ABSTRACT
Abdominal tumor segmentation is a crucial yet challenging step during the screening and diagnosis of tumors. While 3D segmentation models provide powerful performance, they demand substantial computational resources. Additionally, in 3D data, tumors often represent a small portion, leading to imbalanced data and potentially overlooking crucial information. Conversely, 2D segmentation models have a lightweight structure, but disregard the inter-slice correlation, risking the loss of tumor in edge slices. To address these challenges, this paper proposes a novel Position-Aware and Key Slice Feature Sharing 2D tumor segmentation model (PAKS-Net). Leveraging the Swin-Transformer, we effectively model the global features within each slice, facilitating essential information extraction. Furthermore, we introduce a Position-Aware module to capture the spatial relationship between tumors and their corresponding organs, mitigating noise and interference from surrounding organ tissues. To enhance the edge slice segmentation accuracy, we employ key slices to assist in the segmentation of other slices to prioritize tumor regions. Through extensive experiments on three abdominal tumor segmentation CT datasets and a lung tumor segmentation CT dataset, PAKS-Net demonstrates superior performance, reaching 0.893, 0.769, 0.598 and 0.738 tumor DSC on the KiTS19, LiTS17, pancreas and LOTUS datasets, surpassing 3D segmentation models, while remaining computationally efficient with fewer parameters.
Subject(s)
Abdominal Neoplasms , Humans , Abdominal Neoplasms/diagnostic imaging , Tomography, X-Ray Computed/methods , Imaging, Three-Dimensional/methods , AlgorithmsABSTRACT
BACKGROUND: Osimertinib, a third-generation epidermal growth factor receptor-tyrosine kinase inhibitor (EGFR-TKI), is the preferred treatment for EGFR-mutated lung cancer. However, acquired resistance inevitably develops. While non-coding RNAs have been implicated in lung cancer through various functions, the molecular mechanisms responsible for osimertinib resistance remain incompletely elucidated. METHODS: RNA-sequencing technology was employed to determine differentially expressed lncRNAs (DE-lncRNAs) and mRNAs (DE-mRNAs) between H1975 and H1975OR cell lines. Starbase 2.0 was utilized to predict DE-lncRNA and DE-mRNA interactions, constructing ceRNA networks. Subsequently, functional and pathway enrichment analysis were performed on target DE-mRNAs to identify pathways associated with osimertinib resistance. Key target DE-mRNAs were then selected as potential risk signatures for lung adenocarcinoma (LUAD) prognostic modeling using multivariate Cox regression analyses. The Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR) and immunohistochemistry staining were used for result validation. RESULTS: Functional analysis revealed that the identified DE-mRNAs primarily enriched in EGFR-TKI resistance pathways, especially in the PI3K/Akt signaling pathway, where their concerted actions may lead to osimertinib resistance. Specifically, upregulation of LINC00313 enhanced COL1A1 expression by acting as a miR-218-5p sponge, triggering an upstream response that activates the PI3K/Akt pathway, potentially contributing to osimertinib resistance. Furthermore, the expressions of LINC00313 and COL1A1 were validated by qRT-PCR, and the activation of the PI3K/Akt pathway was confirmed by immunohistochemistry staining. CONCLUSIONS: Our results suggest that the LINC00313/miR-218-5p/COL1A1 axis potentially contributes to osimertinib resistance through the PI3K/Akt signaling pathway, providing novel insights into the molecular mechanisms underlying acquired osimertinib resistance in LUAD. Additionally, our study may aid in the identification of potential therapeutic targets for overcoming resistance to osimertinib.
