AIM: To construct an intelligent segmentation scheme for precise localization of central serous chorioretinopathy (CSC) leakage points, thereby enabling ophthalmologists to deliver accurate laser treatment without navigational laser equipment. METHODS: A dataset with dual labels (point-level and pixel-level) was first established based on fundus fluorescein angiography (FFA) images of CSC and subsequently divided into training (102 images), validation (40 images), and test (40 images) datasets. An intelligent segmentation method was then developed, based on the You Only Look Once version 8 Pose Estimation (YOLOv8-Pose) model and segment anything model (SAM), to segment CSC leakage points. Next, the YOLOv8-Pose model was trained for 200 epochs, and the best-performing model was selected to form the optimal combination with SAM. Additionally, the classic five types of U-Net series models [i.e., U-Net, recurrent residual U-Net (R2U-Net), attention U-Net (AttU-Net), recurrent residual attention U-Net (R2AttU-Net), and nested U-Net (UNet++)] were initialized with three random seeds and trained for 200 epochs, resulting in a total of 15 baseline models for comparison. Finally, based on the metrics including Dice similarity coefficient (DICE), intersection over union (IoU), precision, recall, precision-recall (PR) curve, and receiver operating characteristic (ROC) curve, the proposed method was compared with baseline models through quantitative and qualitative experiments for leakage point segmentation, thereby demonstrating its effectiveness. RESULTS: With the increase of training epochs, the mAP50-95, Recall, and precision of the YOLOv8-Pose model showed a significant increase and tended to stabilize, and it achieved a preliminary localization success rate of 90% (i.e., 36 images) for CSC leakage points in 40 test images. Using manually expert-annotated pixel-level labels as the ground truth, the proposed method achieved outcomes with a DICE of 57.13%, an IoU of 45.31%, a precision of 45.91%, a recall of 93.57%, an area under the PR curve (AUC-PR) of 0.78 and an area under the ROC curve (AUC-ROC) of 0.97, which enables more accurate segmentation of CSC leakage points. CONCLUSION: By combining the precise localization capability of the YOLOv8-Pose model with the robust and flexible segmentation ability of SAM, the proposed method not only demonstrates the effectiveness of the YOLOv8-Pose model in detecting keypoint coordinates of CSC leakage points from the perspective of application innovation but also establishes a novel approach for accurate segmentation of CSC leakage points through the “detect-then-segment” strategy, thereby providing a potential auxiliary means for the automatic and precise real-time localization of leakage points during traditional laser photocoagulation for CSC.

Retinal laser surgery serves as a crucial clinical approach for treating central serous chorioretinopathy (CSCR). However, the preoperative process requires ophthalmologist to manually annotate keypoint on both fundus fluorescein angiography (FFA) and color fundus (CF) images for image registration, which is time-consuming and susceptible to subjective variability. Meanwhile, the existing rigid registration methods are relatively cumbersome in detecting keypoints and establishing their corresponding relationships. Consequently, exploring a new rigid registration method for multi-modal retinal images has potential clinical value and research significance. This paper proposes a novel rigid registration method for multi-modal retinal images based on intelligent matching of keypoint pairs. Concretely, a keypoint pair dataset tailored for clinical CSCR multi-modal retinal images was first constructed to support subsequent research. Then, the YOLOv8-pose based framework for intelligent matching of keypoint pairs was successfully introduced, which unifies keypoint localization and correspondence establishment into a single task. Third, a keypoint re-localization technique was developed to enhance the positional accuracy of keypoints, followed by a keypoint pair selection strategy to optimize the choice of keypoints for affine transformation. Finally, detailed quantitative and qualitative experiments were conducted to investigate the effectiveness of the proposed method. The experimental results on the private dataset demonstrate that the relocalized keypoints achieved average reductions of 37.78

Accurate measurement of minimum macular hole (MH) diameter is essential for diagnosis and treatment. The manual measurement approach by ophthalmologists is time-consuming, poorly reproducible, and exhibits high inter-observer variability. Threshold-based methods are sensitive to image quality, and perform inadequately in low-contrast optical coherence tomography (OCT) images. Deep learning can achieve higher measurement accuracy but shows limited generalization capability. We propose an automated measurement method comprising three sub-tasks. Specifically, a tailored MH dataset is first created by cropping publicly available OCT images to minimize interference from non-MH regions. Subsequently, an image processing pipeline, which consists of denoising, binarization, morphological operations, and edge detection, is implemented to extract the contours on both sides of MH. Finally, an automated measurement algorithm is designed to locate the closest points on the bilateral contours of MH and thereby calculate the minimum diameter. Extensive experiments are conducted to validate the effectiveness of this research. More concretely, on the public dataset, the D_a metric obtained by the proposed method closely aligns with the D_m metric. The ADE values are predominantly within 1 pixel, achieving as low as 0.00 in some cases. Simultaneously, the method demonstrates favorable performance in the RDE metric, with values ranging from 0.00

