RESUMO
The hybrid deep models of Vision Transformer (ViT) and Convolution Neural Network (CNN) have emerged as a powerful class of backbones for vision tasks. Scaling up the input resolution of such hybrid backbones naturally strengthes model capacity, but inevitably suffers from heavy computational cost that scales quadratically. Instead, we present a new hybrid backbone with HIgh-Resolution Inputs (namely HIRI-ViT), that upgrades prevalent four-stage ViT to five-stage ViT tailored for high-resolution inputs. HIRI-ViT is built upon the seminal idea of decomposing the typical CNN operations into two parallel CNN branches in a cost-efficient manner. One high-resolution branch directly takes primary high-resolution features as inputs, but uses less convolution operations. The other low-resolution branch first performs down-sampling and then utilizes more convolution operations over such low-resolution features. Experiments on both recognition task (ImageNet-1K dataset) and dense prediction tasks (COCO and ADE20 K datasets) demonstrate the superiority of HIRI-ViT. More remarkably, under comparable computational cost ( â¼ 5.0 GFLOPs), HIRI-ViT achieves to-date the best published Top-1 accuracy of 84.3% on ImageNet with 448×448 inputs, which absolutely improves 83.4% of iFormer-S by 0.9% with 224×224 inputs.
RESUMO
Stress can be a double-edged sword. Given the intricacy of the innovation process, the link between job stress and individual innovation behavior remains uncertain. To clarify the relationship between challenge stressors and the innovative behavior of higher education teachers, this study was based on the conservation of resources (COR) theory and adopted the structural equation modeling method to explore the impact of challenge stressors on the innovative behavior of higher education teachers and reveal its influencing mechanism and boundary conditions. By analyzing the data from 208 questionnaires of higher education teachers, the findings reveal that challenge stressors positively influence innovative behavior, with task crafting serving as a critical bridge. Additionally, a favorable employment relationship climate enhances the positive impact of challenge stressors. Gender dynamics are also explored, adding nuance to the understanding of this relationship. These results shed light on the inherent mechanisms governing the relationship between challenge stressors and innovative behavior among higher education teachers, and underscore the significance of task crafting. In addition, the discoveries provided fresh insights and ideas for investigating how organizational climate affects individual innovative behavior.
Assuntos
Estresse Ocupacional , Humanos , Masculino , Feminino , Adulto , Criatividade , Inquéritos e Questionários , DocentesRESUMO
As two major types of primary liver cancers, the tumor immune microenvironment (TIME) of hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma (ICC) have been well studied separately. However, a systemic assessment of the similarities and differences between the TIME of HCC and ICC is still lacking. In this study, we pictured a landscape of combined TIME of HCC and ICC by sequencing and integrating 41 single-cell RNA-seq samples from four different tissue types of both malignancies. We found that T cells in HCC tumors generally exhibit higher levels of immunosuppression and exhaustion than those in ICC tumors. Myeloid cells in HCC and ICC tumors also exhibit distinct phenotypes and may serve as a key factor driving the differences between their TIMEs. Besides, we identified a cluster of EGR1+ macrophages specifically enriched in HCC tumors. Together, our study provides new insights into cellular composition, states and interactions in the TIMEs of HCC and ICC, which could pave the way for the development of future therapeutic targets for liver cancers.
Assuntos
Neoplasias dos Ductos Biliares , Carcinoma Hepatocelular , Colangiocarcinoma , Neoplasias Hepáticas , Humanos , Carcinoma Hepatocelular/genética , Carcinoma Hepatocelular/patologia , Neoplasias Hepáticas/genética , Neoplasias Hepáticas/patologia , Neoplasias dos Ductos Biliares/genética , Neoplasias dos Ductos Biliares/patologia , Colangiocarcinoma/genética , Colangiocarcinoma/patologia , Ductos Biliares Intra-Hepáticos/patologia , Microambiente TumoralRESUMO
Transformer with self-attention has led to the revolutionizing of natural language processing field, and recently inspires the emergence of Transformer-style architecture design with competitive results in numerous computer vision tasks. Nevertheless, most of existing designs directly employ self-attention over a 2D feature map to obtain the attention matrix based on pairs of isolated queries and keys at each spatial location, but leave the rich contexts among neighbor keys under-exploited. In this work, we design a novel Transformer-style module, i.e., Contextual Transformer (CoT) block, for visual recognition. Such design fully capitalizes on the contextual information among input keys to guide the learning of dynamic attention matrix and thus strengthens the capacity of visual representation. Technically, CoT block first contextually encodes input keys via a 3×3 convolution, leading to a static contextual representation of inputs. We further concatenate the encoded keys with input queries to learn the dynamic multi-head attention matrix through two consecutive 1×1 convolutions. The learnt attention matrix is multiplied by input values to achieve the dynamic contextual representation of inputs. The fusion of the static and dynamic contextual representations are finally taken as outputs. Our CoT block is appealing in the view that it can readily replace each 3×3 convolution in ResNet architectures, yielding a Transformer-style backbone named as Contextual Transformer Networks (CoTNet). Through extensive experiments over a wide range of applications (e.g., image recognition, object detection, instance segmentation, and semantic segmentation), we validate the superiority of CoTNet as a stronger backbone. Source code is available at https://github.com/JDAI-CV/CoTNet.
