Deep learning models for ischemic stroke lesion segmentation in medical images: A survey.

This paper provides a comprehensive review of deep learning models for ischemic stroke lesion segmentation in medical images. Ischemic stroke is a severe neurological disease and a leading cause of death and disability worldwide. Accurate segmentation of stroke lesions in medical images such as MRI and CT scans is crucial for diagnosis, treatment planning and prognosis. This paper first introduces common imaging modalities used for stroke diagnosis, discussing their capabilities in imaging lesions at different disease stages from the acute to chronic stage. It then reviews three major public benchmark datasets for evaluating stroke segmentation algorithms: ATLAS, ISLES and AISD, highlighting their key characteristics. The paper proceeds to provide an overview of foundational deep learning architectures for medical image segmentation, including CNN-based and transformer-based models. It summarizes recent innovations in adapting these architectures to the task of stroke lesion segmentation across the three datasets, analyzing their motivations, modifications and results. A survey of loss functions and data augmentations employed for this task is also included. The paper discusses various aspects related to stroke segmentation tasks, including prior knowledge, small lesions, and multimodal fusion, and then concludes by outlining promising future research directions. Overall, this comprehensive review covers critical technical developments in the field to support continued progress in automated stroke lesion segmentation.

Classification of recurrent major depressive disorder using a residual denoising autoencoder framework: Insights from large-scale multisite fMRI data.

BACKGROUND AND OBJECTIVE: Recurrent major depressive disorder (rMDD) has a high recurrence rate, and symptoms often worsen with each episode. Classifying rMDD using functional magnetic resonance imaging (fMRI) can enhance understanding of brain activity and aid diagnosis and treatment of this disorder. METHODS: We developed a Residual Denoising Autoencoder (Res-DAE) framework for the classification of rMDD. The functional connectivity (FC) was extracted from fMRI data as features. The framework addresses site heterogeneity by employing the Combat method to harmonize feature distribution differences. A feature selection method based on Fisher scores was used to reduce redundant information in the features. A data augmentation strategy using a Synthetic Minority Over-sampling Technique algorithm based on Extended Frobenius Norm measure was incorporated to increase the sample size. Furthermore, a residual module was integrated into the autoencoder network to preserve important features and improve the classification accuracy. RESULTS: We tested our framework on a large-scale, multisite fMRI dataset, which includes 189 rMDD patients and 427 healthy controls. The Res-DAE achieved an average accuracy of 75.1 % (sensitivity = 69 %, specificity = 77.8 %) in cross-validation, thereby outperforming comparison methods. In a larger dataset that also includes first-episode depression (comprising 832 MDD patients and 779 healthy controls), the accuracy reached 70 %. CONCLUSIONS: We proposed a deep learning framework that can effectively classify rMDD and 33 identify the altered FC associated with rMDD. Our study may reveal changes in brain function 34 associated with rMDD and provide assistance for the diagnosis and treatment of rMDD.

Ratiometric fluorescence sensing NADH using AIE-dots transducers at the point of care.

Reduced nicotinamide adenine dinucleotide (NADH) has a strong impact on physiological metabolism, and its concentration is related to metabolic and neurodegenerative diseases. A more reliable and accurate detection method for NADH quantitation is needed for early disease diagnosis and point-of-care testing. Aggregation-induced emission (AIE) materials are widely used to improve the sensitivity in analytes assays due to their anti-aggregation-caused quenching property. Here we developed TPA-BQD-Py AIE-dots transducers and evaluated its performance in NADH detection. The NADH concentration-dependent ratiometric sensing was based on electron transfer from TPA-BQD-Py AIE-dots to NADH with variable fluorescence intensity at 584 nm and 470 nm, resulting in high sensitivity (limit of detection at 110 nM), photostability, selectivity, and a rapid and reversible response. We further developed the application of TPA-BQD-Py AIE-dots transducers in in vivo NADH imaging using a smartphone and digital camera, respectively, demonstrating the potential for NADH point-of-care testing.

Combining methylated SDC2 test in stool DNA, fecal immunochemical test, and tumor markers improves early detection of colorectal neoplasms.