Manual preoperative image registration for central serous chorioretinopathy (CSCR) is labor-intensive and irreproducible. While rigid registration robustly aligns images globally, it misses fine details. Non-rigid registration, though excellent for local refinement, performs poorly with large discrepancies. Therefore, this study presents a coarse-to-fine registration method for multimodal retinal images to address the aforementioned issues. First, a three-step coarse registration strategy is designed that integrates keypoint pair detection and matching via a YOLOv8-pose network, further optimizes keypoints through a post-processing technique, and achieves initial alignment via affine transformation. On this basis, a dual-component fine registration strategy is then implemented, where disentanglement learning eliminates modality-specific variations while preserving essential vessel structures required for registration, and deformable network generates optimized deformation field to refine the coarse alignment locally, ultimately enabling high-precision image registration. Comprehensive qualitative and quantitative experiments were conducted on the CSCR clinical dataset, which includes both color fundus (CF) and fundus fluorescence angiography (FFA) images, to evaluate the proposed method. With Dice and Dices scores of 0.6759 and 0.4977, the method performs comparably to existing approaches, suggesting its potential application value for CSCR preoperative planning.

Sepsis is a common indirect insult leading to acute respiratory distress syndrome (ARDS). Circulating extracellular vesicles (EVs) have been reported to participate in the pathogenesis of sepsis. However, the alteration of EV-bound S100A8/A9 during septic shock, along with the role of S100A8/A9 in driving acute lung injury, remains unexplored. EVs were isolated from the plasma of patients upon admission with sepsis or septic shock, as well as from healthy controls. Levels of EV S100A8/A9 were assayed via ELISA. To examine the effects and underlying mechanisms of septic shock EVs in acute lung injury, these EVs were administered intratracheally into wild-type C57BL/6 mice or mice with a deficiency of advanced glycation end-products (RAGE). In addition, a mouse model of polymicrobial sepsis was introduced using cecal ligation and puncture (CLP). Levels of EV S100A8/A9 were significantly elevated in patients with sepsis or septic shock compared to healthy controls. Receiver operating characteristic (ROC) analysis demonstrated that EV S100A8/A9 effectively distinguished between septic shock and sepsis and had predictive potential for the development of ARDS. Notably, the levels of S100A8/A9 in EVs and alveolar macrophages from CLP mice were significantly higher than those in sham mice. Intratracheal administration of septic shock EVs directly induced acute lung injury and M1 macrophage polarization in a lipopolysaccharide-independent manner. Septic shock EVs were efficiently taken up by alveolar macrophages in vivo, leading to a significant increase in S100A8/A9 levels, which was inhibited by preincubating the EVs with an S100A8/A9 neutralizing antibody. Additionally, mice with deficiency in RAGE, a receptor for S100A8/A9, were partially protected from acute lung injury induced by septic shock EVs. In vitro, septic shock EVs prompted a proinflammatory response in bone marrow-derived macrophages. This response was blocked by preincubating the EVs with the S100A8/A9 neutralizing antibody. Our results suggested that EV S100A8/A9 has potential value in distinguishing septic shock from sepsis and predicting the development of ARDS. Septic shock EVs-induced lung injury is at least partially mediated through S100A8/A9-RAGE pathway, involving the activation of alveolar macrophages.

BackgroundFundus vessel segmentation is crucial for the early diagnosis of ocular diseases. However, existing deep learning-based methods, although effective for detecting coarse vessels, still face challenges in segmenting fine vessels and heavily rely on time-consuming and labor-intensive pixel-level annotations.MethodsTo alleviate these limitations, this study proposes an image-text guided segmentation model enhanced with the Squeeze-and-Excitation (SE) module and gated residual learning. Concretely, the multimodal fundus vessel datasets with text labels are primarily constructed, effectively supporting our pioneering effort to successfully introduce an image-text model into fundus vessel segmentation. Secondly, an improved image-text model is meticulously designed, focusing on the following two aspects: (1) embedding the SE module in the CNN backbone to adaptively recalibrate channel weights for enhanced vessel feature representation; (2) integrating gated residual learning into the ViT backbone to dynamically regulate the information flow between image and text features.ResultsExtensive quantitative and qualitative experiments on two publicly available datasets, including DRIVE and ROSE-1, demonstrate that the proposed model achieves superior segmentation performance. Specifically, on the DRIVE dataset, the model attains an F1-score of 82.01%, an accuracy of 95.72%, a sensitivity of 83.25%, and a specificity of 97.43%. On the ROSE-1 dataset, the model records an F1-score of 86.34%, an accuracy of 94.61%, a sensitivity of 90.14%, and a specificity of 95.80%. Compared with most deep learning methods, these results reveal the competitiveness of the improved model, indicating its feasibility and potential value in fundus vessel segmentation, which is expected to expand a new research approach in this field.