ABSTRACT
Semi-supervised learning has garnered significant interest as a method to alleviate the burden of data annotation. Recently, semi-supervised medical image segmentation has garnered significant interest that can alleviate the burden of densely annotated data. Substantial advancements have been achieved by integrating consistency-regularization and pseudo-labeling techniques. The quality of the pseudo-labels is crucial in this regard. Unreliable pseudo-labeling can result in the introduction of noise, leading the model to converge to suboptimal solutions. To address this issue, we propose learning from reliable pseudo-labels. In this paper, we tackle two critical questions in learning from reliable pseudo-labels: which pseudo-labels are reliable and how reliable are they? Specifically, we conduct a comparative analysis of two subnetworks to address both challenges. Initially, we compare the prediction confidence of the two subnetworks. A higher confidence score indicates a more reliable pseudo-label. Subsequently, we utilize intra-class similarity to assess the reliability of the pseudo-labels to address the second challenge. The greater the intra-class similarity of the predicted classes, the more reliable the pseudo-label. The subnetwork selectively incorporates knowledge imparted by the other subnetwork model, contingent on the reliability of the pseudo labels. By reducing the introduction of noise from unreliable pseudo-labels, we are able to improve the performance of segmentation. To demonstrate the superiority of our approach, we conducted an extensive set of experiments on three datasets: Left Atrium, Pancreas-CT and Brats-2019. The experimental results demonstrate that our approach achieves state-of-the-art performance. Code is available at: https://github.com/Jiawei0o0/mutual-learning-with-reliable-pseudo-labels.
Subject(s)
Heart Atria , Supervised Machine Learning , Humans , Reproducibility of Results , Tomography, X-Ray Computed , Image Processing, Computer-AssistedABSTRACT
Curvature of carbon materials has gained significant attention as catalysts due to their distinctive properties and potential applications. This review comprehensively summarizes how the bending of carbon materials can improve electrocatalytic performance, with special attention to the applications of various bent carbon materials (such as carbon nanotubes, graphene, and fullerene) in electrocatalysts and a large number of related density functional theory (DFT) theoretical calculations. Extensive mechanism research has provided a wealth of evidence indicating that the curvature of carbon materials has a profound impact on catalytic activity. This improvement in catalytic performance by curved carbon materials is attributed to factors like a larger active surface area, modulation of electronic structure, and better dispersal of catalytic active sites. A comprehensive understanding and utilization of these effects enable the design of highly efficient carbon-based catalysts for applications in energy conversion, environmental remediation, and chemical synthesis.
ABSTRACT
BACKGROUND/AIMS: To improve the accuracy of pterygium screening and detection through smartphones, we established a fusion training model by blending a large number of slit-lamp image data with a small proportion of smartphone data. METHOD: Two datasets were used, a slit-lamp image dataset containing 20 987 images and a smartphone-based image dataset containing 1094 images. The RFRC (Faster RCNN based on ResNet101) model for the detection model. The SRU-Net (U-Net based on SE-ResNeXt50) for the segmentation models. The open-cv algorithm measured the width, length and area of pterygium in the cornea. RESULTS: The detection model (trained by slit-lamp images) obtained the mean accuracy of 95.24%. The fusion segmentation model (trained by smartphone and slit-lamp images) achieved a microaverage F1 score of 0.8981, sensitivity of 0.8709, specificity of 0.9668 and area under the curve (AUC) of 0.9295. Compared with the same group of patients' smartphone and slit-lamp images, the fusion model performance in smartphone-based images (F1 score of 0.9313, sensitivity of 0.9360, specificity of 0.9613, AUC of 0.9426, accuracy of 92.38%) is close to the model (trained by slit-lamp images) in slit-lamp images (F1 score of 0.9448, sensitivity of 0.9165, specificity of 0.9689, AUC of 0.9569 and accuracy of 94.29%). CONCLUSION: Our fusion model method got high pterygium detection and grading accuracy in insufficient smartphone data, and its performance is comparable to experienced ophthalmologists and works well in different smartphone brands.