Objective: To explore the value of testing methylated SDC2 (SDC2) in stool DNA combined with fecal immunochemical test (FIT) and serum tumor markers (TM) for the early detection of colorectal neoplasms. Methods: A total of 533 patients, including 150 with CRC (67 with early-stage CRC), 23 with APL, 85 with non-advanced adenomas and general polyps, and 275 with benign lesions and healthy controls. SDC2 was detected by methylation-specific PCR, FIT (hemoglobin, Hb and transferrin, TF) was detected by immunoassay, and the relationships between SDC2, FIT, and clinicopathological features were analyzed. Pathological biopsy or colonoscopy were used as gold standards for diagnosis, and the diagnostic efficacy of SDC2 combined with FIT and TM in CRC and APL evaluated using receiver operating characteristic (ROC) curves. Results: SDC2 positive rates in early-stage CRC and APL were 77.6% (38/49) and 41.2% (7/17), respectively, and combination of SDC2 with FIT increased the positive rates to 98.0% (48/49) and 82.4% (14/17). The positive rates of SDC2 combined with FIT assay in the APL and CRC groups at stages 0-IV were 82.4% (14/17), 85.7% (6/7), 100% (16/16), 100% (26/26), 97.4% (38/39), and 100% (22/22), respectively. Compared to the controls, both the CRC and APL groups showed significantly higher positive detection rates of fecal SDC2 and FIT (χ2 = 114.116, P < 0.0001 and χ2 = 85.409, P < 0.0001, respectively). Our results demonstrate a significant difference in the qualitative methods of SDC2 and FIT for the detection of colorectal neoplasms (McNemar test, P < 0.0001). ROC curve analysis revealed that the sensitivities of SDC2 and FIT, alone or in combination, for the detection of early CRC and APL were 69.9%, 86.3%, and 93.9%, respectively (all P<0.0001). When combined with CEA, the sensitivity increased to 97.3% (P<0.0001). Conclusions: SDC2 facilitates colorectal neoplasms screening, and when combined with FIT, it enhances detection. Furthermore, the combination of SDC2 with FIT and CEA maximizes overall colorectal neoplasm detection.

Classification of recurrent major depressive disorder using a new time series feature extraction method through multisite rs-fMRI data.

BACKGROUND: Major depressive disorder (MDD) has a high rate of recurrence. Identifying patients with recurrent MDD is advantageous in adopting prevention strategies to reduce the disabling effects of depression. METHOD: We propose a novel feature extraction method that includes dynamic temporal information, and inputs the extracted features into a graph convolutional network (GCN) to achieve classification of recurrent MDD. We extract the average time series using an atlas from resting-state functional magnetic resonance imaging (fMRI) data. Pearson correlation was calculated between brain region sequences at each time point, representing the functional connectivity at each time point. The connectivity is used as the adjacency matrix and the brain region sequences as node features for a GCN model to classify recurrent MDD. Gradient-weighted Class Activation Mapping (Grad-CAM) was used to analyze the contribution of different brain regions to the model. Brain regions making greater contribution to classification were considered to be the regions with altered brain function in recurrent MDD. RESULT: We achieved a classification accuracy of 75.8 % for recurrent MDD on the multi-site dataset, the Rest-meta-MDD. The brain regions closely related to recurrent MDD have been identified. LIMITATION: The pre-processing stage may affect the final classification performance and harmonizing site differences may improve the classification performance. CONCLUSION: The experimental results demonstrate that the proposed method can effectively classify recurrent MDD and extract dynamic changes of brain activity patterns in recurrent depression.

Risk Prediction Model for Crohn's Disease Based on Hematological Indicators.

BACKGROUND: The similarity between Crohn's disease (CD) and non-CD, especially with ulcerative colitis (UC) or intestinal tuberculosis (ITB), makes the diagnostic error rate not low. Therefore, there is an urgent need for an efficient, fast, and simple predictive model that can be applied in clinical practice. The purpose of this study is to establish the risk prediction model for CD based on five routine laboratory tests by logistic-regression algorithm, to construct the early warning model for CD and the corresponding visual nomograph, and to provide an accurate and convenient reference for the risk determination and differential diagnosis of CD, in order to assist clinicians to better manage CD and reduce patient suffering. METHODS: Using a retrospective analysis, a total of 310 cases were collected from 2020 to 2022 at The Sixth Affiliated Hospital, Sun Yat-sen University, who were diagnosed by comprehensive clinical diagnosis, including 100 patients with CD, 50 patients with ulcerative colitis (UC), 110 patients with non-inflammatory bowel disease (non-IBD) diseases (65 cases of intestinal tuberculosis, radioactive enterocolitis 39, and colonic diverticulitis 6), and 50 healthy individuals (NC) in the non-CD group. Risk prediction models were established by measuring ESR, Hb, WBC, ALb, and CH levels in hematology. The models were evaluated and visualized using logistic-regression algorithm. RESULTS: 1) ESR, WBC, and WBC/CH ratios in the CD group were higher than those in the non-CD group, while ALb, Hb, CH, WBC/ESR ratio, and Hb/WBC ratio were lower than those in the non-CD group, and the differences were statistically significant (all p < 0.05). 2) CD occurrence had a strong correlation with the WBC/CH ratio, with the correlation coefficient exceeding 0.4; CD occurrence was correlated with other indicators. 3) A risk prediction model containing age, gender, ESR, ALb, Hb, CH, WBC, WBC/CH, WBC/ESR, and Hb/WBC characteristics was constructed using a logistic-regression algorithm. The sensitivity, specificity, positive predictive value, negative predictive value, and area under the curve of the model were 83.0%, 76.2%, 59.0%, 90.5%, and 0.86, respectively. The model based on the corresponding index also had high diagnostic accuracy (AUC = 0.88) for differentiating CD from ITB. Visual nomograph based on the logistic-regression algorithm was also constructed for clinical application reference. CONCLUSIONS: In this study, a CD risk prediction model was established and visualized by five conventional hema-tological indices: ESR, Hb, WBC, ALb, and CH, in addition to a high diagnostic accuracy for the differential diagnosis of CD and ITB.

