A5072177950- Retinal Imaging and Analysis
- AI in cancer detection
- Medical Image Segmentation Techniques
- Advanced Neural Network Applications
- Digital Imaging for Blood Diseases
- Retinal and Optic Conditions
- Cutaneous Melanoma Detection and Management
- Cell Image Analysis Techniques
- melanin and skin pigmentation
- Glaucoma and retinal disorders
- Big Data and Business Intelligence
- COVID-19 diagnosis using AI
- Brain Tumor Detection and Classification
- Medical Imaging and Analysis
- Pluripotent Stem Cells Research
- Sports Analytics and Performance
- Sports Performance and Training
- Retinal Diseases and Treatments
- Atherosclerosis and Cardiovascular Diseases
Allen Institute for Cell Science
2024
SRM Institute of Science and Technology
2024
University of Notre Dame
2019-2022
Medical image segmentation plays a vital role in disease diagnosis and analysis. However, data-dependent difficulties such as low contrast, noisy background, complicated objects of interest render the problem challenging. These diminish dense prediction make it tough for known approaches to explore data-specific attributes robust feature extraction. In this paper, we study medical by focusing on extraction achieve improved prediction. We propose new deep convolutional neural network (CNN),...
Background: Today's machine-learning based dermatologic research has largely focused on pigmented/non-pigmented lesions concerning skin cancers. However, studies machine-learning-aided diagnosis of depigmented non-melanocytic lesions, which are more difficult to diagnose by unaided eye, very few. Objective: We aim assess the performance deep learning methods for diagnosing vitiligo deploying Convolutional Neural Networks (CNNs) and comparing their accuracy with that human raters different...
Accurate vessel segmentation in retinal images is vital for retinopathy diagnosis and analysis. However, existence of very thin vessels low image contrast along with pathological conditions (e.g., capillary dilation or microaneurysms) render the task difficult. In this work, we present a novel approach focusing on improving segmentation. We develop deep convolutional neural network (CNN), which exploits specific characteristics input data to use supervision, improved accuracy. particular,...
Convolutional neural networks (CNNs) for biomedical image analysis are often of very large size, resulting in high memory requirement and latency operations. Searching an acceptable compressed representation the base CNN a specific imaging application typically involves series time-consuming training/validation experiments to achieve good compromise between network size accuracy. To address this challenge, we propose CC-Net, new complexity-guided compression scheme segmentation. Given model,...
From diagnosing cardiovascular diseases to analyzing the progression of diabetic retinopathy, accurate retinal artery/vein (A/V) classification is critical. Promising approaches for A/V classification, ranging from conventional graph based methods recent convolutional neural network (CNN) models, have been known. However, inability traditional utilize deep hierarchical features extracted by CNNs and limitations current CNN incorporate vessel topology information hinder their effectiveness....
Compression is a standard procedure for making convolutional neural networks (CNNs) adhere to some specific computing resource constraints. However, searching compressed architecture typically involves series of time-consuming training/validation experiments determine good compromise between network size and performance accuracy. To address this, we propose an image complexity-guided compression technique biomedical segmentation. Given any constraints, our framework utilizes data complexity...
The epithelial to mesenchymal transition (EMT) is a widely studied but poorly defined state change due the variety of ways in which it has been characterized cells. There need for reproducible cell model systems that enable integration and comparison different types measured observations cells across many distinct cellular contexts. We present human induced pluripotent stem (hiPS) as such system by demonstrating its utility through comparative analysis hiPS cell-EMT 2D 3D culture geometries....
From diagnosing neovascular diseases to detecting white matter lesions, accurate tiny vessel segmentation in fundus images is critical. Promising results for have been known. However, their effectiveness segmenting vessels still limited. In this paper, we study retinal by incorporating into our framework the overall segmentation. To achieve this, propose a new deep convolutional neural network (CNN) which divides two separate objectives. Specifically, consider and as individual Then,...
Medical image segmentation using deep neural networks has been highly successful. However, the effectiveness of these is often limited by inadequate dense prediction and inability to extract robust features. To achieve refined prediction, we propose densely decoded (ddn), selectively introducing 'crutch' network connections. Such connections in each upsampling stage decoder (1) enhance target localization incorporating high resolution features from encoder, (2) improve facilitating...
Convolutional neural networks (CNNs) for biomedical image segmentation are often of very large size, resulting in high memory costs and latency operations. To ensure CNNs' accommodation key computing resource constraints specific applications, network compression is commonly used. However, time-consuming training/validation experiments involved when searching a compressed CNN imaging application, order to achieve desired compromise between the size accuracy. Recognizing that images tend have...