A5000597263- Radiomics and Machine Learning in Medical Imaging
- MRI in cancer diagnosis
- Hepatocellular Carcinoma Treatment and Prognosis
- Liver Disease Diagnosis and Treatment
- Advanced Neural Network Applications
- Advanced MRI Techniques and Applications
- Adversarial Robustness in Machine Learning
- Renal cell carcinoma treatment
- Liver Disease and Transplantation
- Medical Image Segmentation Techniques
- Brain Tumor Detection and Classification
- Medical Imaging and Analysis
- Mesenchymal stem cell research
- Renal and related cancers
- Nutrition and Health in Aging
- AI in cancer detection
- Advanced Neuroimaging Techniques and Applications
- Advanced Vision and Imaging
- Anesthesia and Pain Management
- 3D Printing in Biomedical Research
- Drug Transport and Resistance Mechanisms
- Endometrial and Cervical Cancer Treatments
- Image Processing Techniques and Applications
- Anomaly Detection Techniques and Applications
- Bone Tissue Engineering Materials
Beijing Friendship Hospital
2016-2025
Zhongshan Hospital
2014-2025
Fudan University
2014-2025
Hebei Medical University
2025
First Affiliated Hospital of Hebei Medical University
2025
Capital Medical University
2016-2024
Weihai Municipal Hospital
2024
Dalian Minzu University
2022
University of Michigan–Ann Arbor
2016-2021
National Clinical Research Center for Digestive Diseases
2016-2021
This paper studies single-image depth perception in the wild, i.e., recovering from a single image taken unconstrained settings. We introduce new dataset "Depth Wild" consisting of images wild annotated with relative between pairs random points. also propose algorithm that learns to estimate metric using annotations depth. Compared state art, our is simpler and performs better. Experiments show algorithm, combined existing RGB-D data annotations, significantly improves wild.
In Autonomous Driving (AD) systems, perception is both security and safety critical. Despite various prior studies on its issues, all of them only consider attacks camera- or LiDAR-based AD alone. However, production systems today predominantly adopt a Multi-Sensor Fusion (MSF) based design, which in principle can be more robust against these under the assumption that not fusion sources are (or be) attacked at same time. this paper, we present first study issues MSF-based systems. We...
Highly expressive models such as deep neural networks (DNNs) have been widely applied to various applications. However, recent studies show that DNNs are vulnerable adversarial examples, which carefully crafted inputs aiming mislead the predictions. Currently, majority of these focused on perturbation added image pixels, while manipulation is not physically realistic. Some works tried overcome this limitation by attaching printable 2D patches or painting patterns onto surfaces, but can be...
Background The Liver Imaging Reporting and Data System (LI‐RADS) is widely adopted for noninvasive diagnosis of hepatocellular carcinoma (HCC). It's updated to version 2018 recently, with some major changes compared v2017. However, the diagnostic performance LI‐RADS v2018 its difference v2017 are yet be validated. Purpose To compare performances on MR diagnosing HCC between v2018. Study Type Retrospective. Subjects In all, 181 patients 217 hepatic observations (146 HCCs, 16 non‐HCC...
We aim to develop and validate a three-dimensional convolutional neural network (3D-CNN) model for automatic liver segment segmentation on MRI images.This retrospective study evaluated an automated method using deep that was trained, validated, tested with 367, 157, 158 portal venous phase MR images, respectively. The Dice similarity coefficient (DSC), mean surface distance (MSD), Hausdorff (HD), volume ratio (RV) were used quantitatively measure the accuracy of segmentation. time consumed...
BACKGROUND Ulcerative colitis (UC) is a complex inflammatory bowel disease, and its etiology pathogenesis remain incompletely elucidated. AIM To analyze the effects of Saccharomyces boulardii in combination with sulfasalazine on intestinal microbiota barrier function patients UC. METHODS A retrospective analysis clinical data from 127 UC admitted to our hospital between January 2021 2023 was conducted. All met complete inclusion exclusion criteria. Based treatment interventions received,...
ABSTRACT Accurate medical image segmentation is crucial for clinical diagnosis and disease treatment. However, there are still great challenges most existing methods to extract accurate features from images because of blurred boundaries various appearances. To overcome the above limitations, we propose a novel network named TS‐Net that effectively combines advantages CNN Transformer enhance feature extraction ability. Specifically, design Multi‐scale Convolution Modulation (MCM) module...
