Zhe Li

ORCID: 0000-0003-4716-614X
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Advanced Image and Video Retrieval Techniques
  • Radiomics and Machine Learning in Medical Imaging
  • Multimodal Machine Learning Applications
  • Lung Cancer Diagnosis and Treatment
  • Agronomic Practices and Intercropping Systems
  • Catalytic C–H Functionalization Methods
  • Pharmaceutical and Antibiotic Environmental Impacts
  • Domain Adaptation and Few-Shot Learning
  • Microplastics and Plastic Pollution
  • Oxidative Organic Chemistry Reactions
  • Image Retrieval and Classification Techniques
  • Advanced Radiotherapy Techniques
  • Functional Brain Connectivity Studies
  • Thyroid Cancer Diagnosis and Treatment
  • Heavy metals in environment
  • Machine Learning in Healthcare
  • Peanut Plant Research Studies
  • Head and Neck Anomalies
  • Intracerebral and Subarachnoid Hemorrhage Research
  • Salivary Gland Tumors Diagnosis and Treatment
  • Environmental and Social Impact Assessments
  • Soil Management and Crop Yield
  • Colorectal and Anal Carcinomas
  • Acute Ischemic Stroke Management
  • Agriculture, Soil, Plant Science

Xidian University
2024

University of Science and Technology of China
2008-2024

Shandong Academy of Sciences
2024

Qilu University of Technology
2024

Palo Alto University
2023

Stanford University
2023

Radar (United States)
2023

Air Force Medical University
2016

South China Botanical Garden
1998-2009

Chinese Academy of Sciences
1998-2009

Image-text retrieval is a fundamental task in bridging the semantics between vision and language. The key challenge lies accurately efficiently learning semantic alignment two heterogeneous modalities. Existing image-text approaches can be roughly classified into paradigms. first independent-embedding paradigm to learn global embeddings of modalities, which achieve efficient while failing effectively capture cross-modal fine-grained interaction information images texts. second...

10.1109/tcsvt.2024.3358411 article EN IEEE Transactions on Circuits and Systems for Video Technology 2024-01-25

Abstract Introduction Cognitive deficits caused by heroin‐induced white matter ( WM ) impairments hinder addicts' engagement in and benefit from treatment. The predictive value of integrity heroin addicts during methadone maintenance treatment MMT for future relapse is unclear. Methods Forty‐eight patients were given baseline diffusion tensor imaging scans divided into relapsers HR , 25 cases) abstainers HA 23 according to the results 6‐month follow‐up. Intergroup comparisons performed...

10.1002/brb3.436 article EN cc-by Brain and Behavior 2016-01-24

Image-text matching is a fundamental task to bridge vision and language. The critical challenge lies in accurately learning the semantic similarity between these two heterogeneous modalities. For visual textual features, existing methods typically default static dimensional correspondence mechanism, i.e., using single dimension as measure-unit perform one-to-one correspondence, examine similarity, e.g., cosine/Euclidean distance or weighted similarity. In this paper, different from...

10.1109/tcsvt.2023.3307554 article EN IEEE Transactions on Circuits and Systems for Video Technology 2023-08-22

Carbon−oxygen bond-forming reductive elimination from transient Pd(IV) aryl/acetate complexes was recently implicated as the product release step in Pd(II)-catalyzed arene oxygenation reactions. The mechanistic details of C−O bond formation these intermediates remain elusive and, therefore, are subjected to a systematic theoretical investigation present study. Three proposed mechanisms examined including (A) pre-equilibrium dissociation benzoate ligand followed by resulting five-coordinate...

10.1021/om800067u article EN Organometallics 2008-07-09

Distant metastasis (DM) is the leading cause of death in advanced lung cancer, which diagnosed by positron emission tomography (PET) scanning. Compared with expensive price and nocuous contrast medium PET, using computed (CT) for DM diagnosis more economical convenient clinical practice. However, most existing methods only analyze tumor regions to extract local features prediction, neglects rich whole-lung information. To alleviate this problem, we propose a novel deep learning framework...