RESUMO
Unsupervised learning is just at a tipping point where it could really take off. Among these approaches, contrastive learning has led to state-of-the-art performance. In this paper, we construct a novel probabilistic graphical model that effectively incorporates the low rank promoting prior into the framework of contrastive learning, referred to as LORAC. In contrast to the existing conventional self-supervised approaches that only considers independent learning, our hypothesis explicitly requires that all the samples belonging to the same instance class lie on the same subspace with small dimension. This heuristic poses particular joint learning constraints to reduce the degree of freedom of the problem during the search of the optimal network parameterization. Most importantly, we argue that the low rank prior employed here is not unique, and many different priors can be invoked in a similar probabilistic way, corresponding to different hypotheses about underlying truth behind the contrastive features. Empirical evidences show that the proposed algorithm clearly surpasses the state-of-the-art approaches on multiple benchmarks, including image classification, object detection, instance segmentation and keypoint detection. Code is available: https://github.com/ssl-codelab/lorac.
RESUMO
Recent advances have presented several strategies to mitigate the computations of self-attention mechanism with high-resolution inputs. Many of these works consider decomposing the global self-attention procedure over image patches into regional and local feature extraction procedures that each incurs a smaller computational complexity. Despite good efficiency, these approaches seldom explore the holistic interactions among all patches, and are thus difficult to fully capture the global semantics. In this paper, we propose a novel Transformer architecture that elegantly exploits the global semantics for self-attention learning, namely Dual Vision Transformer (Dual-ViT). The new architecture incorporates a critical semantic pathway that can more efficiently compress token vectors into global semantics with reduced order of complexity. Such compressed global semantics then serve as useful prior information in learning finer local pixel level details, through another constructed pixel pathway. The semantic pathway and pixel pathway are integrated together and are jointly trained, spreading the enhanced self-attention information in parallel through both of the pathways. Dual-ViT is henceforth able to capitalize on global semantics to boost self-attention learning without compromising much computational complexity. We empirically demonstrate that Dual-ViT provides superior accuracy than SOTA Transformer architectures with comparable training complexity.
RESUMO
Background: The aim of this study was to investigate the prognostic factors influencing the outcome of patients with Barcelona Clinic Liver Cancer stage C hepatocellular carcinoma (HCC) receiving salvage surgery after conversion therapy based on tyrosine kinase inhibitors (TKIs) and anti-programmed death-1 (PD-1) antibodies. Methods: From June 2018 to December 2022, patients receiving salvage surgery after conversion therapy based on PD-1 and TKIs at the Faculty of Hepato-Pancreato-Biliary Surgery, Chinese PLA General Hospital were retrospectively recruited for this study. Overall survival (OS) and recurrence-free survival (RFS) were observed as the primary end point in the Cox analysis of prognostic factors among this study. Results: The 6- and 12-month RFS rates were 77.0% and 64.8%, respectively, while the 6-, 12-, 24-, and 36-month OS rates were 98.4%, 93.4%, 76.8%, and 69.8%, respectively. The median OS and RFS were not reached. On multivariable Cox regression analyses, low serum alpha fetoprotein (AFP) level (≤20 ng/mL) after conversion therapy [hazard ratio (HR) 0.186, 95% CI: 0.039-0.887; P=0.035) and microvascular invasion (MVI) grade II (HR 3.054, 95% CI: 1.000-9.329; P=0.050) were independent factors associated with a higher OS and RFS. Conclusions: For patients with Barcelona Clinic Liver Cancer stage C (BCLC-C) HCC, lower AFP level after conversion therapy (<20 ng/mL) and MVI II were associated with a higher OS and lower RFS rate, respectively.