Subject(s)
Conjunctiva/abnormalities , Pterygium , Smartphone , Humans , Pterygium/diagnosis , Cornea , Slit LampABSTRACT
Purpose: Accurate quantification measurement of tear meniscus is vital for the precise diagnosis of dry eye. In current clinical practice, the measurement of tear meniscus height (TMH) relies on doctors' manual operation. This study aims to propose a novel automatic artificial intelligence (AI) system to evaluate TMH. Methods: A total of 510 photographs obtained by the oculus camera were labeled. Three thousand and five hundred images were finally attained by data enhancement to train the neural network model parameters, and 60 were used to evaluate the model performance in segmenting the cornea and tear meniscus region. One hundred images were used to test generalization ability of the model. We modified a segmentation model of the cornea and the tear meniscus based on the UNet-like network. The output of the segmentation model is followed by a calculation module that calculates and reports the TMH. Results: Compared with ground truth (GT) manually labeled by clinicians, our modified model achieved a Dice Similarity Coefficient (DSC) and Intersection over union (Iou) of 0.99/0.98 in the corneal segmentation task and 0.92/0.86 for the detection of tear meniscus on the validation set, respectively. On the test set, the TMH automatically measured by our AI system strongly correlates with the results manually calculated by the ophthalmologists. Conclusions: We developed a fully automated and reliable AI system to obtain TMH. After large-scale clinical testing, our method could be used for dry eye screening in clinical practice.
Subject(s)
Dry Eye Syndromes , Meniscus , Humans , Artificial Intelligence , Neural Networks, Computer , Cornea , Dry Eye Syndromes/diagnosisABSTRACT
Conventional manufacturing techniques to fabricate microfluidic chips, such as soft lithography and hot embossing process, have limitations that include difficulty in preparing multiple-layered structures, cost- and labor-consuming fabrication process, and low productivity. Digital light processing (DLP) technology has recently emerged as a cost-efficient microfabrication approach for the 3D printing of microfluidic chips; however, the fabrication resolution for microchannels is still limited to sub-100 microns at best. Here, we developed an innovative DLP printing strategy for high resolution and scalable microchannel fabrication by dosing- and zoning-controlled vat photopolymerization (DZC-VPP). Specifically, we proposed a modified mathematical model to precisely predict the accumulated UV irradiance for resin photopolymerization, thereby providing guidance for the fabrication of microchannels with enhanced resolution. By fine-tuning the printing parameters, including optical irradiance, exposure time, projection region, and step distance, we can precisely tailor the penetration irradiance stemming from the photopolymerization of the neighboring resin layers, thereby preventing channel blockage due to UV overexposure or compromised bonding stability owing to insufficient resin curing. Remarkably, this strategy can allow the preparation of microchannels with cross-sectional dimensions of 20 µm × 20 µm using a commercial printer with a pixel size of 10 µm × 10 µm; this is significantly higher resolution than previous reports. In addition, this method can enable the scalable and biocompatible fabrication of microfluidic drop-maker units that can be used for cell encapsulation. In general, the current DZC-VPP method can enable major advances in precise and scalable microchannel fabrication and represents a significant step forward for widespread applications of microfluidics-based techniques in biomedical fields.
ABSTRACT
Background: Corneal fluorescein sodium staining is a valuable diagnostic method for various ocular surface diseases. However, the examination results are highly dependent on the subjective experience of ophthalmologists. Objectives: To develop an artificial intelligence system based on deep learning to provide an accurate quantitative assessment of sodium fluorescein staining score and the size of cornea epithelial patchy defect. Design: A prospective study. Methods: We proposed an artificial intelligence system for automatically evaluating corneal staining scores and accurately measuring patchy corneal epithelial defects based on corneal fluorescein sodium staining images. The design incorporates two segmentation models and a classification model to forecast and assess the stained images. Meanwhile, we compare the evaluation findings from the system with ophthalmologists with varying expertise. Results: For the segmentation task of cornea boundary and cornea epithelial patchy defect area, our proposed method can achieve the performance of dice similarity coefficient (DSC) is 0.98/0.97 and Hausdorff distance (HD) is 3.60/8.39, respectively, when compared with the manually labeled gold standard. This method significantly outperforms the four leading algorithms (Unet, Unet++, Swin-Unet, and TransUnet). For the classification task, our algorithm achieves the best performance in accuracy, recall, and F1-score, which are 91.2%, 78.6%, and 79.2%, respectively. The performance of our developed system exceeds seven different approaches (Inception, ShuffleNet, Xception, EfficientNet_B7, DenseNet, ResNet, and VIT) in classification tasks. In addition, three ophthalmologists were selected to rate corneal staining images. The results showed that the performance of our artificial intelligence system significantly outperformed the junior doctors. Conclusion: The system offers a promising automated assessment method for corneal fluorescein staining, decreasing incorrect evaluations caused by ophthalmologists' subjective variance and limited knowledge.