BackgroundFundus vessel segmentation is vital for diagnosing ophthalmic diseases like central serous chorioretinopathy (CSC), diabetic retinopathy, and glaucoma. Accurate segmentation provides crucial vessel morphology details, aiding the early detection and intervention of ophthalmic diseases. However, current algorithms struggle with fine vessel segmentation and maintaining sensitivity in complex regions. Challenges also stem from imaging variability and poor generalization across multimodal datasets, highlighting the need for more advanced algorithms in clinical practice.MethodsThis paper aims to explore a new vessel segmentation method to alleviate the above problems. We propose a fundus vessel segmentation model based on a combination of double skip connections, deep supervision, and TransUNet, namely DS2TUNet. Initially, the original fundus images are improved through grayscale conversion, normalization, histogram equalization, gamma correction, and other preprocessing techniques. Subsequently, by utilizing the U-Net architecture, the preprocessed fundus images are segmented to obtain the final vessel information. Specifically, the encoder firstly incorporates the ResNetV1 downsampling, dilated convolution downsampling, and Transformer to capture both local and global features, which upgrades its vessel feature extraction ability. Then, the decoder introduces the double skip connections to facilitate upsampling and refine segmentation outcomes. Finally, the deep supervision module introduces multiple upsampling vessel features from the decoder into the loss function, so that the model can learn vessel feature representations more effectively and alleviate gradient vanishing during the training phase.ResultsExtensive experiments on publicly available multimodal fundus datasets such as DRIVE, CHASE_DB1, and ROSE-1 demonstrate that the DS2TUNet model attains F1-scores of 0.8195, 0.8362, and 0.8425, with Accuracy of 0.9664, 0.9741, and 0.9557, Sensitivity of 0.8071, 0.8101, and 0.8586, and Specificity of 0.9823, 0.9869, and 0.9713, respectively. Additionally, the model also exhibits excellent test performance on the clinical fundus dataset CSC, with F1-score of 0.7757, Accuracy of 0.9688, Sensitivity of 0.8141, and Specificity of 0.9801 based on the weight trained on the CHASE_DB1 dataset. These results comprehensively validate that the proposed method obtains good performance in fundus vessel segmentation, thereby aiding clinicians in the further diagnosis and treatment of fundus diseases in terms of effectiveness and feasibility.