RESUMO
Recent progress on 2D object detection has featured Cascade RCNN, which capitalizes on a sequence of cascade detectors to progressively improve proposal quality, towards high-quality object detection. However, there has not been evidence in support of building such cascade structures for 3D object detection, a challenging detection scenario with highly sparse LiDAR point clouds. In this work, we present a simple yet effective cascade architecture, named 3D Cascade RCNN, that allocates multiple detectors based on the voxelized point clouds in a cascade paradigm, pursuing higher quality 3D object detector progressively. Furthermore, we quantitatively define the sparsity level of the points within 3D bounding box of each object as the point completeness score, which is exploited as the task weight for each proposal to guide the learning of each stage detector. The spirit behind is to assign higher weights for high-quality proposals with relatively complete point distribution, while down-weight the proposals with extremely sparse points that often incur noise during training. This design of completeness-aware re-weighting elegantly upgrades the cascade paradigm to be better applicable for the sparse input data, without increasing any FLOP budgets. Through extensive experiments on both the KITTI dataset and Waymo Open Dataset, we validate the superiority of our proposed 3D Cascade RCNN, when comparing to state-of-the-art 3D object detection techniques. The source code is publicly available at https://github.com/caiqi/Cascasde-3D.
RESUMO
Hepatocellular carcinoma (HCC) represents a paradigm of the relation between tumor microenvironment (TME) and tumor development. Here, we generate a single-cell atlas of the multicellular ecosystem of HCC from four tissue sites. We show the enrichment of central memory T cells (TCM) in the early tertiary lymphoid structures (E-TLSs) in HCC and assess the relationships between chronic HBV/HCV infection and T cell infiltration and exhaustion. We find the MMP9+ macrophages to be terminally differentiated tumor-associated macrophages (TAMs) and PPARγ to be the pivotal transcription factor driving their differentiation. We also characterize the heterogeneous subpopulations of malignant hepatocytes and their multifaceted functions in shaping the immune microenvironment of HCC. Finally, we identify seven microenvironment-based subtypes that can predict prognosis of HCC patients. Collectively, this large-scale atlas deepens our understanding of the HCC microenvironment, which might facilitate the development of new immune therapy strategies for this malignancy.
Assuntos
Carcinoma Hepatocelular , Neoplasias Hepáticas , Estruturas Linfoides Terciárias , Carcinoma Hepatocelular/patologia , Ecossistema , Humanos , Neoplasias Hepáticas/patologia , Microambiente TumoralRESUMO
Recent advances in video object detection have characterized the exploration of temporal coherence across frames to enhance object detector. Nevertheless, previous solutions either rely on additional inputs (e.g., optical flow) to guide feature aggregation, or complex post-processing to associate bounding boxes. In this paper, we introduce a simple but effective design that learns instance identifiers for instance association in a meta-learning paradigm, which requires no auxiliary inputs or post-processing. Specifically, we present Meta-Learnt Instance Identifier Networks (namely MINet) that novelly meta-learns instance identifiers to recognize identical instances across frames in a single forward-pass, leading to the robust online linking of instances. Technically, depending on the detection results of previous frames, we teach MINet to learn the weights of an instance identifier on the fly, which can be well applied to up-coming frames. Such meta-learning paradigm enables instance identifiers to be flexibly adapted to novel frames at inference. Furthermore, MINet writes/updates the detection results of previous instances into memory and reads from memory when performing inference to encourage temporal consistency for video object detection. Our MINet is appealing in the sense that it is pluggable to any object detection model. Extensive experiments on ImageNet VID dataset demonstrate the superiority of MINet. More remarkably, by integrating MINet into Faster R-CNN, we achieve 80.2% mAP on ImageNet VID dataset.