ABSTRACT
Thrombotic thrombocytopenic purpura (TTP) is a rare and life-threatening thrombotic microangiopathy characterized by microangiopathic hemolytic anemia, severe thrombocytopenia, and organ ischemia associated with disseminated microvascular platelet-rich thrombus. Before the introduction of plasma therapy, acute TTP was almost universally fatal, which improved survival from < 10 to 80-90%. However, patients who survived an acute attack were at high risk for recurrence and long-term morbidity. It was reported that daratumumab can eradicate persistent ADAMTS13-inhibiting autoantibodies and restore ADAMTS13 activity in two patients with relapsed immune-mediated TTP without associated adverse drug reactions. Here we report a case series of patients with initial diagnosed acquired TTP treated with combination regimens containing daratumumab. All the patients achieved clinical response after the initial treatment. Three patients achieved clinical remission, one patient relapsed and one patient suffered an exacerbation during follow-up. The two patients were retreated with glucocorticoids, plasma exchange combined with daratumumab, and clinical remission was achieved again. Combination of daratumumab in the treatment of initial diagnosed acquired thrombotic thrombocytopenic purpura can rapidly restore ADAMST13 activity and turn negative for ADAMST13 inhibitors, resulting in long-term remission in patients.
Subject(s)
Purpura, Thrombotic Thrombocytopenic , Humans , Purpura, Thrombotic Thrombocytopenic/diagnosis , Purpura, Thrombotic Thrombocytopenic/drug therapy , Antibodies, Monoclonal/therapeutic use , Plasma Exchange/methods , ADAMTS13 ProteinABSTRACT
Recent studies show that deep person re-identification (re-ID) models are vulnerable to adversarial examples, so it is critical to improving the robustness of re-ID models against attacks. To achieve this goal, we explore the strengths and weaknesses of existing re-ID models, i.e., designing learning-based attacks and training robust models by defending against the learned attacks. The contributions of this paper are three-fold: First, we build a holistic attack-defense framework to study the relationship between the attack and defense for person re-ID. Second, we introduce a combinatorial adversarial attack that is adaptive to unseen domains and unseen model types. It consists of distortions in pixel and color space (i.e., mimicking camera shifts). Third, we propose a novel virtual-guided meta-learning algorithm for our attack-defense system. We leverage a virtual dataset to conduct experiments under our meta-learning framework, which can explore the cross-domain constraints for enhancing the generalization of the attack and the robustness of the re-ID model. Comprehensive experiments on three large-scale re-ID benchmarks demonstrate that: 1) Our combinatorial attack is effective and highly universal in cross-model and cross-dataset scenarios; 2) Our meta-learning algorithm can be readily applied to different attack and defense approaches, which can reach consistent improvement; 3) The defense model trained on the learning-to-learn framework is robust to recent SOTA attacks that are not even used during training.