Parkinson's disease (PD) is a complex neurodegenerative disease in the elderly with motor and non-motor characteristics. PD is characterized by a unique set of clinical manifestations, including resting tremor, bradykinesia, and postural gait disorders. Patients may also experience non-motor symptoms such as depression, constipation and sleep disorders.[1] If the disease progresses, gradual loss of brain function and premature death may occur. PD is initially associated with the loss of nigrostriatal dopaminergic neurons, with Lewy bodies found in intact nigrostriatal cells. PD affects 1–2% of adults over the age of 65 years and 4% of adults over the age of 80 years. Approximately 60,000 Americans are diagnosed with PD each year and there are currently more than one million Americans suffering from the disease. Given the increase in life expectancy, the amount of people living with PD is expected to increase by over 50% by 2030.[2] In the Braak phase of PD, α-synuclein aggregates pathologically in the olfactory bulb and gut, and then spreads to the central nervous system. In addition, risk factors for PD have been identified, such as exposure to air pollution and environmental pollutants which contains metals, pesticides and the microbiome. A previous animal study has proven that exposure to environmental particles leads to neuropathological damages to dopaminergic neurons, which is a feature of PD.[3] Reducing the burden of PD can be achieved by a two-pronged strategy: implementing measures to decrease changeable factors such as behavioral or environmental risk factors and development of drugs that target the mechanisms of pathological genes or environmental exposure related to PD.[4] Therefore, understanding the relationship between environmental risk factors and PD pathogenesis can promote relevant prevention and treatment strategies. Due to the widespread distribution of natural and man-made sources, nitrogen dioxide (NO2) is a ubiquitous atmospheric pollutant that can cause respiratory irritation when inhaled at high level concentrations. Natural sources of NO2 include volcanism, hill fire, lightning, and the stratosphere. The association between NO2 exposures and sanitation is often attributed to the exposure to ultrafine particles from transport-related emissions. Jo's study[3] found that the increase of the risk of PD was highly associated with exposure to NO2 (Hazard ratio for highest compared with lowest quartile, 1.41; 95% confidence interval, 1.02 to 1.95; P value of tendency = 0.045), so did another study5 that evaluated the risk of PD based on the concentration of NO2. Meanwhile, several studies have demonstrated the direct toxic effects of NO2 on the central nerve system. Studies have indicated that NO2 inhalation exacerbates amyloid β42 (Aβ42) accumulation and causes cognitive impairment through the metabolism of prostaglandin E2,[6] causes synaptic dysfunction accompanied by auopathy,[7] and increases the concentrations of pro-inflammatory markers in the brain[8] (Table 1). Besides, oxidative stress and systemic inflammation following the inhalation of NO2 have been reported.[9] The direct toxic effects of NO2 are consistent with a number of epidemiological studies suggesting associations between neurological disorders and NO2, such as stroke, PD, and amyotrophic lateral sclerosis.Table 1:: The potential role of nitrogen dioxide (NO2) in neurodegenerative diseases (including Parkinson's disease)We reviewed the related articles systematically in the online databases PubMed, Embase and Cochrane on March 3, 2022, using the following search strategies: nitrogen dioxide or nitrogen peroxide, and Parkinson's disease or primary parkinsonism, in various combinations as needed. In this manuscript, we explore the potential role of NO2 inhalation in PD to provide insight into the pathogenic mechanisms involved and to guide possible future treatment and prevention methods. The potential role of NO2 inhalation in mitochondrial dysfunction and excessive reactive oxygen species (ROS) generation: NO2 is reported to be a highly concentrated pollutant that may be involved in the pathogenesis of PD and plays an important role in neuronal cell death.[10] The impairment of mitochondrial has been implicated in a number of neurological disorders, and mitochondrial production of ROS is associated with these physiological signaling cascades. The interaction of mitochondrial membrane damage with compounds disrupts the efficiency of the coupling between oxidation and phosphorylation, resulting in a large bioenergetic deficit that is essential for the survival of cells and organisms. Mitochondria generate a membrane potential in the form of a proton gradient through the inner mitochondrial membrane by using oxidizable substrates.[8] NO2 exposure may induce a reduction in mitochondrial intima potential. The membrane potential provides the impetus for adenosine triphosphate synthesis, and a drop in membrane potential affects adenosine triphosphate production in cells, ultimately leading to cell death.[11] Meanwhile, after NO2 inhalation, thiazolyl blue tetrazolium bromide metabolism, cytochrome C oxidase activity and the expression levels of four respiratory complex subunits are decreased, which means that mitochondrial respiratory function is affected after NO2 inhalation, and mitochondrial respiratory function and neural activity are closely linked.[12] The main source of ROS production is the mitochondrial respiratory enzyme complexes. The activities of these complexes are decreased after NO2 inhalation, which is an indirect indication of oxidative damage and results in loss of dopamine release from striatal axons, reduction of dendritic structures and dopamine release in the substantia nigra. After NO2 inhalation, the increased malonaldehyde level and overproduction of ROS in the cortex has been confirmed in the study by Li et al.[10] These results confirm the close association between mitochondrial dysfunction or excessive ROS and NO2 inhalation, which may induce neurotoxicity and finally result in PD (Figure 1).Figure 1:: Nitrogen dioxide (NO2) inhalation may disrupt the efficiency of the coupling between oxidation and phosphorylation, induce a decrease in mitochondrial intima potential which contributes to the decrease of adenosine triphosphate (ATP) generation.Note: NO2 can decrease the activities of mitochondrial respiratory enzyme complexes that results in excessive reactive oxygen species (ROS) generation. The decrease of ATP generation and excessive ROS generation finally result in neuron death. Created with Pathway Builder 2.0.The potential role of NO2 inhalation in tau pathology and insulin resistance: Hyperphosphorylated tau protein is the main source of pathological tau inclusions that accumulate in neurogenic fibers and eventually lead to several neurodegenerative diseases such as PD.