and Shangri-La regions). Combined with the published mitochondrial genome of A. laboriosa from Nepal, comparative genomic analyses were conducted to gain insight into the genetic diversity of giant honeybees from different geographical distributions. The mitochondrial genomes of A. laboriosa from Chongqing and Shangri-La regions were 15,579 and 15,683 bp in length, respectively, both larger than that from Nepal with the length of 15,510 bp. Three mitochondrial genomes all harbor 37 common genes and present the same AT bias and the frequency of codon usage. However, the fragments including COX1, SSUrRNA, LSUrRNA, and the AT-rich region of the mitochondrial genome from Shangri-La region demonstrate distinctive insertions and deletions compared to those from Chongqing and Nepal regions. Phylogenetic trees of mitochondrial genomes show that A. laboriosa from Chongqing is most closely related to that from Nepal, rather than to Shangri-La. Genetic distance between Shangri-La and Chongqing or Nepal was even larger than that between the various subspecies of Apis mellifera. Overall, these results unmark that A. laboriosa in different geographical distributions can exhibit high genetic diversity at the mitochondrial genomic level, and therein, A. laboriosa from Shangri-La may be the subspecies. All these studies will contribute to our understanding of the geographical distribution and genetic differentiation of black giant honeybee in Asian region.

CASF-Net: Cross-attention and cross-scale fusion network for medical image segmentation.

BACKGROUND: Automatic segmentation of medical images has progressed greatly owing to the development of convolutional neural networks (CNNs). However, there are two uncertainties with current approaches based on convolutional operations: (1) how to eliminate the general limitations that CNNs lack the ability of modeling long-range dependencies and global contextual interactions, and (2) how to efficiently discover and integrate global and local features that are implied in the image. Notably, these two problems are interconnected, yet previous approaches mainly focus on the first problem and ignore the importance of information integration. METHODS: In this paper, we propose a novel cross-attention and cross-scale fusion network (CASF-Net), which aims to explicitly tap the potential of dual-branch networks and fully integrate the coarse and fine-grained feature representations. Specifically, the well-designed dual-branch encoder hammers at modeling non-local dependencies and multi-scale contexts, significantly improving the quality of semantic segmentation. Moreover, the proposed cross-attention and cross-scale module efficiently perform multi-scale information fusion, being capable of further exploring the long-range contextual information. RESULTS: Extensive experiments conducted on three different types of medical image segmentation tasks demonstrate the state-of-the-art performance of our proposed method both visually and numerically. CONCLUSIONS: This paper assembles the feature representation capabilities of CNN and transformer and proposes cross-attention and cross-scale fusion algorithms. The promising results show new possibilities of using cross-fusion mechanisms in more downstream medical image tasks.

Adaptive Diffusion Pairwise Fused Lasso LMS Algorithm Over Networks.

The topic of identification for sparse vector in a distributed way has triggered great interest in the area of adaptive filtering. Grouping components in the sparse vector has been validated to be an efficient way for enhancing identification performance for sparse parameter. The technique of pairwise fused lasso, which can promote similarity between each possible pair of nonnegligible components in the sparse vector, does not require that the nonnegligible components have to be distributed in one or multiple clusters. In other words, the nonnegligible components may be randomly scattered in the unknown sparse vector. In this article, based on the technique of pairwise fused lasso, we propose the novel pairwise fused lasso diffusion least mean-square (PFL-DLMS) algorithm, to identify sparse vector. The objective function we construct consists of three terms, i.e., the mean-square error (MSE) term, the regularizing term promoting the sparsity of all components, and the regularizing term promoting the sparsity of difference between each pair of components in the unknown sparse vector. After investigating mean stability condition of mean-square behavior in theoretical analysis, we propose the strategy of variable regularizing coefficients to overcome the difficulty that the optimal regularizing coefficients are usually unknown. Finally, numerical experiments are conducted to verify the effectiveness of the PFL-DLMS algorithm in identifying and tracking sparse parameter vector.

Online Detection System for Wheat Machine Harvesting Impurity Rate Based on DeepLabV3.

Wheat, one of the most important food crops in the world, is usually harvested mechanically by combine harvesters. The impurity rate is one of the most important indicators of the quality of wheat obtained by mechanized harvesting. To realize the online detection of the impurity rate in the mechanized harvesting process of wheat, a vision system based on the DeepLabV3+ model of deep learning for identifying and segmenting wheat grains and impurities was designed in this study. The DeepLabV3+ model construction considered the four backbones of MobileNetV2, Xception-65, ResNet-50, and ResNet-101 for training. The optimal DeepLabV3+ model was determined through the accuracy rate, comprehensive evaluation index, and average intersection ratio. On this basis, an online detection method of measuring the wheat impurity rate in mechanized harvesting based on image information was constructed. The model realized the online detection of the wheat impurity rate. The test results showed that ResNet-50 had the best recognition and segmentation performance; the accuracy rate of grain identification was 86.86%; the comprehensive evaluation index was 83.63%; the intersection ratio was 0.7186; the accuracy rate of impurity identification was 89.91%; the comprehensive evaluation index was 87.18%; the intersection ratio was 0.7717; and the average intersection ratio was 0.7457. In terms of speed, ResNet-50 had a fast segmentation speed of 256 ms per image. Therefore, in this study, ResNet-50 was selected as the backbone network for DeepLabV3+ to carry out the identification and segmentation of mechanically harvested wheat grains and impurity components. Based on the manual inspection results, the maximum absolute error of the device impurity rate detection in the bench test was 0.2%, and the largest relative error was 17.34%; the maximum absolute error of the device impurity rate detection in the field test was 0.06%; and the largest relative error was 13.78%. This study provides a real-time method for impurity rate measurement in wheat mechanized harvesting.