ABSTRACT
Bone remodeling is a tightly coupled process between bone forming osteoblasts (OBs) and bone resorbing osteoclasts (OCs) to maintain bone architecture and systemic mineral homeostasis throughout life. However, the mechanisms responsible for the coupling between OCs and OBs have not been fully elucidated. Herein, we first validate that secreted extracellular vesicles by osteoclasts (OC-EVs) promote osteogenic differentiation of mesenchymal stem cells (MSCs) and further demonstrate the efficacy of osteoclasts and their secreted EVs in treating tibial bone defects. Furthermore, we show that OC-EVs contain several osteogenesis-promoting proteins as cargo. By employing proteomic and functional analysis, we reveal that mature osteoclasts secrete thrombin cleaved phosphoprotein 1 (SPP1) through extracellular vesicles which triggers MSCs osteogenic differentiation into OBs by activating Transforming Growth Factor ß1 (TGFß1) and Smad family member 3 (SMAD3) signaling. In conclusion, our findings prove an important role of SPP1, present as cargo in OC-derived EVs, in signaling to MSCs and driving their differentiation into OBs. This biological mechanism implies a paradigm shift regarding the role of osteoclasts and their signaling toward the treatment of skeletal disorders which require bone formation.
Subject(s)
Extracellular Vesicles , Osteoclasts , Osteoclasts/metabolism , Osteogenesis , Transforming Growth Factor beta1/metabolism , Proteomics , Bone Regeneration , Osteoblasts , Cell Differentiation , Extracellular Vesicles/metabolismABSTRACT
In this paper, we study the cross-view geo-localization problem to match images from different viewpoints. The key motivation underpinning this task is to learn a discriminative viewpoint-invariant visual representation. Inspired by the human visual system for mining local patterns, we propose a new framework called RK-Net to jointly learn the discriminative Representation and detect salient Keypoints with a single Network. Specifically, we introduce a Unit Subtraction Attention Module (USAM) that can automatically discover representative keypoints from feature maps and draw attention to the salient regions. USAM contains very few learning parameters but yields significant performance improvement and can be easily plugged into different networks. We demonstrate through extensive experiments that (1) by incorporating USAM, RK-Net facilitates end-to-end joint learning without the prerequisite of extra annotations. Representation learning and keypoint detection are two highly-related tasks. Representation learning aids keypoint detection. Keypoint detection, in turn, enriches the model capability against large appearance changes caused by viewpoint variants. (2) USAM is easy to implement and can be integrated with existing methods, further improving the state-of-the-art performance. We achieve competitive geo-localization accuracy on three challenging datasets, i. e., University-1652, CVUSA and CVACT. Our code is available at https://github.com/AggMan96/RK-Net.
ABSTRACT
Epidermal growth factor receptor tyrosine kinase inhibitors (EGFR-TKIs) are widely used for patients with EGFR-mutated lung cancer. Despite its initial therapeutic efficacy, most patients eventually develop drug resistance, which leads to a poor prognosis in lung cancer patients. Previous investigations have proved that non-coding RNAs including long non-coding RNAs (lncRNAs), circular RNAs (circRNAs), and microRNAs (miRNAs) contribute to drug resistance by various biological functions, whereas how they regulate EGFR-TKI resistance remains unclear. In this study, we examined gene expression using the microarray technology on gefitinib-resistant NSCLC cells to obtain differentially expressed (DE) lncRNAs and mRNAs. A total of 45 DE-lncRNAs associated with overall survival and 1799 target DE-mRNAs were employed to construct a core lncRNA-miRNA-mRNA network to illustrate underlying molecular mechanisms of how EGFR-TKI resistance occurs in NSCLC. We found that target DE-mRNAs were mainly enriched in pathways involved in EGFR-TKI resistance, especially the target DE-mRNAs regulated by LINC01128 were significantly enriched in the PI3K/Akt signaling pathway, where the synergy of these target DE-mRNAs may play a key role in EGFR-TKI resistance. In addition, downregulated LINC01128, acting as a specific miRNA sponge, decreases PTEN via sponging miR-25-3p. Furthermore, signaling reactions caused by the downregulation of PTEN would activate the PI3K/Akt signaling pathway, which may lead to EGFR-TKI resistance. In addition, a survival analysis indicated the low expression of LINC01128, and PTEN is closely related to poor prognosis in lung adenocarcinoma (LUAD). Therefore, the LINC01128/miR-25-3p/PTEN axis may promote EGFR-TKI resistance via the PI3K/Akt signaling pathway, which provides new insights into the underlying molecular mechanisms of drug resistance to EGFR-TKIs in NSCLC. In addition, our study sheds light on developing novel therapeutic approaches to overcome EGFR-TKI resistance in NSCLC.