[13] Tau phosphorylation increased in the cerebral cortex and hippocampus in a dose-dependent manner after inhalation of NO2, clearly indicating that the inhalation of NO2 induces taupathy by enhancing tau phosphorylation. There is evidence that accumulation of highly phosphorylated tau in the brain deletions related to tau significantly impairs the function of synapses by altering glutamate receptor expression and function. Glutamate receptor expression was reduced in the cerebral cortex and hippocampus after exposure to NO2. Based on these findings, NO2 exposure shows a great association with synaptic transmission dysfunction. Tau can influence the activity of synapses by directly interacting with post-synaptic signaling complexes in addition to regulating glutamate receptor content.[14] The hyperphosphorylated tau protein, following exposure to NO2, initiates the reduction cascade of the synaptic protein. Insulin is a toxicological target of air pollutants such as NO2 which can impair insulin sensitivity and induce insulin resistance, based on previous epidemiological and experimental studies.[15] Findings focusing on the function of insulin and its related receptor in the central nervous system have shown that the neuronal insulin signaling can directly regulate the tau function and disrupt intracellular insulin signaling molecules, leading to tauopathy.[16] Physiological functions of insulin are performed in various tissues targeted by insulin through several intracellular signaling cascades. The insulin receptor substrate-1/phosphoinositide-3-kinase/protein kinase B (Akt) pathway is the most remarkable cascades. Tyrosine phosphorylation of insulin receptor substrate-1 and Akt was significantly reduced after exposure to NO2, indicating the impairment of insulin signaling function. In response to inhaled NO2, abnormal tau phosphorylation helped increase glycogen synthase kinase 3β by Akt inhibition activity.[7] As a result, NO2 inhalation contributed to the tauopathy and disturbed insulin signaling, which is also involved in mediating the insulin receptor substrate-1/Akt/glycogen synthase kinase 3β signaling pathway (Figure 2).Figure 2:: The potential mechanism of nitrogen dioxide (NO2) causing taupathy.Note: NO2 inhalation decreases tyrosine phosphorylation of insulin receptor substrate-1 (IRS-1) and protein kinase B (AKT) and increase the level of glycogen synthase kinase 3β (GSK-3β) which finally results in taupathy and insulin signaling impairment. Created with Pathway Builder 2.0.The potential role of NO2 inhalation in Αβ42 accumulation aggravated: Aβ42 is one of the isoforms of Aβ that accumulates when the gene encoding amyloid precursor protein is mutated or otherwise causes an abnormal increase in β-secretase activity.[17] The typical pathological feature of PD is the degenerative absence of dopaminergic neurons in the substantia nigra of the midbrain.[18] A study found that Aβ42 could cause progressive degeneration of dopaminergic neurons, significantly altering the morphology of dopaminergic neurons, damaging proteins and membrane structures, and leading to neuronal necrosis or apoptosis.[19] NO2 inhalation was shown to selectively cause Aβ42 deposition and dose-dependent impairment of mouse memory and cognitive ability. These findings suggest that inhaling NO2 may enhance the production of Aβ42 and lead to deterioration of spatial learning and memory. Through the Kyoto Encyclopedia of Genes and Genomes and Biocarta pathways, the significantly differentially expressed genes were mainly involved in the function of synapse and the abilities of learning and memory, including long-term potentiation/depression, the pathway of NO signaling, glutamatergic synapses, neurotrophin signaling, and the pathway of calcium signaling.[6] The arachidonic acid metabolism pathway was also involved in the Kyoto Encyclopedia of Genes and Genomes analysis. Arachidonic acid-derived prostaglandin E2, mainly derived from the cyclooxygenase-2 reaction, plays a critical role in stimulating Aβ formation, promoting neuroinflammatory responses, and regulating synaptic events (Figure 3).[20] Thus, cyclooxygenase-2-mediated arachidonic acid metabolism seems to be a potential mechanism to promote neurodegenerative disease progression. Furthermore, improving the endogenous 2-arachidonoylglycerol levels by suppressing monoacylglycerol lipase effectively inhibits the neuroinflammation and accumulation of Aβ induced by the excessive release of prostaglandin E2,[21] thereby reducing the damage to spatial learning and memory caused by inhalation of NO2, which provides a mechanistic basis for treating PD-related diseases in contaminated areas.Figure 3:: After nitrogen dioxide (NO2) inhalation, arachidonic acid metabolism pathway may be involved in amyloid β42 (Aβ42) accumulation and neuroinflammation by increase the level of cyclooxygenase-2 (COX-2) and prostaglandin E2 (PGE2).Note: COX-2-mediated AA metabolism seems to be a potential mechanism to promote necrosis and Apoptosis of neuron. Created with Pathway Builder 2.0.Limitations: First of all, the real air environment is a complex system that contains various components. Therefore, previous evidence may not be sufficient to confirm the exposure response correlation between NO2 and PD progression because of the different components in air pollutant mixtures. In addition, the studies included in this review investigated the potential mechanisms of NO2 in PD using animal models. Further studies should focus on the relationship between pathophysiological changes of populations exposed to high-level NO2 and the morbidity of PD in the future. Conclusion: With an aging population, PD urgently needs more attention. Many studies have proven that NO2 can increase the incidence of PD. We need to continuously study the role of NO2 in the pathophysiology of PD, so as to provide guidance for the further prevention and treatment of PD. Gang Chen is an Editorial Board member of Medical Gas Research. He is blindedfrom reviewing or making decisions on the manuscript. The article was subject to the journal 's standard procedures, with peer review handled independently of this Editorial Board member and his research group.