Semiconducting Polymer Dots for Point-of-Care Biosensing and In Vivo Bioimaging: A Concise Review.

In recent years, semiconducting polymer dots (Pdots) have attracted much attention due to their excellent photophysical properties and applicability, such as large absorption cross section, high brightness, tunable fluorescence emission, excellent photostability, good biocompatibility, facile modification and regulation. Therefore, Pdots have been widely used in various types of sensing and imaging in biological medicine. More importantly, the recent development of Pdots for point-of-care biosensing and in vivo imaging has emerged as a promising class of optical diagnostic technologies for clinical applications. In this review, we briefly outline strategies for the preparation and modification of Pdots and summarize the recent progress in the development of Pdots-based optical probes for analytical detection and biomedical imaging. Finally, challenges and future developments of Pdots for biomedical applications are given.

Altered Effective Connectivity Among the Cerebellum and Cerebrum in Patients with Major Depressive Disorder Using Multisite Resting-State fMRI.

Major depressive disorder (MDD) is a serious and widespread psychiatric disorder. Previous studies mainly focused on cerebrum functional connectivity, and the sample size was relatively small. However, functional connectivity is undirected. And, there is increasing evidence that the cerebellum is also involved in emotion and cognitive processing and makes outstanding contributions to the symptomology and pathology of depression. Therefore, we used a large sample size of resting-state functional magnetic resonance imaging (rs-fMRI) data to investigate the altered effective connectivity (EC) among the cerebellum and other cerebral cortex in patients with MDD. Here, from the perspective of data-driven analysis, we used two different atlases to divide the whole brain into different regions and analyzed the alterations of EC and EC networks in the MDD group compared with healthy controls group (HCs). The results showed that compared with HCs, there were significantly altered EC in the cerebellum-neocortex and cerebellum-basal ganglia circuits in MDD patients, which implied that the cerebellum may be a potential biomarker of depressive disorders. And, the alterations of EC brain networks in MDD patients may provide new insights into the pathophysiological mechanisms of depression.

Nucleic acid amplification strategies for volume-amplified magnetic nanoparticle detection assay.

Magnetic nanoparticles (MNPs) can be quantified based on their magnetic relaxation properties by volumetric magnetic biosensing strategies, for example, alternating current susceptometry. Volume-amplified magnetic nanoparticle detection assays (VAMNDAs) employ analyte-initiated nucleic acid amplification (NAA) reactions to increase the hydrodynamic size of MNP labels for magnetic sensing, achieving attomolar to picomolar detection limits. VAMNDAs offer rapid and user-friendly analysis of nucleic acid targets but present inherence defects determined by the chosen amplification reactions and sensing principles. In this mini-review, we summarize more than 30 VAMNDA publications and classify their detection models for NAA-induced MNP size increases, highlighting the performances of different linear, cascade, and exponential NAA strategies. For some NAA strategies that have not yet been reported in VAMNDA, we predicted their performances based on the reaction kinetics and feasible detection models. Finally, challenges and perspectives are given, which may hopefully inspire and guide future VAMNDA studies.

The alterations of brain functional connectivity networks in major depressive disorder detected by machine learning through multisite rs-fMRI data.

