A5092150892- Radiomics and Machine Learning in Medical Imaging
- AI in cancer detection
- Advanced Image Processing Techniques
- Advanced Vision and Imaging
- Gene expression and cancer classification
- Digital Imaging for Blood Diseases
- Advanced X-ray and CT Imaging
- Medical Imaging Techniques and Applications
- Image and Signal Denoising Methods
- Colorectal Cancer Screening and Detection
- Data-Driven Disease Surveillance
- Advanced Neuroimaging Techniques and Applications
- Image Enhancement Techniques
- Computer Graphics and Visualization Techniques
- Digital Radiography and Breast Imaging
- Text and Document Classification Technologies
- Image Retrieval and Classification Techniques
- 3D Shape Modeling and Analysis
- Medical Image Segmentation Techniques
Tsinghua University
2015-2025
Tsinghua–Berkeley Shenzhen Institute
2023-2025
Southeast University
2019-2023
Ocean University of China
2019-2023
In this paper, we propose a novel method, namely GR-PSN, which learns surface normals from photometric stereo images and generates the under distant illumination different lighting directions materials. The framework is composed of two subnetworks, named GeometryNet ReconstructNet, are cascaded to perform shape reconstruction image rendering in an end-to-end manner. ReconstructNet introduces additional supervision for surface-normal recovery, forming closed-loop structure with GeometryNet....
Gigapixel Whole Slide Images (WSIs) aided patient diagnosis and prognosis analysis are promising directions in computational pathology. However, limited by expensive time-consuming annotation costs, WSIs usually only have weak annotations, including 1) WSI-level Annotations (WA) 2) Limited Patch-level (LPA). Currently, Multiple Instance Learning (MIL) often exploits WA, while LPA assign pseudo-labels for unlabeled data. Intuitively, can serve as a practical guide MIL, but the unreliable...
Detection of biomarkers breast cancer incurs additional costs and tissue burden. We propose a deep learning-based algorithm (BBMIL) to predict classical biomarkers, immunotherapy-associated gene signatures, prognosis-associated subtypes directly from hematoxylin eosin stained histopathology images. BBMIL showed the best performance among comparative algorithms on prediction immunotherapy related subtypes.
Automated classification of lymph node metastasis (LNM) plays an important role in the diagnosis and prognosis. However, it is very challenging to achieve satisfactory performance LNM classification, because both morphology spatial distribution tumor regions should be taken into account. To address this problem, article proposes a two-stage dMIL-Transformer framework, which integrates morphological information based on theory multiple instance learning (MIL). In first stage, double Max-Min...
Underwater images are generally of low quality, limiting the performance subsequent perceptual tasks, such as underwater object detection and recognition. However, only a few methods can improve quality by simultaneously restoring super-resolving images. In this paper, we propose an end-to-end trainable model based on generative adversarial networks (GANs) called Simultaneous Restoration Super-Resolution GAN (SRSRGAN) to obtain clear super-resolution automatically. particular, our leverages...
As DECT becomes widely accepted in the field of diagnostic radiology, there is growing interest using dual-energy imaging to improve other scenarios. In this context, a new mobile dual-source CBCT being developed for scenarios such as radiotherapy and interventional radiology. The device performs measurements by utilizing two X-ray sources mounted side-by-side z-axis direction, causing problem mismatch fields view high-energy low-energy z-axis. To solve problem, study proposes method based...
The rise of large language models (LLMs) has raised concerns about machine-generated text (MGT), including ethical and practical issues like plagiarism misinformation. Building a robust highly generalizable MGT detection system become increasingly important. This work investigates the generalization capabilities detectors in three aspects: First, we construct MGTAcademic, large-scale dataset focused on academic writing, featuring human-written texts (HWTs) MGTs across STEM, Humanities,...
In this multicountry study, we aim to explore the effectiveness of self-supervised learning (SSL) in colorectal cancer (CRC)-related predictive tasks using large amount unlabeled digital pathology imaging data.We adopted SimSiam conduct pretraining on two whole-slide image CRC data sets from United States and Australia. The SSL pretrained encoder is then used several tasks, including supervised (tissue classification, microsatellite instability v stability classification), weakly (polyp type...
<title>Abstract</title> Outbreaks of infectious diseases have caused tremendous human suffering and incalculable economic losses, are a global public health problem that threatens society. Therefore, it is necessary to model the spatial temporal distribution characteristics diseases, explore transmission trend establish an infection early warning take corresponding preventive control measures, which can make prevention work more targeted forward-looking. Given complex correlation variation...
Biomarkers of Breast cancer, such as estrogen receptors, progesterone receptor, gene signatures, and prognosis-related subtypes, can guide cancer treatment, but the detection these biomarkers incurs additional costs tissue burden on management disease. Here, we proposed a deep learning-based algorithm (Breast Biomarker Multiple Instance Learning, BBMIL) to predict classical biomarkers, immunotherapy-associated prognosis-associated subtypes directly from hematoxylin eosin (H&E) stained...