Accurately quantifying the height of central serous chorioretinopathy (CSCR) lesion is of great significance for assisting ophthalmologists in diagnosing CSCR and evaluating treatment efficacy. The manual measurement results dominated by single optical coherence tomography (OCT) B-scan image in clinical practice face the dilemma of weak reference, poor reproducibility, and experience dependence. In this context, this paper constructs two schemes: Scheme Ⅰ draws on the idea of ensemble learning, namely, integrating multiple models for locating starting key point in the height direction of lesion in the inference stage, which appropriately improves the performance of a single model. Scheme Ⅱ designs an adaptive gradient threshold (AGT) technique, followed by the construction of cascading strategy, which involves preliminary location of starting key point through deep learning, and then employs AGT for precise adjustment. This strategy not only achieves effective location for starting key point, but also significantly reduces the large appetite of deep learning model for training samples. Subsequently, AGT continues to play a crucial role in locating the terminal key point in the height direction of lesion, further demonstrating its feasibility and effectiveness. Quantitative and qualitative key point location experiments in the height direction of lesion on 1152 samples, as well as the final height measurement display, consistently conveys the superiority of the constructed schemes, especially the cascading strategy, expanding another potential tool for the comprehensive analysis of CSCR.

The lesion boundary of central serous chorioretinopathy (CSCR) is the guarantee to guide the ophthalmologist to accurately arrange the laser spots, so as to enable this ophthalmopathy to be treated precisely. Currently, the accuracy and rapidity of manually locating CSCR lesion boundary in clinic based on single-modal fundus image are limited by imaging quality and ophthalmologist experience, which is also accompanied by poor repeatability, weak reliability and low efficiency. Consequently, a multi-modal fundus image-based lesion boundary auxiliary location method is developed. Firstly, the initial location module (ILM) is employed to achieve the preliminary location of key boundary points of CSCR lesion area on the optical coherence tomography (OCT) B-scan image, then followed by the joint location module (JLM) created based on reinforcement learning for further enhancing the location accuracy. Secondly, the scanning line detection module (SLDM) is constructed to realize the location of lesion scanning line on the scanning laser ophthalmoscope (SLO) image, so as to facilitate the cross-modal mapping of key boundary points. Finally, a simple yet effective lesion boundary location module (LBLM) is designed to assist the automatic cross-modal mapping of key boundary points and enable the final location of lesion boundary. Extensive experiments show that each module can perform well on its corresponding sub task, such as JLM, which makes the correction rate (CR) of ILM increase to 92.11%, comprehensively indicating the effectiveness and feasibility of this method in providing effective lesion boundary guidance for assisting ophthalmologists to precisely arrange the laser spots, and also opening a new research idea for the automatic location of lesion boundary of other fundus diseases.

Accurately and rapidly measuring the diameter of central serous chorioretinopathy (CSCR) lesion area is the key to judge the severity of CSCR and evaluate the efficacy of the corresponding treatments. Currently, the manual measurement scheme based on a single or a small number of optical coherence tomography (OCT) B-scan images encounters the dilemma of incredibility. Although manually measuring the diameters of all OCT B-scan images of a single patient can alleviate the previous issue, the situation of inefficiency will thus arise. Additionally, manual operation is subject to subjective factors of ophthalmologists, resulting in unrepeatable measurement results. Therefore, an automatic image processing method (i.e., a joint framework) based on artificial intelligence (AI) is innovatively proposed for locating the key boundary points of CSCR lesion area to assist the diameter measurement. Firstly, the initial location module (ILM) benefiting from multitask learning is properly adjusted and tentatively achieves the preliminary location of key boundary points. Secondly, the location task is formulated as a Markov decision process, aiming at further improving the location accuracy by utilizing the single agent reinforcement learning module (SARLM). Finally, the joint framework based on the ILM and SARLM is skillfully established, in which ILM provides an initial starting point for SARLM to narrow the active region of agent, and SARLM makes up for the defect of low generalization of ILM by virtue of the independent exploration ability of agent. Experiments reveal the AI-based method which joins the multitask learning, and single agent reinforcement learning paradigms enable agents to work in local region, alleviating the time-consuming problem of SARLM, performing location task in a global scope, and improving the location accuracy of ILM, thus reflecting its effectiveness and clinical application value in the task of rapidly and accurately measuring the diameter of CSCR lesions.

