A5065448472- AI in cancer detection
- Radiomics and Machine Learning in Medical Imaging
- COVID-19 diagnosis using AI
- Advanced Data Compression Techniques
- Advanced Neural Network Applications
- Colorectal Cancer Screening and Detection
- Medical Image Segmentation Techniques
- Medical Imaging and Analysis
- Retinal Imaging and Analysis
- Image Retrieval and Classification Techniques
- Image and Object Detection Techniques
- Domain Adaptation and Few-Shot Learning
- Digital Imaging for Blood Diseases
- Image Processing Techniques and Applications
- Spectroscopy Techniques in Biomedical and Chemical Research
- Identification and Quantification in Food
- Vehicle License Plate Recognition
- Machine Learning in Healthcare
- Spectroscopy and Chemometric Analyses
- Cutaneous Melanoma Detection and Management
- Brain Tumor Detection and Classification
- Artificial Intelligence in Healthcare
- Artificial Intelligence in Healthcare and Education
- Material Properties and Processing
North Sichuan Medical University
2024
Tianjin University
2023-2024
Deep polyp segmentation methods have shown remarkable potential in boosting diagnostic efficiency. Nevertheless, these rely on sufficient pixel-wise annotated data, which is time-consuming and labor-intensive to acquire clinical practice. This challenge further escalated under the scenario due massive video frames. To alleviate annotating burden, this letter, we propose a label-efficient framework named HybridVPS, drastically reduces annotation cost while maintaining satisfactory...
Image segmentation plays an important role in vision understanding. Recently, the emerging foundation models continuously achieved superior performance on various tasks. Following such success, this paper, we prove that Segment Anything Model 2 (SAM2) can be a strong encoder for U-shaped models. We propose simple but effective framework, termed SAM2-UNet, versatile image segmentation. Specifically, SAM2-UNet adopts Hiera backbone of SAM2 as encoder, while decoder uses classic design....
Automatic polyp segmentation is critical for early prevention and diagnosis of colorectal cancer. However, diverse foreground appearance complicated background interference severely degrade the performance pixel-level prediction. The excessive computational overheads further hinder practical clinical applications existing methods. In this paper, we propose a novel Lesion-Aware Contextual Interaction Network (LACINet), which aims to explore long-range dependencies global contexts with...
Abstract The realm of clinical medicine stands on the brink a revolutionary breakthrough as large language models (LLMs) emerge formidable allies, propelled by prowess deep learning and wealth data. Yet, amidst disquieting specter misdiagnoses haunting halls medical treatment, LLMs offer glimmer hope, poised to reshape landscape. However, their mettle acumen, particularly in crucible real-world professional scenarios replete with intricate logical interconnections, re-main shrouded...
In medical image analysis, multi-organ semi-supervised segmentation faces challenges such as insufficient labels and low contrast in soft tissues. To address these issues, existing studies typically employ techniques using pseudo-labeling consistency regularization. However, methods mainly rely on individual data samples for training, ignoring the rich neighborhood information present feature space. this work, we argue that supervisory can be directly extracted from geometry of Inspired by...
Accurate identification of breast masses is crucial in diagnosing cancer; however, it can be challenging due to their small size and being camouflaged surrounding normal glands. Worse still, also expensive clinical practice obtain adequate pixel-wise annotations for training deep neural networks. To overcome these two difficulties with one stone, we propose a semi- weakly-supervised learning framework mass segmentation that utilizes limited strongly-labeled samples sufficient weakly-labeled...
Deep learning-based diagnostic systems have demonstrated potential in skin disease diagnosis. However, their performance can easily degrade on test domains due to distribution shifts caused by input-level corruptions, such as imaging equipment variability, brightness changes, and image blur. This will reduce the reliability of model deployment real-world scenarios. Most existing solutions focus adapting source through retraining different target domains. Although effective, this process is...
Recent advances in large foundation models, such as the Segment Anything Model (SAM), have demonstrated considerable promise across various tasks. Despite their progress, these models still encounter challenges specialized medical image analysis, especially recognizing subtle inter-class differences Diabetic Retinopathy (DR) lesion segmentation. In this paper, we propose a novel framework that customizes SAM for text-prompted DR segmentation, termed TP-DRSeg. Our core idea involves...
The data obtained from Raman spectroscopy is characterized by high dimension and complexity. In this study, dimensional reduction techniques, including LASSO, random forest, principal component analysis, were employed to refine the datasets. An optimal model was then constructed using Artificial Neural Networks, which achieved an average accuracy of 0.943 when subjected 10 times 5-fold cross-validation. Based on feature importance ranking map four algorithms, species differences five...