Your browser doesn't support javascript.
loading
Mostrar: 20 | 50 | 100
Resultados 1 - 20 de 44
Filtrar
1.
Sci Rep ; 12(1): 7162, 2022 May 03.
Artigo em Inglês | MEDLINE | ID: mdl-35504892

RESUMO

Screening of mRNAs and lncRNAs associated with prognosis and immunity of lung adenocarcinoma (LUAD) and used to construct a prognostic risk scoring model (PRS-model) for LUAD. To analyze the differences in tumor immune microenvironment between distinct risk groups of LUAD based on the model classification. The CMap database was also used to screen potential therapeutic compounds for LUAD based on the differential genes between distinct risk groups. he data from the Cancer Genome Atlas (TCGA) database. We divided the transcriptome data into a mRNA subset and a lncRNA subset, and use multiple methods to extract mRNAs and lncRNAs associated with immunity and prognosis. We further integrated the mRNA and lncRNA subsets and the corresponding clinical information, randomly divided them into training and test set according to the ratio of 5:5. Then, we performed the Cox risk proportional analysis and cross-validation on the training set to construct a LUAD risk scoring model. Based on the risk scoring model, patients were divided into distinct risk group. Moreover, we evaluate the prognostic performance of the model from the aspects of Area Under Curve (AUC) analysis, survival difference analysis, and independent prognostic analysis. We analyzed the differences in the expression of immune cells between the distinct risk groups, and also discuss the connection between immune cells and patient survival. Finally, we screened the potential therapeutic compounds of LUAD in the Connectivity Map (CMap) database based on differential gene expression profiles, and verified the compound activity by cytostatic assays. We extracted 26 mRNAs and 74 lncRNAs related to prognosis and immunity by using different screening methods. Two mRNAs (i.e., KLRC3 and RAET1E) and two lncRNAs (i.e., AL590226.1 and LINC00941) and their risk coefficients were finally used to construct the PRS-model. The risk score positions of the training and test set were 1.01056590 and 1.00925190, respectively. The expression of mRNAs involved in model construction differed significantly between the distinct risk population. The one-year ROC areas on the training and test sets were 0.735 and 0.681. There was a significant difference in the survival rate of the two groups of patients. The PRS-model had independent predictive capabilities in both training and test sets. Among them, in the group with low expression of M1 macrophages and resting NK cells, LUAD patients survived longer. In contrast, the monocyte expression up-regulated group survived longer. In the CMap drug screening, three LUAD therapeutic compounds, such as resveratrol, methotrexate, and phenoxybenzamine, scored the highest. In addition, these compounds had significant inhibitory effects on the LUAD A549 cell lines. The LUAD risk score model constructed using the expression of KLRC3, RAET1E, AL590226.1, LINC00941 and their risk coefficients had a good independent prognostic power. The optimal LUAD therapeutic compounds screened in the CMap database: resveratrol, methotrexate and phenoxybenzamine, all showed significant inhibitory effects on LUAD A549 cell lines.


Assuntos
Adenocarcinoma , Neoplasias Pulmonares , RNA Longo não Codificante , Adenocarcinoma/tratamento farmacológico , Adenocarcinoma/genética , Proteínas de Transporte , Antígenos de Histocompatibilidade Classe I/metabolismo , Humanos , Pulmão/patologia , Neoplasias Pulmonares/tratamento farmacológico , Neoplasias Pulmonares/genética , Neoplasias Pulmonares/metabolismo , Masculino , Proteínas de Membrana/metabolismo , Metotrexato , Fenoxibenzamina , RNA Longo não Codificante/genética , RNA Longo não Codificante/metabolismo , RNA Mensageiro/genética , RNA Mensageiro/metabolismo , Resveratrol , Microambiente Tumoral/genética
2.
Korean J Radiol ; 23(5): 555-565, 2022 May.
Artigo em Inglês | MEDLINE | ID: mdl-35506529