RESUMO
BACKGROUND: This study aims to retrospectively analyse the safety of ambulatory laparoscopic cholecystectomy (ALC) and identify risk factors for delayed discharge after ALC in the elderly. METHODS: Consecutive patients who were scheduled to undergo ALC were assigned to the elderly group (age ≥ 65 years) or the non-elderly group. The primary outcome was postoperative discharge within 24 h (D24). Secondary outcomes were perioperative mortality, reasons for delayed discharge (psychosocial reasons (DP), complications (DC), drainage (DD) and conversion to open surgery (DCO)), intraoperative data and readmission within 30 days after discharge (readmission). Differences were statistically significant when P < 0.05. RESULTS: There were 7657 patients assigned to the elderly group (n = 1143) or the non-elderly group (n = 6514). The differences between elderly patients and non-elderly patients in the operation time (51.0 (37.0-70.0) versus 50.0 (35.0-65.0) min), blood loss (10.0 (5.0-10.0) versus 5.0 (5.0-10.0) mL), D24 (75.5% versus 81.7%) and DD (7.8% versus 3.2%) were statistically significant (P < 0.05, respectively). The differences between elderly patients and non-elderly patients in DP (8.2% versus 6.7%), DC (7.8% versus 7.9%), DCO (0.7% versus 0.5%) and readmission (0.5% versus 0.4%) were not statistically significant (P > 0.05, respectively). Independent risk factors for delayed discharge after ALC in the elderly were male sex, octogenarian status, prolonged operation time, arrhythmia, type 2 diabetes mellitus, a previous operation in the upper abdomen, acute inflammation of gallbladder and a gallbladder wall thicker than 3 mm (P < 0.05, respectively). CONCLUSION: ALC in the elderly is feasible and safe.
Assuntos
Colecistectomia Laparoscópica , Diabetes Mellitus Tipo 2 , Idoso , Idoso de 80 Anos ou mais , Colecistectomia Laparoscópica/efeitos adversos , Feminino , Vesícula Biliar , Humanos , Masculino , Pessoa de Meia-Idade , Estudos Retrospectivos , Fatores de Tempo , Resultado do TratamentoRESUMO
Extrahepatic portosystemic shunt belongs to a family of rare vascular abnormalities. The clinical importance and manifestations of this vascular abnormality range from asymptomatic cases to liver or metabolic dysfunctions of various degrees. Congenital extrahepatic portosystemic shunt, also termed as Abernethy malformation, is a very rare congenital vascular malformation in which splenomesenteric blood drains into a systemic vein, bypassing the liver through a complete or partial extrahepatic shunt. So far, limited cases of congenital extrahepatic portosystemic shunt have been reported. In this review, incidence, mechanisms, complications, diagnoses and treatments of congenital extrahepatic portosystemic shunt are described.
RESUMO
BACKGROUND: Exposure to high sustained +Gz (head-to-foot inertial load) is known to have harmful effects on pilots' body in flight. Although clinical data have shown that liver dysfunction occurs in pilots, the precise cause has not been well defined. AIM: To investigate rat liver function changes in response to repeated +Gz exposure. METHODS: Ninety male Wistar rats were randomly divided into a blank control group (BC group, n = 30), a +6 Gz/5 min stress group (6GS group, n = 30), and a +10 Gz/5min stress group (10GS group, n = 30). The 6GS and 10GS groups were exposed to +6 Gz and +10 Gz, respectively, in an animal centrifuge. The onset rate of +Gz was 0.5 G/s. The sustained time at peak +Gz was 5 min for each exposure (for 5 exposures, and 5-min intervals between exposures for a total exposure and non-exposure time of 50 min). We assessed liver injury by measuring the portal venous flow volume, serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST), liver tissue malondialdehyde (MDA), Na+-K+-ATPase, and changes in liver histology. These parameters were recorded at 0 h, 6 h, and 24 h after repeated +Gz exposures. RESULTS: After repeated +Gz exposures in the 6GS and the 10GS groups, the velocity and flow signal in the portal vein (PV) were significantly decreased as compared to the BC group at 0 h after exposure. Meanwhile, we found that the PV diameter did not change significantly. However, rats in the 6GS group had a much higher portal venous flow volume than the 10GS group at 0 h after exposure. The 6GS group had significantly lower ALT, AST, and MDA values than the 10GS group 0 h and 6 h post exposure. The Na+-K+-ATPase activity in the 6GS group was significantly higher than that in the 10GS group 0 h and 6 h post exposure. Hepatocyte injury, determined pathologically, was significantly lower in the 6GS group than in the 10GS group. CONCLUSION: Repeated +Gz exposures transiently cause hepatocyte injury and affect liver metabolism and morphological structure.
