3D-convolutional neural network
PulseAugur coverage of 3D-convolutional neural network — every cluster mentioning 3D-convolutional neural network across labs, papers, and developer communities, ranked by signal.
4 day(s) with sentiment data
-
Hyperspectral image classification vulnerable to data leakage via spatial overlap
A new research paper highlights a critical issue in hyperspectral image classification: data leakage caused by spatial overlap in patch-based sampling. When training and testing data are drawn from the same image withou…
-
Multi-task learning fails to improve pulmonary nodule malignancy assessment in CT scans
A new research paper published on arXiv explores the effectiveness of multi-task learning for assessing pulmonary nodule malignancy in 3D CT scans. The study compared a single-task 3D convolutional neural network with a…
-
CNN-DNN architecture accelerates diagnostic model training for COVID-19
Researchers have developed a parallel training approach using a hybrid CNN-DNN architecture to accelerate the development of diagnostic models, particularly for diseases like COVID-19. This method was tested on 300 CT s…
-
New AI framework reduces annotation needs for colonoscopy polyp detection
Researchers have developed WSPolypNet, a novel weakly supervised framework designed for polyp localization in colonoscopy videos. This system utilizes a 3D convolutional neural network to generate class activation maps …
-
AI model adapts to Alzheimer's MRI tasks with minimal retraining
Researchers have developed a generalizable feature extractor for Alzheimer's disease-related brain MRI tasks, demonstrating the effectiveness of transfer learning in neuroimaging. By adapting a pre-trained 3D convolutio…
-
Deep learning corrects X-ray micro-CT image jitter
Researchers have developed a novel method for correcting image distortions in X-ray phase-contrast micro computed tomography. This technique utilizes a deep learning model to estimate and compensate for projection jitte…
-
EquiPocket: E(3)-equivariant GNN advances protein binding site prediction
A research paper introduced EquiPocket, an E(3)-equivariant Graph Neural Network designed for predicting protein binding sites, a crucial step in drug discovery. Unlike traditional 3D CNN methods that struggle with irre…
-
New HMD Framework Offers Efficient Hyperspectral Image Classification
Researchers have developed a new framework called Holistic Multivariance Decomposition (HMD) for hyperspectral image classification. This novel approach aims to improve accuracy and efficiency by capturing complex spati…
-
Deep learning tracks workpieces in hot forging environments
Researchers have developed a new framework for tracking workpieces in hot forging environments using deep learning and event-driven finite state machines. This system infers workpiece locations by analyzing data from mu…
-
6G ISAC framework uses AI for railway intrusion detection
Researchers have developed a novel framework for railway safety using integrated sensing and communication (ISAC) technology, which combines sensing and communication capabilities to optimize wireless resource usage. Th…
-
Deep learning framework automates fetal brain biometry from MRI
Researchers have developed a new deep learning framework to automate fetal brain biometry using MRI scans. This four-step pipeline jointly estimates linear measurements and their anatomical landmarks, aiming to improve …
-
AI models for COVID-19 CT scan classification show promising results · 2 papers
Two research papers submitted to arXiv propose novel methods for classifying COVID-19 cases from chest CT scans. The first paper introduces a hybrid 2D/3D Convolutional Neural Network (CNN) that extracts features from m…
-
New AI model enhances driver monitoring with spatiotemporal facial analysis
Researchers have developed a novel spatiotemporal architecture called the Twin Cycle Autoencoder (TCA) for detecting facial Action Units (AUs) in driver monitoring systems. This new model addresses challenges like varia…
-
AI framework optimizes genetic algorithm hyperparameters for material design
Researchers have developed a multi-fidelity framework to optimize genetic algorithm (GA) hyperparameters for lattice material design. This system uses a combination of high-fidelity Fast Fourier Transform (FFT) homogeni…
-
Bayesian optimization framework enhances genetic algorithm hyperparameter tuning for materials science
Researchers have developed a multi-fidelity framework to optimize genetic algorithm (GA) hyperparameters for lattice material design. This framework uses a combination of high-fidelity Fast Fourier Transform (FFT) homog…
-
New LaryngealCT Dataset Benchmarks Deep Learning for Cancer Staging
Researchers have developed LaryngealCT, a new benchmark dataset for staging laryngeal cancer using deep learning models. The dataset comprises 1,029 CT scans aggregated from The Cancer Imaging Archive and has been used …
-
Deep learning predicts surgical risk from CT scans
Researchers have developed a deep learning pipeline to predict postoperative pancreatic fistula (POPF) using preoperative CT scans. The system automates the process from pancreatic segmentation to classification, aiming…
-
RealLiFe achieves real-time light field reconstruction via sparse gradient descent
Researchers have developed RealLiFe, a novel method for real-time light field reconstruction from sparse input images. This technique leverages Hierarchical Sparse Gradient Descent (HSGD) to optimize a coarse Multi-plan…
-
Simple MIL matches complex models for 3D neuroimage classification
Researchers have published a benchmark comparing multiple instance learning (MIL) methods against 3D CNNs and ViTs for classifying 3D neuroimages. The study found that a simple mean pooling MIL approach, without attenti…