RESUMO

OBJECTIVE: To assess the efficacy and safety of ultrasound (US)-guided radiofrequency ablation (RFA) in patients with primary hyperparathyroidism (PHPT). MATERIALS AND METHODS: This prospective study enrolled 39 participants (14 male, 25 female; mean age, 59.5 ± 15.3 [range, 18-87] years) between September 1, 2018, and January 31, 2021. All participants had parathyroid lesions causing PHPT, proven biochemically and through imaging. The imaging features of the PHPT nodules, including the shape, margin, size, composition, and location, were evaluated before treatment. Serum intact parathyroid hormone, calcium, and phosphorus levels; parathyroid nodule volume; and PHPT-related symptoms were recorded before and after treatment. We calculated the technical success, biochemical cure, and clinical cure rates for these patients. Complications were evaluated during and after the ablation. RESULTS: Complete ablation was achieved in 38 of the 39 nodules in the 39 enrolled participants. All the patients were treated in one session. The technical success rate was 97.4% (38/39). The mean follow-up duration was 13.2 ± 4.6 (range, 6.0-24.9) months. At 6 and 12 months post-RFA, the biochemical cure rates were 82.1% (32/39) and 84.4% (27/32), respectively, and the clinical cure rates were 100% (39/39) and 96.9% (31/32), respectively. Only 2.6% (1/39) of the patients had recurrent PHPT. At 1, 3, 6, and 12 months after technically successful RFA, 44.7% (17/38), 34.3% (12/35), 15.8% (6/38), and 12.5% (4/32) of participants, respectively, had elevated eucalcemic parathyroid hormone levels. Recurrent laryngeal nerve paralysis occurred in 5.1% (2/39) of the patients, who recovered spontaneously within 1-3 months. CONCLUSION: US-guided RFA was effective and safe for PHPT patients. RFA may be an alternative treatment tool for patients who cannot tolerate or refuse to undergo surgery.


Assuntos
Hiperparatireoidismo Primário , Ablação por Radiofrequência , Adolescente , Adulto , Idoso , Idoso de 80 Anos ou mais , Feminino , Humanos , Hiperparatireoidismo Primário/diagnóstico por imagem , Hiperparatireoidismo Primário/cirurgia , Masculino , Pessoa de Meia-Idade , Hormônio Paratireóideo , Estudos Prospectivos , Ablação por Radiofrequência/métodos , Estudos Retrospectivos , Resultado do Tratamento , Ultrassonografia de Intervenção , Adulto Jovem
3.
Natl Sci Rev ; 9(3): nwab192, 2022 Mar.
Artigo em Inglês | MEDLINE | ID: mdl-35382356

RESUMO

Intra-tumor heterogeneity (ITH) is a key challenge in cancer treatment, but previous studies have focused mainly on the genomic alterations without exploring phenotypic (transcriptomic and immune) heterogeneity. Using one of the largest prospective surgical cohorts for hepatocellular carcinoma (HCC) with multi-region sampling, we sequenced whole genomes and paired transcriptomes from 67 HCC patients (331 samples). We found that while genomic ITH was rather constant across stages, phenotypic ITH had a very different trajectory and quickly diversified in stage II patients. Most strikingly, 30% of patients were found to contain more than one transcriptomic subtype within a single tumor. Such phenotypic ITH was found to be much more informative in predicting patient survival than genomic ITH and explains the poor efficacy of single-target systemic therapies in HCC. Taken together, we not only revealed an unprecedentedly dynamic landscape of phenotypic heterogeneity in HCC, but also highlighted the importance of studying phenotypic evolution across cancer types.

4.
World J Clin Cases ; 10(4): 1320-1325, 2022 Feb 06.
Artigo em Inglês | MEDLINE | ID: mdl-35211565

RESUMO

BACKGROUND: The radial nerve (RN) splits into two main branches at the elbow: The superficial branch of RN (SBRN) and the deep branch of RN. The SBRN can be easily damaged in acute trauma due to its superficial feature. CASE SUMMARY: A 55-year-old male patient injured his right wrist 10 mo ago. Debridement, suturing and bandaging were performed in the emergency room. Six months after the scar had healed, he felt numbness and tingling in the dorsal surface of the thumb of the right hand. So the surgery of resection and SBRN anastomosis were performed. The pathological findings showed it as traumatic neuroma. Four months after surgery, the patient felt numbness and tingling in the right dorsal surface of the thumb again. The tenderness was marked in the operated area. Ultrasound indicated that the SBRN was adhered to the surrounding tissue. The patient refused further surgical treatment and underwent ultrasound-guided needle release plus corticosteroid injection of the SBRN. Four weeks later, the tenderness in the surgical area was reduced by 70%, the numbness in the dorsal surface of the thumb of the right hand was reduced by 40% and the nerve swelling evaluated by ultrasound was reduced. Four months passed, he did not feel any numbness or tingling sensation of his right wrist. This is the first report of ultrasound-guided needle release plus corticosteroid injection of the SBRN. CONCLUSION: Ultrasound can evaluate the condition of the RN, and the relationship with surrounding tissues. Ultrasound-guided needle release plus corticosteroid injection is an effective and safe treatment for SBRN adhesion.

