A5053336034- AI in cancer detection
- Medical Image Segmentation Techniques
- Retinal Imaging and Analysis
- Advanced Neural Network Applications
- Domain Adaptation and Few-Shot Learning
- Radiomics and Machine Learning in Medical Imaging
- COVID-19 diagnosis using AI
- Machine Learning and ELM
- Optical Coherence Tomography Applications
- Digital Imaging for Blood Diseases
- Retinal and Macular Surgery
- Multimodal Machine Learning Applications
- Image Retrieval and Classification Techniques
- Advanced Image and Video Retrieval Techniques
- Advanced MRI Techniques and Applications
- NMR spectroscopy and applications
- Skin Protection and Aging
- Spine and Intervertebral Disc Pathology
- Nuclear Physics and Applications
- Advanced X-ray Imaging Techniques
- Immunotherapy and Immune Responses
- Brain Tumor Detection and Classification
- Cutaneous Melanoma Detection and Management
- Medical Imaging and Analysis
- Hip disorders and treatments
Xi'an Jiaotong University
2024
Hangzhou Dianzi University
2021-2024
Xinqiao Hospital
2023-2024
Army Medical University
2023-2024
University of South China
2020-2021
Affiliated Hospital of Jining Medical University
2019
Nuclei instance segmentation within microscopy images is a fundamental task in the pathology work-flow, based on that meaningful nuclear features can be extracted and multiple biological related analysis performed. However, this still challenging because of large variability among different types nuclei. Although deep learning(DL) methods have achieved state-of-the-art results nuclei tasks, these are usually focus improving accuracy require support powerful computing resources. In paper, we...
The acetabular bone defect reconstruction is of great challenge in total hip arthroplasty (THA). Although several solutions such as autologous grafting, trabecular metal augment, or compromising techniques the medial protrusion, high inclination angle, and elevated center have been raised, their efficacy reliability not fully substantiated. Traditional methods may lead to resorption, aggravation defects, unequal length lower limbs, unbalance hip-spine relationship, increased costs, so on....
Abstract Purpose Segmentation of orbital tumors in CT images is great significance for tumor diagnosis, which one the most prevalent diseases eye. However, large variety sizes and shapes makes segmentation task very challenging, especially when available annotation data limited. Methods To this end, paper, we propose a multi‐scale consistent self‐training network (MSCINet) semi‐supervised segmentation. Specifically, exploit semantic‐invariance features by enforcing consistency between...
To explore the efficacy and safety of bortezomib or thalidomide combined with recombinant human erythropoietin (rhEPO) in treatment multiple myeloma (MM).
Domain generalization (DG) for medical image segmentation due to privacy preservation prefers learning from a single-source domain and expects good robustness on unseen target domains. To achieve this goal, previous methods mainly use data augmentation expand the distribution of samples learn invariant content them. However, most these commonly perform global augmentation, leading limited augmented sample diversity. In addition, style is more scattered than source domain, which may cause...
Automated detection and segmentation of every nucleus in microscopy images is a fundamental task biomedical research clinical practice, including nuclear morphology analysis, cancer grading prognosis predicting. However, this still challenging because the atypical shape, size internal organisation. In paper, we propose nuclei instance method with aim jointing simultaneously. The builds on two-stream Convolutional Neural Network(CNN) architecture that explicit utilize single shot multi-box...
Automatic segmentation of fluid in Optical Coherence Tomography (OCT) images is beneficial for ophthalmologists to make an accurate diagnosis. Although semi-supervised OCT networks enhance their performance by introducing additional unlabeled data, the enhancement limited. To address this, we propose Superpixel and Confident Learning Guide Point Annotations Network (SCLGPA-Net) based on teacher-student architecture, which can learn from limited fully-annotated data abundant point-annotated...
The shape and location of the orbital tumor are precious for evaluation, diagnosis, treatment, so it is essential to accurately segment assist clinicians in making reasonable decisions. Nowadays, encoder-decoder-based convolutional neural networks have been widely used medical image segmentation tasks. However, commonly single tensor flow architecture cannot achieve satisfactory performance with significant size variations. In this paper, we propose a size-sensitive deep network named...
Medical image segmentation annotations exhibit variations among experts due to the ambiguous boundaries of segmented objects and backgrounds in medical images. Although using multiple for each fully-supervised has been extensively studied training deep models, obtaining a large amount multi-annotated data is challenging substantial time manpower costs required annotations, resulting most images lacking any annotations. To address this, we propose Multi-annotated Semi-supervised Ensemble...
Retinal disease is one of the primary causes visual impairment, and early diagnosis essential for preventing further deterioration. Nowadays, many works have explored Transformers diagnosing diseases due to their strong representation capabilities. However, retinal exhibit milder forms often present with overlapping signs, which pose great difficulties accurate multi-class classification. Therefore, we propose a new framework named Multi-Scale Patch Message Passing Swin Transformer...
Optic disc and cup segmentation plays a crucial role in automating the screening diagnosis of optic glaucoma. While data-driven convolutional neural networks (CNNs) show promise this area, inherent ambiguity segmenting objects background boundaries task leads to noisy annotations that impact model performance. To address this, we propose an innovative label-denoising method Multiple Pseudo-labels Noise-aware Network (MPNN) for accurate segmentation. Specifically, Generation Guided Denoising...
Medical image segmentation aims to partition a medical into several segments, with each segment corresponding distinct object or anatomical structure of interest, which is crucial for applications such as conducting quantitative analyses and planning treatment procedures. As more multi-site data are acquired, many learning models have been developed by applying model trained on multiple source domains unseen target domains. To address cross-domain heterogeneity, domain generalization...
摘要： 【摘要】 目的 探讨进食藜导致的植物日光性皮炎的临床特点和治疗。方法 对2013—2017年在济宁医学院附属医院皮肤科诊治的11例进食藜引起的植物日光性皮炎患者的一般资料、临床表现、实验室检查、治疗及预后进行回顾性分析。结果 11例患者均为女性，年龄45 ~ 62岁，起病前1 2 d内均有进食藜史和日光暴露史。临床均表现为颜面部、双手背、前臂暴露处对称分布的非凹陷性肿胀性红斑，表面紧张光亮，伴痛痒。6例患者血白细胞计数升高，1例嗜酸性粒细胞计数升高，1例尿蛋白阳性，1例隐血阳性。均系统给予甲泼尼龙、氯雷他定、依巴斯汀、螺内酯、呋塞米、奥美拉唑治疗，2例同时应用人免疫球蛋白，3例口服洛芬待因，并予外用药物对症治疗。10例治疗7 10 d后好转出院，1年后随访均无色素沉着及瘢痕；1例出现双手背皮肤坏死，转入整形科继续治疗，半年后随访告知遗留皮肤瘢痕。结论 进食藜导致的植物日光性皮炎常表现为暴露部位肿胀性红斑，明确诊断后避免进食藜和日晒，并及时给予抗过敏治疗可有效控制病情。