A5057396396- AI in cancer detection
- Radiomics and Machine Learning in Medical Imaging
- Chaos-based Image/Signal Encryption
- Advanced Steganography and Watermarking Techniques
- Digital Media Forensic Detection
- COVID-19 diagnosis using AI
- Thyroid Cancer Diagnosis and Treatment
- Advanced Neural Network Applications
- Flow Measurement and Analysis
- Cervical Cancer and HPV Research
- Photoacoustic and Ultrasonic Imaging
- Ultrasonics and Acoustic Wave Propagation
- Traditional Chinese Medicine Studies
- Leprosy Research and Treatment
- Cybercrime and Law Enforcement Studies
- Advanced Image Processing Techniques
- Authorship Attribution and Profiling
- Medical Imaging and Analysis
- Video Surveillance and Tracking Methods
- Advanced X-ray and CT Imaging
- Multilevel Inverters and Converters
- Domain Adaptation and Few-Shot Learning
- Artificial Intelligence in Law
- Digital Imaging for Blood Diseases
- Retinal Imaging and Analysis
West China Medical Center of Sichuan University
2021-2024
Sichuan University
2022-2024
West China Hospital of Sichuan University
2022-2024
Beijing Academy of Artificial Intelligence
2023-2024
Shanghai Artificial Intelligence Laboratory
2023-2024
Concordia University
2019
Chengdu Guoke Haibo Information Technology (China)
2015
This study set out to develop a computed tomography (CT)-based wavelet transforming radiomics approach for grading pulmonary lesions caused by COVID-19 and validate it using real-world data.This retrospective analyzed 111 patients with 187 from 16 hospitals; all had confirmed underwent non-contrast chest CT. Data were divided into training cohort (72 127 nine hospitals) an independent test (39 60 seven according the hospital in which CT was performed. In all, 73 texture features extracted...
This study investigates the efficiency of deep learning models in automated diagnosis Hashimoto's thyroiditis (HT) using real-world ultrasound data from examinations by computer-assisted (CAD) with artificial intelligence.We retrospectively collected images patients and without HT 2 hospitals China between September 2008 February 2018. Images were divided into a training set (80%) validation (20%). We ensembled 9 convolutional neural networks (CNNs) as final model (CAD-HT) for...
Ultrasound based estimation of fetal biometry is extensively used to diagnose prenatal abnormalities and monitor growth, for which accurate segmentation the anatomy a crucial prerequisite. Although deep neural network-based models have achieved encouraging results on this task, inevitable distribution shifts in ultrasound images can still result severe performance drop real world deployment scenarios. In article, we propose complete examination system deal with troublesome problem by...
The recent advent of in-context learning (ICL) capabilities in large pre-trained models has yielded significant advancements the generalization segmentation models. By supplying domain-specific image-mask pairs, ICL model can be effectively guided to produce optimal outcomes, eliminating necessity for fine-tuning or interactive prompting. However, current existing ICL-based exhibit limitations when applied medical datasets with substantial diversity. To address this issue, we propose a dual...
This paper proposes a robust watermarking approach based on Discrete Cosine Transform domain that combines Quick Response Code and chaotic system.
This paper proposes a novel fragile watermarking scheme for digital image authentication which is based on Singular Value Decomposition(SVD) and grouped blocks. The watermark bits include two types of are inserted into the least significant bit(LSB) plane host using adaptive chaotic map to determine positions. groped blocks break block-wise independence therefore can withstand Vector Quantization attack(VQ attack). inserting positions related statistical information block data, in order...
This paper proposes a robust watermarking approach based on Discrete Cosine Transform (DCT) domain that combines Quick Response (QR) Code and chaotic system. When embed the watermark, high error correction performance strong decoding capability of QR are utilized to decode text watermark information which improves robustness algorithm. Then image is encrypted with system enhance security this approach. Finally embedded carrier image's DCT blocks after they underwent block-based Arnold...
<title>Abstract</title> Ultrasound imaging is pivotal in clinical diagnostics, providing critical insights into a wide range of diseases and organs. However, advancing artificial intelligence (AI) this field hindered by challenges such as the reliance on large labeled datasets limited generalizability task-specific models, largely due to ultrasound’s low signal-to-noise ratio (SNR). To address these issues, we propose Representation Foundation Model (URFM), designed learn robust...
This paper proposes a novel fragile watermarking scheme for digital image authentication which is based on Singular Value Decomposition(SVD) and grouped blocks. The watermark bits include two types of are inserted into the least significant bit(LSB) plane host using adaptive chaotic map to determine positions. groped blocks break block-wise independence therefore can withstand Vector Quantization attack(VQ attack). inserting positions related statistical information block data, in order...
Masked autoencoder (MAE) has attracted unprecedented attention and achieves remarkable performance in many vision tasks. It reconstructs random masked image patches (known as proxy task) during pretraining learns meaningful semantic representations that can be transferred to downstream However, MAE not been thoroughly explored ultrasound imaging. In this work, we investigate the potential of for recognition. Motivated by unique property imaging high noise-to-signal ratio, propose a novel...