5.
IEEE Trans Cybern ; 52(2): 1233-1246, 2022 Feb.
Artigo em Inglês | MEDLINE | ID: mdl-32559172

RESUMO

Deep learning methods are becoming the de-facto standard for generic visual recognition in the literature. However, their adaptations to industrial scenarios, such as visual recognition for machines, product streamlines, etc., which consist of countless components, have not been investigated well yet. Compared with the generic object detection, there is some strong structural knowledge in these scenarios (e.g., fixed relative positions of components, component relationships, etc.). A case worth exploring could be automated visual inspection for trains, where there are various correlated components. However, the dominant object detection paradigm is limited by treating the visual features of each object region separately without considering common sense knowledge among objects. In this article, we propose a novel automated visual inspection framework for trains exploring structural knowledge for train component detection, which is called SKTCD. SKTCD is an end-to-end trainable framework, in which the visual features of train components and structural knowledge (including hierarchical scene contexts and spatial-aware component relationships) are jointly exploited for train component detection. We propose novel residual multiple gated recurrent units (Res-MGRUs) that can optimally fuse the visual features of train components and messages from the structural knowledge in a weighted-recurrent way. In order to verify the feasibility of SKTCD, a dataset that contains high-resolution images captured from moving trains has been collected, in which 18 590 critical train components are manually annotated. Extensive experiments on this dataset and on the PASCAL VOC dataset have demonstrated that SKTCD outperforms the existing challenging baselines significantly. The dataset as well as the source code can be downloaded online (https://github.com/smartprobe/SKCD).


Assuntos
Software
6.
Curr Med Chem ; 29(8): 1369-1378, 2022 Mar 04.
Artigo em Inglês | MEDLINE | ID: mdl-34238143

RESUMO

This review describes how phase-changeable nanoparticles enable highly-efficient high-intensity focused ultrasound ablation (HIFU). HIFU is effective in the clinical treatment of solid malignant tumors; however, it has intrinsic disadvantages for treating some deep lesions, such as damage to surrounding normal tissues. When phase-changeable nanoparticles are used in HIFU treatment, they could serve as good synergistic agents because they are transported in the blood and permeated and accumulated effectively in tissues. HIFU's thermal effects can trigger nanoparticles to undergo a special phase transition, thus enhancing HIFU ablation efficiency. Nanoparticles can also carry anticancer agents and release them in the targeted area to achieve chemo-synergistic therapy response. Although the formation of nanoparticles is complicated and HIFU applications are still in an early stage, the potential for their use in synergy with HIFU treatment shows promising results.


Assuntos
Antineoplásicos , Ablação por Ultrassom Focalizado de Alta Intensidade , Nanopartículas , Neoplasias , Antineoplásicos/uso terapêutico , Ablação por Ultrassom Focalizado de Alta Intensidade/métodos , Humanos , Neoplasias/tratamento farmacológico
7.
Front Chem ; 10: 881812, 2022.
Artigo em Inglês | MEDLINE | ID: mdl-35372260

RESUMO

Cancer is a serious health problem which increasingly causes morbidity and mortality worldwide. It causes abnormal and uncontrolled cell division. Traditional cancer treatments include surgery, chemotherapy, radiotherapy and so on. These traditional therapies suffer from high toxicity and arouse safety concern in normal area and have difficulty in accurately targeting tumour. Recently, a variety of nanomaterials could be used for cancer diagnosis and therapy. Nanomaterials have several advantages, e.g., high concentration in tumour via targeting design, reduced toxicity in normal area and controlled drug release after various rational designs. They can combine with many types of biomaterials in order to improve biocompatibility. In this review, we outlined the latest research on the use of bioresponsive nanomaterials for various cancer imaging modalities (magnetic resonance imaging, positron emission tomography and phototacoustic imaging) and imaging-guided therapy means (chemotherapy, radiotherapy, photothermal therapy and photodynamic therapy), followed by discussing the challenges and future perspectives of this bioresponsive nanomaterials in biomedicine.