BACKGROUND: The current diagnosis of major depressive disorder (MDD) is mainly based on the patient's self-report and clinical symptoms. Machine learning methods are used to identify MDD using resting-state functional magnetic resonance imaging (rs-fMRI) data. However, due to large site differences in multisite rs-fMRI data and the difficulty of sample collection, most of the current machine learning studies use small sample sizes of rs-fMRI datasets to detect the alterations of functional connectivity (FC) or network attribute (NA), which may affect the reliability of the experimental results. METHODS: Multisite rs-fMRI data were used to increase the size of the sample, and then we extracted the functional connectivity (FC) and network attribute (NA) features from 1611 rs-fMRI data (832 patients with MDD (MDDs) and 779 healthy controls (HCs)). ComBat algorithm was used to harmonize the data variances caused by the multisite effect, and multivariate linear regression was used to remove age and sex covariates. Two-sample t-test and wrapper-based feature selection methods (support vector machine recursive feature elimination with cross-validation (SVM-RFECV) and LightGBM's "feature_importances_" function) were used to select important features. The Shapley additive explanations (SHAP) method was used to assign the contribution of features to the best classification effect model. RESULTS: The best result was obtained from the LinearSVM model trained with the 136 important features selected by SVMRFE-CV. In the nested five-fold cross-validation (consisting of an outer and an inner loop of five-fold cross-validation) of 1611 data, the model achieved the accuracy, sensitivity, and specificity of 68.90 %, 71.75 %, and 65.84 %, respectively. The 136 important features were tested in a small dataset and obtained excellent classification results after balancing the ratio between patients with depression and HCs. CONCLUSIONS: The combined use of FC and NA features is effective for classifying MDDs and HCs. The important FC and NA features extracted from the large sample dataset have some generalization performance and may be used as a reference for the altered brain functional connectivity networks in MDD.

Removing Adversarial Noise via Low-Rank Completion of High-Sensitivity Points.

Deep neural networks are fragile under adversarial attacks. In this work, we propose to develop a new defense method based on image restoration to remove adversarial attack noise. Using the gradient information back-propagated over the network to the input image, we identify high-sensitivity keypoints which have significant contributions to the image classification performance. We then partition the image pixels into the two groups: high-sensitivity and low-sensitivity points. For low-sensitivity pixels, we use a total variation (TV) norm-based image smoothing method to remove adversarial attack noise. For those high-sensitivity keypoints, we develop a structure-preserving low-rank image completion method. Based on matrix analysis and optimization, we derive an iterative solution for this optimization problem. Our extensive experimental results on the CIFAR-10, SVHN, and Tiny-ImageNet datasets have demonstrated that our method significantly outperforms other defense methods which are based on image de-noising or restoration, especially under powerful adversarial attacks.

An integrated microfluidic 3D tumor system for parallel and high-throughput chemotherapy evaluation.

The development of a microplatform with multifunctional integration allowing the dynamic and high-throughput exploration of three-dimensional (3D) cultures is promising for biomedical research. Here, we introduce an integrated microfluidic 3D tumor system with pneumatic manipulation and chemical gradient generation to investigate anticancer therapy in a parallel, controllable, dynamic, and high-throughput manner. The stability of the microfluidic system to realize precise and long-term chemical gradient production was developed. Serial manipulations including active cell trapping, array-like tumor self-assembly and formation, reliable gradient generation, parallel multi-concentration drug stimulation, and real-time tumor analysis were achieved in a single microfluidic device. The microfluidic platform was demonstrated to be stable for high-throughput cell trapping and 3D tumor formation with uniform quantities. On-chip analysis of phenotypic tumor responses to diverse chemotherapies with different concentrations can be conducted in this device. The microfluidic advancement holds great potential for applications in the development of high-performance and multi-functional biomimetic tumor systems and in the fields of cancer research and pharmaceutical development.

Acoustic-resolution-based photoacoustic microscopy with non-coaxial arrangements and a multiple vertical scan for high lateral resolution in-depth.

In conventional acoustic-resolution-based photoacoustic microscopy (ARPAM), a focused ultrasound transducer is placed coaxially with the laser beam to obtain the generated ultrasound signals. The information from deep regions can be greatly affected by the shallow targets. More importantly, in ARPAM the irreconcilable conflict between the lateral resolution and depth of fields has always been a major factor that lowers the imaging quality. In this work, an ARPAM system was developed, in which a non-coaxial arrangement of light illumination and acoustic detection was adopted to alleviate the influence of the tissue surface on the deep targets, and a focal zone integral algorithm was applied with a multiple scanning scheme to improve the lateral resolution. The system can achieve a consistent high lateral resolution of 0.5 mm over a large range in the axial direction. Both the phantom experiment and the chicken embryo in vivo results indicate that the proposed method can provide more in-depth information compared with the conventional ARPAM method. With the development of high repetition lasers and the advancement of image scanning technologies, the proposed method may play an important role in cerebral vascular imaging, superficial tumor imaging, and other related biomedical imaging applications.

