- Advanced SAR Imaging Techniques
- Radiomics and Machine Learning in Medical Imaging
- Synthetic Aperture Radar (SAR) Applications and Techniques
- AI in cancer detection
- Obstructive Sleep Apnea Research
- Medical Image Segmentation Techniques
- Remote-Sensing Image Classification
- Non-Invasive Vital Sign Monitoring
- Cervical Cancer and HPV Research
- Radar Systems and Signal Processing
- Lung Cancer Diagnosis and Treatment
- Spectroscopy and Chemometric Analyses
- COVID-19 diagnosis using AI
- Infrared Target Detection Methodologies
- Medical Imaging and Analysis
- Advanced Image Fusion Techniques
- Image and Signal Denoising Methods
- Advanced Chemical Sensor Technologies
- Geophysical Methods and Applications
- Topic Modeling
- Spectroscopy Techniques in Biomedical and Chemical Research
- Image Processing Techniques and Applications
- Brain Tumor Detection and Classification
- Digital Imaging for Blood Diseases
- Minerals Flotation and Separation Techniques
Tianjin University
2015-2025
State Key Laboratory of Chemical Engineering
2025
Institute of Electronics
2011-2024
Northeast Agricultural University
2024
Tianjin Medical University
2021-2024
Second Military Medical University
2024
University of Glasgow
2024
Eastern Hepatobiliary Surgery Hospital
2024
Shandong Normal University
2023
Beihang University
2022
Recently, deep neural networks such as RNN, CNN and Transformer have been applied in the task of sequential recommendation, which aims to capture dynamic preference characteristics from logged user behavior data for accurate recommendation. However, online platforms, is inevitable contain noise, recommendation models are easy overfit on these data. To tackle this problem, we borrow idea filtering algorithms signal processing that attenuates noise frequency domain. In our empirical...
Accurate segmentation and recognition algorithm of lung nodules has great important value reference for early diagnosis cancer. An is proposed 3D CT sequence images in this paper based on Res U-Net network ResNet50 classification network. The common convolutional layers encoding decoding paths are replaced by residual units while the loss function changed to Dice after using cross entropy accelerate convergence. Since small rich information, improved replacing 2D with reducing sizes some...
Recent advances in the Medical Internet of Things (MIoT) and big data enable a prospering platform for pervasive healthcare facilitate transformation from hospital-centered to human-centered healthcare. Wearable devices as human interfaces provide first-hand real-time monitoring, which are key technologies MIoT. Several remarkable surveys have been conducted summarize recent progress wearable sensors systems MIoT medicine. However, few focused on optical sensing (WOS) technologies, is an...
Abstract Coronary vessel segmentation plays a pivotal role in automating the auxiliary diagnosis of coronary heart disease. The continuity and boundary accuracy segmented vessels directly affect subsequent processing. Notably, during segmentation, with severe stenosis can easily cause errors breakage, resulting isolated islands. To address these issues, we propose novel multi-scale U-shaped transformer aggregation topology preservation (UT-BTNet) for angiography. Specifically, considering...
Continuous respiratory monitoring is extensively important in clinical applications. To effectively assess respiration rate (RR), tidal volume (TV), and minute ventilation (MV), we propose experimentally demonstrate a system using an in-line few-mode fiber Mach-Zehnder interferometer (FMF-MZI), which the first to introduce MZI into optimal wearable design for monitoring. The linear region of proposed sensor analyzed positioned by flexible arch structure with curvature sensitivity up 8.53...
Colposcopy is an important method in the diagnosis of cervical lesions. However, experienced colposcopists are lacking at present, and training cycle long. Therefore, artificial intelligence-based colposcopy-assisted examination has great prospects. In this paper, a lesion segmentation model (CLS-Model) was proposed for region from colposcopic post-acetic-acid images accurate results could provide good foundation further research on classification selection biopsy site.First, improved Faster...
Convolutional neural network (CNN)-based methods have become the mainstream in radar ship recognition. However, these suffer from two common problems. First, training samples consist largely of types, giving them an overwhelming numerical advantage over rare types. As a result, CNN-based recognition algorithms fail to classify types correctly. Second, huge high-resolution slices result heavy computational burdens. To solve first problem, namely, class imbalance this letter proposes CNN...
In this study, the effects of polyethylene glycol (PEG) and polyvinylpyrrolidone (PVP) on nucleation crystal growth kinetics acephate were systematically investigated. The differential PVP PEG with varying molecular weights primary secondary uncovered. addition, effect additives rate was also measured, which uncovered that micromolar concentrations (4.25 × 10–5 M) at low supersaturation can substantially promote growth. Moreover, a cooling crystallization experiment designed to further...
In this paper, we propose an approach to improve the sensitivity of optical fiber surface plasmon resonance (SPR) sensor with a pure higher-order mode excited by designed selective coupler (MSC). We calculate proportion power in cladding. Compared LP01 mode, LP11 (LP21 mode) cladding theoretically improves 100% (150%). To generate relatively or LP21 (MSC, 430–580 nm) is designed. The coupling efficiency LP01–LP11 over 80%, and that LP01–LP21 50%. simulation results show increases...
Objective: Cervical cancer is one of the two biggest killers women and early detection cervical precancerous lesions can effectively improve survival rate patients. Manual diagnosis by combining colposcopic images clinical examination results main method at present. Developing an intelligent algorithm based on artificial intelligence inevitable trend to solve objectification quality efficiency diagnosis.Approach: A multimodal fusion convolutional neural network (CMF-CNN) was proposed for...
Traditional method of epileptic seizure detection could not avoid the process manually selecting features. Recently, development deep learning technology has provided a new direction. This paper introduces based on EEG signal using short time Fourier transform(STFT) and convolution neural network(CNN). And verifies feasibility this through actual research data parameter setting. Afterwards, single threshold is adopted to combine multi-channel results. Then, comparison with classical support...
Bone segmentation is important in computed tomography (CT) imaging of the pelvis, which assists physicians early diagnosis pelvic injury, planning operations, and evaluating effects surgical treatment. This study developed a new algorithm for accurate, fast, efficient pelvis. The proposed method consists two main parts: extraction key frames CT images. Key were extracted based on pixel difference, mutual information normalized correlation coefficient. In pelvis phase, skeleton from images...
This paper proposes an improved region-based active contour model for segmenting magnetic resonance imaging (MRI) images of brain tuberculosis by combining a global energy fitting term and local term. First, is utilized to extract image information, which guides the evolving curve globally approximates intensity inside outside contour. Second, proposed describe inhomogeneity based on variance adaptive difference. Third, Fuzzy C-Means (FCM) clustering method applied pre-segment MRI...
Abstract Background Chronic obstructive pulmonary disease (COPD) is a chronic respiratory that seriously threatens people’s health, with high morbidity and mortality worldwide. At present, the clinical diagnosis methods of COPD are time-consuming, invasive, radioactive. Therefore, it urgent to develop non-invasive rapid severity technique suitable for daily screening in practice. Results This study established an effective model preliminary using lung sounds few channels. Firstly,...