10.1109/tetci.2022.3171311 article EN IEEE Transactions on Emerging Topics in Computational Intelligence 2022-05-17

Image-text matching is a fundamental task in bridging the semantics between vision and language. The key challenge lies establishing accurate alignment two heterogeneous modalities. Existing cross-modal fine-grained methods normally include directions, "word to region" "region word", overall image-text similarity calculated from alignments. However, of these directions typically independent, that is, word" irrelevant, so consistency cannot be guaranteed which inevitably introduces...

10.1109/tcsvt.2024.3369656 article EN IEEE Transactions on Circuits and Systems for Video Technology 2024-02-26

Abstract Background In patients with papillary thyroid cancer (PTC), cervical lymph node metastasis (LNM) must be carefully assessed to determine the extent of dissection required and patient prognosis. Few studies attempted whether ultrasound (US) appearance primary tumor could used predict involvement. This study aimed identify US features that LNM in PTC. Methods was a retrospective pathologically confirmed We evaluated following characteristics: lobe, isthmus, size; position; parenchymal...

10.1186/s12885-021-08875-5 article EN cc-by BMC Cancer 2021-11-20

Prognosis of primary pontine hemorrhage (PPH) is important for treatment planning and patient management. However, only few clinical factors were reported to have prognostic value PPH. Here, we propose a deep learning (DL) model that mines high-dimensional information from computed tomography (CT) images combines predicting individualized prognosis We proposed multi-task DL learn CT features hematoma perihematomal areas the risk 30-day mortality, 90-day mortality functional outcome PPH...

10.1016/j.nicl.2022.103257 article EN cc-by NeuroImage Clinical 2022-01-01

Abstract: The authors reveal the functions of Ming Great Wall's archery windows as well wisdom behind their construction. They developed a cross-regional, quantitative research method that identifies heritage values, beginning with data collection by drones, proceeding through processing using artificial intelligence, and concluding analysis via landscape archaeology. respects morphology Wall rugged terrain around it. improved an existing neural network introducing automatic labeling image...

10.1353/lib.2023.a925015 article EN Library trends 2023-02-01

The maize/peanut intercropping is an important cereal/legume system in Liaoning. This can reduce wind erosion, increase yield and resource use efficiency. To clarify the effects of nitrogen rate on nodulation characteristic peanut sole semi-humid area, a micro-plot experiment was conducted at Drought-resistant Cultivation Simulation Test Site, Liaoning, Northeast China. Including two cropping (sole intercropping) seven N rates (0, 50, 75, 100, 150, 200 300 kg/ha) based pattern. results...

10.56028/aetr.12.1.1.2024 article EN Advances in Engineering Technology Research 2024-09-05

<h3>Background</h3> There is a critical unmet need for predictive biomarkers of cancer immunotherapy. The tumor microenvironment (TME) plays an important role in determining immunotherapy response and outcomes. Here, we aimed to develop validate multi-modal deep learning model that integrates routine histopathology radiology images predict TME status gastric patients. <h3>Methods</h3> In this retrospective multi-cohort study, developed multitask the simultaneous prediction disease-free...

10.1136/jitc-2023-sitc2023.1291 article EN cc-by-nc Regular and Young Investigator Award Abstracts 2023-10-31

The prediction of adaptive radiation therapy (ART) prior to (RT) for nasopharyngeal carcinoma (NPC) patients is important reduce toxicity and prolong the survival patients. Currently, due complex tumor micro-environment, a single type high-resolution image can provide only limited information. Meanwhile, traditional softmax-based loss insufficient quantifying discriminative power model. To overcome these challenges, we propose supervised multi-view contrastive learning method with an...

10.48550/arxiv.2210.15201 preprint EN other-oa arXiv (Cornell University) 2022-01-01
Coming Soon ...