8.
Int J Hyperthermia ; 39(1): 490-496, 2022.
Artigo em Inglês | MEDLINE | ID: mdl-35285391

RESUMO

OBJECTIVE: To investigate the efficacy of radiofrequency ablation (RFA) as a treatment option for primary hyperparathyroidism (pHPT) and risk factors for postablative eucalcemic parathyroid hormone elevation (ePTH). METHODS: This retrospective study included 51 patients with pHPT who underwent RFA. The patients were divided into the ePTH and normal PTH groups, based on the serum intact parathyroid hormone (iPTH) level one month after ablation. Serum iPTH, calcium, and phosphorus levels, and the volume reduction rates (VRR) of the parathyroid glands were compared between the groups at each follow-up point. Risk factors for ePTH at one month after ablation were examined. RESULTS: After RFA, one (2%) patient had persistent pHPT, and 50 (98%) patients were cured. The incidence rates of ePTH at 1, 3, 6, and 12 months were 48%, 30%, 20%, and 16%, respectively. Serum iPTH levels in the ePTH group were higher than those in the normal PTH group at each follow-up point (all p < 0.05), except 1 day after ablation (p > 0.05). Serum calcium and phosphorus levels, and the VRR of the glands were comparable in both groups at each follow-up point (all p > 0.05), except for calcium levels 3 days after RFA (p < 0.05). Baseline iPTH (odds ratio, 1.067; p = 0.045) and calcium (odds ratio, 3.923; p = 0.038) levels were independent risk factors for ePTH 1 month after RFA. CONCLUSIONS: RFA is safe and effective for the treatment of pHPT. Moreover, ePTH occurrence after RFA was associated with baseline iPTH and calcium levels and did not increase the risk of recurrent pHPT.


Assuntos
Hiperparatireoidismo Primário , Ablação por Radiofrequência , Cálcio , Humanos , Hiperparatireoidismo Primário/cirurgia , Hormônio Paratireóideo , Paratireoidectomia , Ablação por Radiofrequência/efeitos adversos , Estudos Retrospectivos , Fatores de Risco
9.
Front Bioeng Biotechnol ; 9: 784602, 2021.
Artigo em Inglês | MEDLINE | ID: mdl-34869294

RESUMO

Mitochondria are the primary organelles which can produce adenosine triphosphate (ATP). They play vital roles in maintaining normal functions. They also regulated apoptotic pathways of cancer cells. Given that, designing therapeutic agents that precisely target mitochondria is of great importance for cancer treatment. Nanocarriers can combine the mitochondria with other therapeutic modalities in cancer treatment, thus showing great potential to cancer therapy in the past few years. Herein, we summarized lipophilic cation- and peptide-based nanosystems for mitochondria targeting. This review described how mitochondria-targeted nanocarriers promoted highly efficient cancer treatment in photodynamic therapy (PDT), chemotherapy, combined immunotherapy, and sonodynamic therapy (SDT). We further discussed mitochondria-targeted nanocarriers' major challenges and future prospects in clinical cancer treatment.

10.
Curr Med Imaging ; 2021 Dec 08.
Artigo em Inglês | MEDLINE | ID: mdl-34879810

RESUMO

BACKGROUND: Mycosis fungoides (MF) is the most common form of cutaneous T-cell lymphoma with many clinicopathological variants, thus difficult to diagnose in its early stages. CASE PRESENTATION: This case report is about a 76 years old Chinese woman presented with 2 years history of erythematous plaque on the lateral right thigh, after combining clinical manifestations with results of pathological examinations, it is consistent with the diagnosis of MF. DISCUSSION: Mycosis fungoides (MF) is the most common form of cutaneous T-cell lymphoma. Patient in this case had a long course of disease and repeated attacks. Ultrasound shows small patch of liquid dark area of the lesion. Color Doppler image shows rich blood flow which just looks like lacustrine. Thick and nourishing blood vessels could be seen in the depth. CONCLUSION: Our case report using ultrasound to observe MF and demonstrate that ultrasound is helpful in diagnosing and evaluating effectiveness in treating MF.

11.
Annu Int Conf IEEE Eng Med Biol Soc ; 2021: 2115-2118, 2021 11.
Artigo em Inglês | MEDLINE | ID: mdl-34891706

RESUMO

Diabetic retinopathy (DR) is one of the most common eye conditions among diabetic patients. However, vision loss occurs primarily in the late stages of DR, and the symptoms of visual impairment, ranging from mild to severe, can vary greatly, adding to the burden of diagnosis and treatment in clinical practice. Deep learning methods based on retinal images have achieved remarkable success in automatic DR grading, but most of them neglect that the presence of diabetes usually affects both eyes, and ophthalmologists usually compare both eyes concurrently for DR diagnosis, leaving correlations between left and right eyes unexploited. In this study, simulating the diagnostic process, we propose a two-stream binocular network to capture the subtle correlations between left and right eyes, in which, paired images of eyes are fed into two identical subnetworks separately during training. We design a contrastive grading loss to learn binocular correlation for five-class DR detection, which maximizes inter-class dissimilarity while minimizing the intra-class difference. Experimental results on the EyePACS dataset show the superiority of the proposed binocular model, outperforming monocular methods by a large margin.Clinical relevance- Compared to conventional DR grading methods based on monocular images, our approach can provide more accurate predictions and extract graphical patterns from retinal images of both eyes for clinical reference.


