Abstract: Ophthalmologists rely on retinal vessel segmentation. Low contrast, variable vessel size and thickness, and micro-aneurysms and hemorrhages complicate this issue. Early methods used hand-crafted filters and morphological post-processing to capture vessel structures. Deep learning…
Abstract: A deep convolutional neural network for breast cancer screening exam classification was trained and evaluated on over 200000 exams (1000000 images). Our screening population network predicts breast cancer with an AUC of 0.895. Technical…
Abstract: Early diagnosis of retinal diseases, the leading cause of permanent blindness, requires segmenting retinal layers in optical coherence tomography (OCT) images. Thus, segmentation is crucial to vision preservation. Expert ophthalmologists must manually segment retinal…
Abstract: Accurate diagnostics require high-resolution MR images. Hardware and processing limits image resolution. Deep learning has produced impressive image enhancement/super-resolution results recently. We propose a new regularized network that uses image priors—a low-rank structure and…
Abstract: Picture Wise Perception-oriented image and video processing uses Just Noticeable Difference (PW-JND), the minimum difference a human visual system can perceive. Conventional Just Noticeable Difference (JND) models calculate the JND threshold for each pixel…
Abstract: MEF-Net is a fast multi-exposure image fusion (MEF) method for static image sequences of any spatial resolution and exposure number. A fully convolutional network predicts weight maps from a low-resolution input sequence. Guided filters…
Abstract: Multi-modal image super-resolution (MISR) uses an HR image from another modality to find the HR version of a low-resolution (LR) image. This paper designs a model-based MISR deep network architecture. A novel joint multi-modal…
Abstract: Single-image de-raining is difficult because rain streaks vary in size, direction, and density. Rain streaks affect various parts of the image differently. Previous methods used prior information to remove rain streaks from a single…
Abstract: Compressive sensing requires a sensing matrix that is random enough for good signal reconstruction and has desirable properties like orthogonality or circulancy. First, generate a full orthogonal circulant matrix, then select a few rows….
Abstract: Doctors rely on volumetric imaging. Convolutional neural networks (CNN) for volumetric image analysis require detailed training data and GPU memory. This project presents the volumetric image classification problem as a multi-instance classification problem and…
Abstract: Digital entertainment relies heavily on photo-to-face sketch synthesis. An intelligent face sketch synthesis system must be light-resistant. Such a system will perform well in real-world uncontrolled lighting conditions. However, previous face sketch synthesis methods…
Abstract: DR and DME are the leading causes of permanent blindness in working-age people. Automatic DR and DME grading helps ophthalmologists customize patient treatments. However, prior works either grade DR or DME, ignoring the correlation…
WhatsApp us