Diagnostic value of high-frequency oscillations for the epileptogenic zone: A systematic review and meta-analysis.

BACKGROUND: Delineation of the epileptogenic zone (EZ) is crucial during resective surgery in patients with epilepsy. In recent years, high-frequency oscillations, including fast ripples and ripples, have been considered promising biomarkers of the EZ. The objective of this study was to perform a systematic review and meta-analysis to evaluate the diagnostic accuracy of fast ripples and ripples in identifying EZs. METHODS: The PubMed, Embase, and Cochrane databases were searched systematically. The review process followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses guidelines. A bivariate mixed-effects regression approach was used to obtain summary estimates of the sensitivity and specificity of fast ripples and ripples. RESULTS: Thirteen studies were included in this review (12 for fast ripples and eight for ripples). Fast ripples had a pooled sensitivity of 0.80 (95% confidence interval [CI], 0.62-0.91), a specificity of 0.72 (95% CI, 0.55-0.85), and a summary area under the curve of 0.82 (95% CI, 0.79-0.86). The threshold effect was detected for ripples, and the summary area under the curve was 0.75 (95% CI, 0.71-0.79), with a sensitivity of 0.38 (95% CI, 0.18-0.64) and specificity of 0.90 (95% CI, 0.70-0.97). CONCLUSIONS: Fast ripples are a biomarker of the EZ with moderate diagnostic accuracy; in contrast, ripples are not as effective.

The strategy of precise targeting and in situ oxygenating for enhanced triple-negative breast cancer chemophototherapy.

The absence of effective therapeutic targets and tumor hypoxia are the main causes of failure in the treatment of triple-negative breast cancer (TNBC). Biomimetic nanotechnology and tumor microenvironment (TME) responsiveness bring hope and opportunity to address this problem. Here, we develop a core membrane nanoplatform (HM/D-I-BL) using hollow mesoporous manganese dioxide (HM) coated with a biomimetic cancer cell membrane for enhanced chemotherapy/phototherapy via the strategy of precise drug delivery and hypoxia amelioration. Cancer cell membrane modification endows HM/D-I-BL with excellent homologous targeting and immune escape performance. Cellular uptake and fluorescence imaging studies confirmed that HM/D-I-BL can be accurately delivered to tumor sites. HM/D-I-BL also features efficient in situ O2 generation in tumors upon laser irradiation, and subsequently enhanced chemotherapy/phototherapy, pointing to its usefulness as a TME-responsive nanozyme to alleviate tumor hypoxia in the presence of H2O2. In addition, HM/D-I-BL showed good fluorescence and magnetic resonance imaging performances, which offers a reliable multimodal image-guided combination tumor therapy for precision theranostics in the future. In general, this intelligent biomimetic nanoplatform with its homotypic tumor targeting, in situ alleviation of tumor hypoxia and synergetic chemophototherapy would open up a new dimension for the precision treatment of TNBC.

Independent temporal lobe epilepsy in patients with tuberous sclerosis complex.

Tuberous sclerosis complex (TSC) is a rare disease that involves multiple organs, including the brain; approximately 80%-90% of TSC patients exhibit TSC-associated epilepsy. Independent temporal lobe epilepsy (TLE), TSC-unrelated epilepsy, is particularly rare in patients with TSC. Here, we describe three patients with TSC with independent TLEs that were confirmed by stereo-electroencephalography (EEG), postoperative pathological findings, and seizure outcome at follow-up. The patients were retrospectively enrolled at two centers; their ictal epileptiform discharge onsets were determined using electrode contacts in the hippocampus during stereo-EEG. The three patients underwent anterior temporal lobectomies and remained seizure-free at 1-5 years after surgery. Postoperative pathological examinations confirmed hippocampal sclerosis in all three patients. Furthermore, postoperative intelligence quotient improvement was evident in one patient, while the quality of life was improved in two patients at 12 months after surgery.

Network Theory Based EHG Signal Analysis and its Application in Preterm Prediction.

OBJECTIVE: Preterm birth is the leading cause of neonatal morbidity and mortality. Early identification of high-risk patients followed by medical interventions is essential to the prevention of preterm birth. Based on the relationship between uterine contraction and the fundamental electrical activities of muscles, we extracted effective features from EHG signals recorded from pregnant women, and use them to train classifiers with the purpose of providing high precision in classifying term and preterm pregnancies. METHODS: To characterize changes from irregularity to coherence of the uterine activity during the whole pregnancy, network representations of the original electrohysterogram (EHG) signals are established by applying the Horizontal Visibility Graph (HVG) algorithm, from which we extract network degree density and distribution, clustering coefficient and assortativity coefficient. Concerns on the interferences of different noise sources embedded in the EHG signal, we apply Short-Time Fourier Transform (STFT) to expand the original signal in the time-frequency domain. This allows a network representation and the extraction of related features on each frequency component. Feature selection algorithms are then used to filter out unrelated frequency components. We further apply the proposed feature extraction method to EHG signals available in the Term-Preterm EHG database (TPEHG), and use them to train classifiers. We adopt the Partition-Synthesis scheme which splits the original imbalanced dataset into two sets, and synthesizes artificial samples separately within each subset to solve the problem of dataset imbalance. RESULTS: The optimally selected network-based features, not only contribute to the identification of the essential frequency components of uterine activities related to preterm birth, but also to improved performance in classifying term/preterm pregnancies, i.e., the SVM (Support Vector Machine) classifier trained with the available samples in the TPEHG gives sensitivity, specificity, overall accuracy, and auc values as high as 0.89, 0.93, 0.91, and 0.97, respectively.