BackgroundThe location of retinal vessels is an important prerequisite for Central Serous Chorioretinopathy (CSC) Laser Surgery, which does not only assist the ophthalmologist in marking the location of the leakage point (LP) on the fundus color image but also avoids the damage of the laser spot to the vessel tissue, as well as the low efficiency of the surgery caused by the absorption of laser energy by retinal vessels. In acquiring an excellent intra- and cross-domain adaptability, the existing deep learning (DL)-based vessel segmentation scheme must be driven by big data, which makes the densely annotated work tedious and costly.MethodsThis paper aims to explore a new vessel segmentation method with a few samples and annotations to alleviate the above problems. Firstly, a key solution is presented to transform the vessel segmentation scene into the few-shot learning task, which lays a foundation for the vessel segmentation task with a few samples and annotations. Then, we improve the existing few-shot learning framework as our baseline model to adapt to the vessel segmentation scenario. Next, the baseline model is upgraded from the following three aspects: (1) A multi-scale class prototype extraction technique is designed to obtain more sufficient vessel features for better utilizing the information from the support images; (2) The multi-scale vessel features of the query images, inferred by the support image class prototype information, are gradually fused to provide more effective guidance for the vessel extraction tasks; and (3) A multi-scale attention module is proposed to promote the consideration of the global information in the upgraded model to assist vessel localization. Concurrently, the integrated framework is further conceived to appropriately alleviate the low performance of a single model in the cross-domain vessel segmentation scene, enabling to boost the domain adaptabilities of both the baseline and the upgraded models.ResultsExtensive experiments showed that the upgraded operation could further improve the performance of vessel segmentation significantly. Compared with the listed methods, both the baseline and the upgraded models achieved competitive results on the three public retinal image datasets (i.e., CHASE_DB, DRIVE, and STARE). In the practical application of private CSC datasets, the integrated scheme partially enhanced the domain adaptabilities of the two proposed models.

The diameter of central serous chorioretinopathy (CSCR) lesion is one of the important indicators to evaluate the severity of CSCR and the efficacy of corresponding treatment schemes. Traditional manual measurement by ophthalmologists is usually based on a single or a small number of optical coherence tomography (OCT) B-scan images. This measurement scheme may not be convincing, vulnerable to subjective factors and lower efficiency. To alleviate the above situation, this paper proposes an intelligent key boundary point location method for all B-scan images of a single patient to assist in the rapid and accurate diameter measurement of the CSCR lesion area. Firstly, an initial location module (ILM) based on the multi-task learning paradigm is appropriately adjusted and introduced into the key boundary point location task, which preliminarily realizes the rapid location of key boundary points. Secondly, to further ameliorate the ILM, a gradient based correction module (GBCM) is designed, followed by the construction of the cascade model (ILM-GBCM) which improves the location accuracy of key boundary points as a whole. Extensive experiments based on five different convolutional neural network (CNN) backbones are carried out, revealing the feasibility of ILM in this task and the effectiveness of ILM-GBCM. On 912 testing images, the maximum correction ratio reaches 83.66%, and the minimum location time at the image level is as low as 0.1754 s, which not only confirms the necessity of correction operation, but also greatly reduce the time cost of ophthalmologists' manual measurement operation in clinic.

At present, laser surgery is one of the effective ways to treat the chronic central serous chorioretinopathy (CSCR), in which the location of the leakage area is of great importance. In order to alleviate the pressure on ophthalmologists to manually label the biomarkers as well as elevate the biomarker segmentation quality, a semiautomatic biomarker segmentation method is proposed in this paper, aiming to facilitate the accurate and rapid acquisition of biomarker location information. Firstly, the multimodal fundus images are introduced into the biomarker segmentation task, which can effectively weaken the interference of highlighted vessels in the angiography images to the location of biomarkers. Secondly, a semiautomatic localization technique is adopted to reduce the search range of biomarkers, thus enabling the improvement of segmentation efficiency. On the basis of the above, the low-rank and sparse decomposition (LRSD) theory is introduced to construct the baseline segmentation scheme for segmentation of the CSCR biomarkers. Moreover, a joint segmentation framework consisting of the above method and region growing (RG) method is further designed to improve the performance of the baseline scheme. On the one hand, the LRSD is applied to offer the initial location information of biomarkers for the RG method, so as to ensure that the RG method can capture effective biomarkers. On the other hand, the biomarkers obtained by RG are fused with those gained by LRSD to make up for the defect of undersegmentation of the baseline scheme. Finally, the quantitative and qualitative ablation experiments have been carried out to demonstrate that the joint segmentation framework performs well than the baseline scheme in most cases, especially in the sensitivity and F1-score indicators, which not only confirms the effectiveness of the framework in the CSCR biomarker segmentation scene but also implies its potential application value in CSCR laser surgery.