Tissue engineering approaches using the combination of porous ceramics and bone marrow mesenchymal stem cells (BMSCs) represent a promising substitute for repairing large defects. Nevertheless, optimal conditions constructing tissue-engineered have yet to be determined. It remains unclear if transplantation predifferentiated BMSCs is superior undifferentiated or freshly isolated mononucleated (BMNCs) in terms new formation vivo. The aim this study was investigate effect vitro osteogenic...
Background: With the development of new magnetic resonance imaging (MRI) techniques, an increasing number articles have been published regarding hepatocellular carcinoma (HCCMRI) in past decade. However, few studies statistically analyzed these articles. In this study, we aim to systematically evaluate scientific outcomes HCCMRI research and explore hotspots from 2008 2017. Methods: The included 2017 were downloaded Web Science Core Collection verified by two experienced radiologists. Excel...
In this paper, we address the shape-from-shading problem by training deep networks with synthetic images. Unlike conventional approaches that combine learning and imagery, propose an approach does not need any external shape dataset to render Our consists of two synergistic processes: evolution complex shapes from simple primitives, a network for shape-from-shading. The generates better guided training, while improves using evolved shapes. We show our achieves state-of-the-art performance on...
Cirrhotic patients with sarcopenia have poor prognoses and higher mortality. The third lumbar vertebra (L3) skeletal muscle index (SMI) is widely used to assess sarcopenia. However, L3 generally outside the scanning volume on standard liver MRI.To investigate SMIs change between slices in cirrhotic relationships SMI at 12th thoracic (T12), first (L1) second (L2) levels L3-SMI accuracy of estimated L3-SMIs diagnosing sarcopenia.Prospective.A total 155 (109 sarcopenia, 67 male; 46 without 18...
The purpose of this study was to evaluate the use diffusion-weighted imaging (DWI) for detection and characterisation focal hepatic lesions compared with T(2) weighted imaging.45 patients 97 (51 malignant 46 benign lesions) were included in retrospective study. Malignant 12 hepatocellular carcinomas, 26 metastases 13 intrahepatic cholangiocarcinomas. Benign 19 haemangiomas 27 cysts. MRI protocol upper abdomen images, in- opposed-phase T(1) images dynamic images. Breath-hold fat-suppressed...
The concept of establishing a dental stem cell (DSC) bank for oral and maxillofacial regeneration has become great interest but it remains at primitive stage. routine application serum-containing conditions human DSC (hDSC) culture is in controversy considering that the animal-originated serum can cause serious ethical concerns lead to increasingly irrelevant variables, errors, poor repeatability experiment results. Thus, this study aimed establish safe, stable efficient hDSC serum-free...
Background: The value of magnetic resonance elastography (MRE) in portal hypertension (PH) has yet to be determined the context chronic liver disease (CLD). This study examined MRE for prediction hepatic venous pressure gradient (HVPG) and high-risk esophageal varices (EVs) a CLD cohort with generally high HVPG. Methods: Patients who underwent both HVPG measurement two-dimensional examination at Beijing Friendship Hospital between April 2018 March 2022 were prospectively included....
Abstract Purpose To develop a nomogram for preoperative assessment of microvascular invasion (MVI) in hepatocellular carcinoma (HCC) based on the radiological features enhanced CT and to verify two imaging techniques (CT MRI) an external centre. Method A total 346 patients were retrospectively included (training, n = 185, images; testing 1, 90, 2, 71, MRI images), including 229 MVI-negative 117 MVI-positive patients. The clinical information images analysed, independent variables associated...
Apparent diffusion coefficient (ADC) measurement in diffusion-weighted imaging (DWI) has been reported to be a helpful biomarker for detection and characterization of lesion. In view the importance ADC reproducibility, aim this study was probe variability healthy hepatic values measured at 3 MR scanners from different vendors with field strengths, investigate reproducibility normalized (nADC) value spleen as reference organ. Thirty enrolled volunteers received DWI GE 1.5T, Siemens Philips...
Purpose. To evaluate the diagnostic performance of deep learning with a multichannel fusion three-dimensional convolutional neural network (MCF-3DCNN) in differentiation pathologic grades hepatocellular carcinoma (HCC) based on dynamic contrast-enhanced magnetic resonance images (DCE-MR images). Methods and Materials. Fifty-one histologically proven HCCs from 42 consecutive patients January 2015 to September 2017 were included this retrospective study. Pathologic examinations revealed nine...