A5055595805- Medical Image Segmentation Techniques
- Domain Adaptation and Few-Shot Learning
- Medical Imaging Techniques and Applications
- COVID-19 diagnosis using AI
- Advanced Neural Network Applications
- Medical Imaging and Analysis
- Adversarial Robustness in Machine Learning
- Cancer-related molecular mechanisms research
- Integrated Circuits and Semiconductor Failure Analysis
- Generative Adversarial Networks and Image Synthesis
- Advanced MRI Techniques and Applications
- Radiomics and Machine Learning in Medical Imaging
- Anomaly Detection Techniques and Applications
University of California, Irvine
2023-2024
University of Oxford
2024
Yale University
2023
Image registration is an essential step in many medical image analysis tasks. Traditional methods for are primarily optimization-driven, finding the optimal deformations that maximize similarity between two images. Recent learning-based methods, trained to directly predict transformations images, run much faster, but suffer from performance deficiencies due domain shift. Here we present a new neural network based framework, called NIR (Neural Registration), which on optimization utilizes...
Diffeomorphic image registration, offering smooth transformation and topology preservation, is required in many medical analysis tasks. Traditional methods impose certain modeling constraints on the space of admissible transformations use optimization to find optimal between two images. Specifying right challenging: registration quality can be poor if too restrictive, while hard solve general. Recent learning-based methods, utilizing deep neural networks learn directly, achieve fast...
Advances in self-supervised learning have drawn attention to developing techniques extract effective visual representations from unlabeled images. Contrastive (CL) trains a model consistent features by generating different views. Recent success of Masked Autoencoders (MAE) highlights the benefit generative modeling learning. The approaches encode input into compact embedding and empower model's ability recovering original input. However, our experiments, we found vanilla MAE mainly recovers...
Recent advancements in self-supervised learning have demonstrated that effective visual representations can be learned from unlabeled images. This has led to increased interest applying the medical domain, where images are abundant and labeled difficult obtain. However, most approaches modeled as image level discriminative or generative proxy tasks, which may not capture finer necessary for dense prediction tasks like multi-organ segmentation. In this paper, we propose a novel contrastive...
Image registration is an essential step in many medical image analysis tasks. Traditional methods for are primarily optimization-driven, finding the optimal deformations that maximize similarity between two images. Recent learning-based methods, trained to directly predict transformations images, run much faster, but suffer from performance deficiencies due model generalization and inefficiency handling individual specific deformations. Here we present a new neural net based framework,...