The angiography and color fundus images are of great assistance for the localization of central serous chorioretinopathy (CSCR) lesions. However, it brings much inconvenience to ophthalmologists because of these two modalities working independently in guiding laser surgery. Hence, a novel fundus image fusion method in non-subsampled contourlet transform (NSCT) domain, aiming to integrate the multi-modal CSCR information, is proposed. Specifically, the source images are initially decomposed into high-frequency and low-frequency components based on NSCT. Then, an improved deep learning-based method is employed for the fusion of low-frequency components, which helps to alleviate the tedious process of manually designing fusion rules and enhance the smoothness of the fused images. The fusion of high-frequency components based on pulse-coupled neural network (PCNN) is closely followed to facilitate the integration of detailed information. Finally, the fused images can be obtained by applying an inverse transform on the above fusion components. Qualitative and quantitative experiments demonstrate the proposed scheme is superior to the baseline methods of multi-scale transform (MST) in most cases, which not only implies its potential in multi-modal fundus image fusion, but also expands the research direction of MST-based fusion methods.

针对BP网络算法预测光伏组件电压易陷入局部最优解,提出一种新型智能算法--自适应差分进化算法优化BP神经网络(BPNN).太阳能无人机光伏组件电压的预测是通过自适应差分进化算法对BP神经网络的初始值和阈值进行优化,经过不断趋同、异化、迭代,输出最优个体,并按照编码规则将其解码后得到BP神经网络的初始权值和初始阈值,建立SaDE-BPNN电压预测模型.为了证明新方法的优良性能,选取均值绝对误差、均值绝对误差、均方根误差和相对误差4种精度指标对模型的精度进行评价.实验结果表明,SaDE优化BPNN后的平均绝对误差比BPNN低约30％.SaDE优化后的BPNN的均值绝对误差和均方根误差均小于BPNN,分别约为0.65和0.043.以上数据表明,新方法提高了预测的精度,实现了全局优化,能够显著提高预测效果.

Central serous chorioretinopathy (CSCR) is a common fundus disease. Early detection of CSCR is of great importance to prevent visual loss. Therefore, a novel automatic detection method is presented in this paper which integrates technologies including discrete wavelet transform (DWT) image decomposition, local binary patterns (LBP) based texture feature extraction, and multi-instance learning (MIL). LBP is selected due to its robustness to low contrast and low quality images, which can reduce the interference of image itself on the detection method. DWT image decomposition provides high-frequency components with rich details for extracting LBP texture features, which can remove redundant information that is not necessary for diagnosis of CSCR in the raw image. The tedious task of accurately locating and segmenting CSCR lesions is avoided by using MIL. Experiments on 358 optical coherence tomography (OCT) B-scan images demonstrate the effectiveness of our method. Even under the condition of single threshold, the accuracy of 99.58% is obtained at K = 35 by only using a high-frequency feature fusion scheme, which is competitive with the existing methods. Additionally, through further detail innovation, such as multi-threshold optimization (MTO) and integrated decision-making (IDM), the performance of our method is further improved and the detection accuracy is 100% at K = 40.

A gas path fault diagnosis scheme for turborfan engines based on deep belief network (DBN) is presented. The scheme is constructed according to the diagnosis principles of gas path faults and is composed of a turbofan engine reference model and a DBN diagnosis model. The DBN diagnosis model is a stacked network of several restricted Boltzmann machines (RBM) and was trained with the contrastive divergence algorithm and the back propagation algorithm. To optimize the DBN performance, the orthogonal tests L25 (5 7 ) were adopted to determine the hyper-parameters, such as learning rate, hidden layer number, hidden layer neuron number, etc. The proposed DBN-based scheme was applied to diagnose the gas path faults of a turbofan engine model and compared with BP-based and SVM-based schemes. The results show that the fault diagnosis accuracy of the DBN-based scheme is as high as 96.59%, and the DBN-based scheme has dramatic performance advantages over the other two schemes.

利用传统电导增量法跟踪最大功率点时,若跟踪步长较大,则跟踪速度较快,但跟踪精度较差;反之,则跟踪精度较好,但跟踪速度较慢.当外界环境发生变化时,利用传统电导增量法得到的功率变化曲线振荡幅度较大,功率损失较多.改进粒子群算法能够对外界环境的突变迅速作出响应,利用该方法得到的功率变化曲线振荡幅度较小,但是很难精确地定位到最大功率点(MPP).因此,文章提出一种混合控制的最大功率点跟踪(MPPT)策略,先利用改进粒子群算法快速跟踪到MPP附近,然后利用小步长电导增量法对MPP进行精细搜索.仿真结果表明,该跟踪策略在一定程度上能够增加跟踪系统的响应速度、跟踪精度,减小功率变化曲线的振荡幅度.

针对传统的航空发动机故障诊断方法正确率较低,并且对异常数据不敏感的问题,将智能诊断算法引入航空发动机气路故障诊断领域.以涡轴发动机为例,分析了常见气路部件故障类型的成因和表现,并在Tensorflow上建立基于深度信念网络的故障诊断模型.与传统的故障诊断方法相比,具有更高的故障诊断正确率.