Design and Simulation of a Passive Micromixer with Gourd-Shaped Channel.

A gourd-shaped contraction-expansion design is proposed for a passive planar micromixer in this study. The mixing performance of the micromixer is analyzed numerically and compared with a T-shaped planar micromixer. The gourd-shaped contraction-expansion structure can enhance the vortex-formation and mixing abilities of the micromixer. The numerical simulation reveals that the gourd-shaped structure can improved vortex generation and mixing efficiency within a high Reynolds number range. The micromixer with an optimized waist width of 50 µm reaches a mixing efficiency of approximately 83.25% and maintains a moderate pressure drop of 4860 Pa at Re = 100. This study can shed light on the design of new 2D micromixers from the point view of bionics.

Acoustic field simulation method for arbitrarily shaped transducer with dynamically refined sub-elements.

BACKGROUND: Simulation of the emitted acoustic field is crucial to the design of ultrasound transducers. The method based on the spatial impulse response (SIR) and aperture discretization provides a powerful tool to study the acoustic field emitted by a transducer with complex aperture geometry and sophisticated apodization/excitation pattern. METHODS: In this work, a new method based on the dynamically refined sub-elements (SE) is employed to discrete the aperture and generate the SIR. Then, these SIRs are convoluted with the excitation pulse to get the acoustic pressure (AP) signal. When calculating the SIR with this method, the slowly changed time flight from a SE to a field point (FP) is approximated with a step function, and the fast changed length of intersection between a SE and a spherical wave centered at a FP is accurately estimated with the areas of the sub-parts (SP) which are given by the dynamically refined SE. RESULTS: Simulations of the acoustic field created by a focusing transducer array and a hollow structured point focusing transducer indicate that the proposed new method can give similar data accuracy with a sampling frequency 16 times lower than the conventional time tracing SE (TTSE) based method. The computational cost is also reduced by nearly one order of magnitude. CONCLUSIONS: A new method is proposed to simulate the acoustic field emitted by transducers with complex geometrical structure and sophisticated apodization/excitation patterns. The required sampling frequency with the new algorithm is greatly reduced compared to that of the conventional TTSE-based method; thus, the efficiency of the acoustic field calculation is improved significantly.

[Study of hybrid time-frequency method and characterization of atrial fibrillation from surface ECG].

With the non-invasive analysis of time-frequency features and instantaneous frequency (IFs) of atrial fibrillation (AF) from surface ECG signals, some important information reflecting the dynamic behavior of atria with AF can be extracted. In this paper is proposed a hybrid time-frequency analysis method, which uses the respective advantages of Gabor expansion and quadratic Wigner distribution. Our study showed that the time-frequency representation of atrial fibrillation signals was formulated into the combinations of time-frequency atoms series. By controlling the trade-off of resolution and interference terms via Manhattan distance threshold, this method in combination with moment-based computation obtained more robust estimation of IFs. The comparative analysis of 10 pairs of non-terminating and terminating types of AF signals suggested that hybrid estimation of IFs can detect the reduction of a majority of the fibrillatory rate when AF will end. Meanwhile, this method decreases compute burden and is a more robust way relative to peak-based or spectrogram method. So, the proposed method would have prospective applications in clinical management of atrial fibrillation.

A Novel spectral Analysis Method of Atrial Fibrillation Signal Based on Hilbert-Huang Transform.

Time/frequency analysis has been extensively used in biomedical signal processing. By extracting some essential features from the electro-physiological signals, these methods are able to determine the clinical pathology mechanisms of some diseases. Fourier spectrum analysis provides a common framework for examining the distribution of global energy in the frequency domain. However, this method assumes that the signal should be stationary, which limits its application in non-stationary system. Atrial fibrillation (AF) is a complex nonlinear pathological phenomena, and recently receives a significant amount of research effort. In this work we develop a new signal processing method using Hilbert-huang transform to perform spectral analysis of the atrial fibrillation signals (AFs). This method provides a new analysis tool for AFs. Our experimental results show that it improves the spectral resolution and enables us to understand the episode of AF more precisely.