Assuntos
Aceleração/efeitos adversos , Hipergravidade/efeitos adversos , Hepatopatias/fisiopatologia , Fígado/lesões , Estresse Fisiológico , Medicina Aeroespacial , Animais , Velocidade do Fluxo Sanguíneo , Centrifugação/efeitos adversos , Modelos Animais de Doenças , Hepatócitos/metabolismo , Hepatócitos/patologia , Humanos , Fígado/irrigação sanguínea , Fígado/citologia , Fígado/patologia , Hepatopatias/etiologia , Testes de Função Hepática , Masculino , Pilotos , Veia Porta/fisiopatologia , Ratos , Ratos WistarRESUMO
Primary hepatic neuroendocrine tumor (PHNET) is an extremely rare liver tumor. Patients often have no clinical symptoms or have only non-specific symptoms, such as abdominal pain and abdominal mass. The clinical manifestations, disease development, treatment methods, and treatment outcomes of PHNET vary greatly among cases. Here we report a case of PHNET with a confirmed 26-year survival before surgery. The patient was a 56-year-old female. A large right hepatic mass was detected when the patient was 30 years old. The tumor could not be removed during exploratory laparotomy, and constriction of the right hepatic artery and biopsy were conducted. Pathological results indicated a diagnosis of benign tumor, but a confirmed diagnosis was not reached. Twenty-six years after the patient had been living with the tumor, she sought treatment again because of tumor progression. After systematic evaluation of the resectability, the tumor was resected. Based on the examination results of the gastrointestinal tract and lungs, intraoperative examination results, pathological findings, and long-term follow-up results, the diagnosis of PHNET was confirmed. This case represents the longest reported survival time for a PHNET patient before removal of the tumor.
Assuntos
Neoplasias Hepáticas/patologia , Fígado/patologia , Tumores Neuroendócrinos/patologia , Período Pré-Operatório , Biópsia , China , Feminino , Hepatectomia , Humanos , Fígado/diagnóstico por imagem , Neoplasias Hepáticas/diagnóstico por imagem , Neoplasias Hepáticas/cirurgia , Pessoa de Meia-Idade , Tumores Neuroendócrinos/diagnóstico por imagem , Tumores Neuroendócrinos/cirurgia , Tamanho do Órgão , Fatores de Tempo , Tomografia Computadorizada por Raios XRESUMO
BACKGROUND: Delayed discharge is the existing obstacle to further enhancing quality of recovery after Day-surgery laparoscopic cholecystectomy (LC/DS). This study aims to analyze the reasons for delayed discharge after LC/DS. METHODS: The 745 patients with delayed discharge after LC/DS were retrospectively studied. The reasons for delayed discharge and data related to patients were collected and analyzed. Psychosocial reasons were defined by meeting discharge criteria but refusing to discharge, and complications were defined and graded using the "Clavien-Dindo" classification system. Differences were statistically significant when p-value < 0.01 level. RESULTS: The reasons for delayed discharge included psychosocial reasons (P, n = 324), conversion to open surgery (CO, n = 21) and Clavien-Dindo I (n = 72), II (n = 307), IIIa (n = 17), IVa (n = 4) complications. Group P had a shorter length of postoperative hospital stay (PHT) compared to groups I, II, IIIa, IVa and CO (p < 0.01, respectively). Group II had a longer operation time (p < 0.01) but no longer length of PHT (p = 0.814) compared to group I. The length of PHT in group IIIa was longer than that in groups I and II (p < 0.01, respectively), but the length of PHT in group IVa was no longer than that in groups I, II, and IIIa (p = 0.047, p = 0.044 and p = 0.849, respectively). Group CO had a longer operation time (p < 0.01, respectively), a more blood loss (p < 0.01, respectively) and a longer length of PHT (p < 0.01, respectively) compared to groups P, I, II, and IIIa. CONCLUSION: Patients who are delayed discharge due to psychosocial reasons have a rapid postoperative recovery. The slower postoperative recovery and upgraded complication classifications are related and optimized medical procedures promote the recovery. It is reasonable for patients who undergo conversion to open surgery to experience a slow postoperative recovery.