ABSTRACT
Carotid artery lumen diameter (CALD) and carotid artery intima-media thickness (CIMT) are essential factors for estimating the risk of many cardiovascular diseases. The automatic measurement of them in ultrasound (US) images is an efficient assisting diagnostic procedure. Despite the advances, existing methods still suffer the issue of low measuring accuracy and poor prediction stability, mainly due to the following disadvantages: (1) ignore anatomical prior and prone to give anatomically inaccurate estimation; (2) require carefully designed post-processing, which may introduce more estimation errors; (3) rely on massive pixel-wise annotations during training; (4) can not estimate the uncertainty of the predictions. In this study, we propose the Anatomical Prior-guided ReInforcement Learning model (APRIL), which innovatively formulate the measurement of CALD & CIMT as an RL problem and dynamically incorporate anatomical prior (AP) into the system through a novel reward. With the guidance of AP, the designed keypoints in APRIL can avoid various anatomy impossible mis-locations, and accurately measure the CALD & CIMT based on their corresponding locations. Moreover, this formulation significantly reduces human annotation effort by only using several keypoints and can help to eliminate the extra post-processing steps. Further, we introduce an uncertainty module for measuring the prediction variance, which can guide us to adaptively rectify the estimation of those frames with considerable uncertainty. Experiments on a challenging carotid US dataset show that APRIL can achieve MAE (in pixel/mm) of 3.02±2.23 / 0.18±0.13 for CALD, and 0.96±0.70 / 0.06±0.04 for CIMT, which significantly surpass popular approaches that use more annotations.
Subject(s)
Cardiovascular Diseases , Carotid Intima-Media Thickness , Carotid Arteries/diagnostic imaging , Humans , Tumor Necrosis Factor Ligand Superfamily Member 13 , UltrasonographyABSTRACT
BACKGROUND This study assessed the role of different immune phenotypes of T cells in virus-induced acute exacerbation of chronic obstructive pulmonary disease (AECOPD). MATERIAL AND METHODS The study involved 103 participants, including individuals with virus-induced AECOPD (n=32), non-virus-induced AECOPD (n=31), and stable COPD (n=20) and individuals who were healthy smokers (n=20). The immune phenotypes of T cells in peripheral blood were evaluated via flow cytometry analysis, and the differences were analyzed. RESULTS Patients with virus-induced AECOPD (virus group) had a higher COPD assessment test score on admission than those in the group with non-virus-induced AECOPD (nonvirus group; 25.6±3.8 vs 21.9±4.8, P=0.045). A lower CD4⺠human leukocyte antigen-DR (HLA-DR)+ frequency was found in the peripheral blood of the virus group compared with the nonvirus group (2.2 vs 4.2, P=0.015), and the frequency of CD4⺠CD25high CD127low HLA-DR⺠in CD4⺠in the virus group was lower than in the nonvirus group (1.1 vs 3.6, P=0.011). The CD3âº, CD4âº, CD8âº, CD4⺠central memory T cell, CD4⺠effector memory T cell (Tem), CD4⺠end-stage T cell, and CD8⺠Tem levels in lymphocytes of peripheral blood were lower in exacerbation groups relative to those in the stable COPD and healthy smoking groups, but similar between exacerbation groups. Similar frequencies and levels of T cells between different stagings of COPD were also identified. CONCLUSIONS The expression of HLA-DR on the cell surface of CD4⺠regulatory T cells (Tregs) was lower in the peripheral blood of patients with virus-induced AECOPD. The expression of HLA-DR in CD4⺠Tregs suggested the effect of respiratory viruses on adaptive immunity of patients with AECOPD to some extent.