Assuntos
Diabetes Mellitus , Retinopatia Diabética , Retinopatia Diabética/diagnóstico , Fundo de Olho , Humanos
12.
Annu Int Conf IEEE Eng Med Biol Soc ; 2021: 3395-3398, 2021 11.
Artigo em Inglês | MEDLINE | ID: mdl-34891968

RESUMO

Deep learning has achieved promising segmentation performance on 3D left atrium MR images. However, annotations for segmentation tasks are expensive, costly and difficult to obtain. In this paper, we introduce a novel hierarchical consistency regularized mean teacher framework for 3D left atrium segmentation. In each iteration, the student model is optimized by multi-scale deep supervision and hierarchical consistency regularization, concurrently. Extensive experiments have shown that our method achieves competitive performance as compared with full annotation, outperforming other state-of-the-art semi-supervised segmentation methods.


Assuntos
Átrios do Coração , Aprendizado de Máquina Supervisionado , Átrios do Coração/diagnóstico por imagem , Humanos , Imageamento Tridimensional , Estudantes
13.
Annu Int Conf IEEE Eng Med Biol Soc ; 2021: 3582-3585, 2021 11.
Artigo em Inglês | MEDLINE | ID: mdl-34892013

RESUMO

Accurate automatic liver and tumor segmentation plays a vital role in treatment planning and disease monitoring. Recently, deep convolutional neural network (DCNNs) has obtained tremendous success in 2D and 3D medical image segmentation. However, 2D DCNNs cannot fully leverage the inter-slice information, while 3D DCNNs are computationally expensive and memory intensive. To address these issues, we first propose a novel dense-sparse training flow from a data perspective, in which, densely adjacent slices and sparsely adjacent slices are extracted as inputs for regularizing DCNNs, thereby improving the model performance. Moreover, we design a 2.5D light-weight nnU-Net from a network perspective, in which, depthwise separable convolutions are adopted to improve the efficiency. Extensive experiments on the LiTS dataset have demonstrated the superiority of the proposed method.Clinical relevance- The proposed method can effectively segment livers and tumors from CT scans with low complexity, which can be easily implemented into clinical practice.


Assuntos
Processamento de Imagem Assistida por Computador , Neoplasias , Abdome , Humanos , Fígado/diagnóstico por imagem , Redes Neurais de Computação
14.
IEEE Trans Cybern ; PP2021 Dec 17.
Artigo em Inglês | MEDLINE | ID: mdl-34919527

RESUMO

Traffic prediction based on massive speed data collected from traffic sensors plays an important role in traffic management. However, it is still challenging to obtain satisfactory performance due to the complex and dynamic spatial-temporal correlations among the data. Recently, many research works have demonstrated the effectiveness of graph neural networks (GNNs) for spatial-temporal modeling. However, such models are restricted by conditional distribution during training, and may not perform well when the target is outside the primary region of interest in the distribution. In this article, we address this problem with a stagewise learning mechanism, in which we redefine speed prediction as a conditional distribution learning followed by speed regression. We first perform a conditional distribution learning for each observed speed class, and then obtain speed prediction by optimizing regression learning, based on the learned conditional distribution. To effectively learn the conditional distribution, we introduce a mean-residue loss, consisting of two parts: 1) a mean loss, which penalizes the differences between the mean of the estimated conditional distribution and the ground truth and 2) a residue loss, which penalizes residue errors of the long tails in the distribution. To optimize the subsequent regression based on distribution information, we combine the mean absolute error (MAE) as another part of the loss function. We also incorporate a GNN-based architecture with our proposed learning mechanism. Mean-residue loss is employed to supervise the hidden speed representation in the network at each time interval, followed by a shared layer to recalibrate the hidden temporal dependencies in the conditional distribution. The experimental results based on three public traffic datasets have demonstrated that the effectiveness of the proposed method outperforms state-of-the-art methods.