Subject(s)
HLA-DR Antigens/metabolism , Pulmonary Disease, Chronic Obstructive/immunology , T-Lymphocytes, Regulatory/immunology , Adaptive Immunity , Aged , CD4-Positive T-Lymphocytes/immunology , CD8-Positive T-Lymphocytes/immunology , China , Female , Flow Cytometry , Gene Expression/genetics , HLA-DR Antigens/analysis , HLA-DR Antigens/immunology , Humans , Male , Middle Aged , Pulmonary Disease, Chronic Obstructive/physiopathology , Pulmonary Disease, Chronic Obstructive/virology , Smoking/immunology , VirusesABSTRACT
Retinal vessel segmentation is a critical procedure towards the accurate visualization, diagnosis, early treatment, and surgery planning of ocular diseases. Recent deep learning-based approaches have achieved impressive performance in retinal vessel segmentation. However, they usually apply global image pre-processing and take the whole retinal images as input during network training, which have two drawbacks for accurate retinal vessel segmentation. First, these methods lack the utilization of the local patch information. Second, they overlook the geometric constraint that retina only occurs in a specific area within the whole image or the extracted patch. As a consequence, these global-based methods suffer in handling details, such as recognizing the small thin vessels, discriminating the optic disk, etc. To address these drawbacks, this study proposes a Global and Local enhanced residual U-nEt (GLUE) for accurate retinal vessel segmentation, which benefits from both the globally and locally enhanced information inside the retinal region. Experimental results on two benchmark datasets demonstrate the effectiveness of the proposed method, which consistently improves the segmentation accuracy over a conventional U-Net and achieves competitive performance compared to the state-of-the-art.
Subject(s)
Deep Learning , Diagnostic Techniques, Ophthalmological , Image Interpretation, Computer-Assisted/methods , Retinal Vessels/diagnostic imaging , Algorithms , Databases, Factual , Humans , Retinal Diseases/diagnostic imagingABSTRACT
This work considers the problem of unsupervised domain adaptation in person re-identification (re-ID), which aims to transfer knowledge from the source domain to the target domain. Existing methods are primary to reduce the inter-domain shift between the domains, which however usually overlook the relations among target samples. This paper investigates into the intra-domain variations of the target domain and proposes a novel adaptation framework w.r.t three types of underlying invariance, i.e., Exemplar-Invariance, Camera-Invariance, and Neighborhood-Invariance. Specifically, an exemplar memory is introduced to store features of samples, which can effectively and efficiently enforce the invariance constraints over the global dataset. We further present the Graph-based Positive Prediction (GPP) method to explore reliable neighbors for the target domain, which is built upon the memory and is trained on the source samples. Experiments demonstrate that 1) the three invariance properties are complementary and indispensable for effective domain adaptation, 2) the memory plays a key role in implementing invariance learning and improves the performance with limited extra computation cost, 3) GPP can facilitate the invariance learning and thus significantly improves the results, and 4) our approach produces new state-of-the-art adaptation accuracy on three re-ID large-scale benchmarks.
ABSTRACT
Sugarcane white leaf (SCWL) is a devastating sugarcane (Saccharum officinarum) disease caused by a 16SrXI group phytoplasma, which is extremely harmful to sugarcane production. To determine the occurrence of SCWL in different varieties in 2018, we conducted a field survey and performed nested PCR detection of SCWL phytoplasma in cane-planting areas of Mangweng and Hepai in Gengma, Yunnan province, which are the areas most severely affected by SCWL in China. The results of the field survey showed that the symptomatic incidence of SCWL differed among varieties. The mean symptomatic incidence of SCWL on variety Yuetang60 was the highest (73.50%), and it was the lowest on Liucheng05-136 (13.67%). Using nested PCR, the SCWL phytoplasma was detected in symptomatic plants of all varieties more than 90% of the time; the SCWL phytoplasma was detected in 91 and 97% of symptomatic plants of Yingyu91-59 and Liucheng05-136 varieties, respectively. The SCWL phytoplasma was detected by PCR in 82% of the asymptomatic plant samples. The results of this study showed that field survey based on white leaf symptoms did not accurately reflect the actual occurrence of the SCWL phytoplasma.