16.
SN Comput Sci ; 2(5): 394, 2021.
Artigo em Inglês | MEDLINE | ID: mdl-34341778

RESUMO

There is no doubt that the COVID-19 epidemic posed the most significant challenge to all governments globally since January 2020. People have to readapt after the epidemic to daily life with the absence of an effective vaccine for a long time. The epidemic has led to society division and uncertainty. With such issues, governments have to take efficient procedures to fight the epidemic. In this paper, we analyze and discuss two official news agencies' tweets of Iran and Turkey by using sentiment- and semantic analysis-based unsupervised learning approaches. The main topics, sentiments, and emotions that accompanied the agencies' tweets are identified and compared. The results are analyzed from the perspective of psychology, sociology, and communication.

18.
Artigo em Inglês | MEDLINE | ID: mdl-33788693

RESUMO

Recently, deep learning-based approaches have achieved superior performance on object detection applications. However, object detection for industrial scenarios, where the objects may also have some structures and the structured patterns are normally presented in a hierarchical way, is not well investigated yet. In this work, we propose a novel deep learning-based method, hierarchical graphical reasoning (HGR), which utilizes the hierarchical structures of trains for train component detection. HGR contains multiple graphical reasoning branches, each of which is utilized to conduct graphical reasoning for one cluster of train components based on their sizes. In each branch, the visual appearances and structures of train components are considered jointly with our proposed novel densely connected dual-gated recurrent units (Dense-DGRUs). To the best of our knowledge, HGR is the first kind of framework that explores hierarchical structures among objects for object detection. We have collected a data set of 1130 images captured from moving trains, in which 17 334 train components are manually annotated with bounding boxes. Based on this data set, we carry out extensive experiments that have demonstrated our proposed HGR outperforms the existing state-of-the-art baselines significantly. The data set and the source code can be downloaded online at https://github.com/ChengZY/HGR.

19.
IEEE J Biomed Health Inform ; 25(10): 3744-3751, 2021 10.
Artigo em Inglês | MEDLINE | ID: mdl-33460386

RESUMO

Image segmentation is one of the most essential biomedical image processing problems for different imaging modalities, including microscopy and X-ray in the Internet-of-Medical-Things (IoMT) domain. However, annotating biomedical images is knowledge-driven, time-consuming, and labor-intensive, making it difficult to obtain abundant labels with limited costs. Active learning strategies come into ease the burden of human annotation, which queries only a subset of training data for annotation. Despite receiving attention, most of active learning methods still require huge computational costs and utilize unlabeled data inefficiently. They also tend to ignore the intermediate knowledge within networks. In this work, we propose a deep active semi-supervised learning framework, DSAL, combining active learning and semi-supervised learning strategies. In DSAL, a new criterion based on deep supervision mechanism is proposed to select informative samples with high uncertainties and low uncertainties for strong labelers and weak labelers respectively. The internal criterion leverages the disagreement of intermediate features within the deep learning network for active sample selection, which subsequently reduces the computational costs. We use the proposed criteria to select samples for strong and weak labelers to produce oracle labels and pseudo labels simultaneously at each active learning iteration in an ensemble learning manner, which can be examined with IoMT Platform. Extensive experiments on multiple medical image datasets demonstrate the superiority of the proposed method over state-of-the-art active learning methods.


Assuntos
Aprendizado de Máquina Supervisionado , Humanos , Processamento de Imagem Assistida por Computador , Isoquinolinas
20.
IEEE Trans Pattern Anal Mach Intell ; 43(4): 1460-1466, 2021 04.
Artigo em Inglês | MEDLINE | ID: mdl-32142419

RESUMO

Is recurrent network really necessary for learning a good visual representation for video based person re-identification (VPRe-id)? In this paper, we first show that the common practice of employing recurrent neural networks (RNNs) to aggregate temporal-spatial features may not be optimal. Specifically, with a diagnostic analysis, we show that the recurrent structure may not be effective learn temporal dependencies than what we expected and implicitly yields an orderless representation. Based on this observation, we then present a simple yet surprisingly powerful approach for VPRe-id, where we treat VPRe-id as an efficient orderless ensemble of image based person re-identification problem. More specifically, we divide videos into individual images and re-identify person with ensemble of image based rankers. Under the i.i.d. assumption, we provide an error bound that sheds light upon how could we improve VPRe-id. Our work also presents a promising way to bridge the gap between video and image based person re-identification. Comprehensive experimental evaluations demonstrate that the proposed solution achieves state-of-the-art performances on multiple widely used datasets (iLIDS-VID, PRID 2011, and MARS).

SELEÇÃO DE REFERÊNCIAS
DETALHE DA PESQUISA
...