Method for endoscopic bleeding detection in videos using high-level visual features and image processing. These features are used to enhance the input of convolutional neural networks (CNNs) like Resnet and VGG.
AngioScore, an AI-based web application, was developed to semi-automatically and objectively calculate the SYNTAX score for coronary artery disease treatment strategy, showing promising results in assessing coronary artery lesions
Comparison of four state-of-the-art neural network models for automatic liver tumor segmentation, with a focus on tumor segmentation over liver parenchyma segmentation
Retrospective longitudinal analysis of antipsychotics effects on death risk.
Evaluation of preprocessing methods in liver segmentation in computed tomography (CT) imaging
Novel residual depth-separable convolution block in MobileNet
Evaluation of Dynamic RNNs, scheduled sampling, BiLSTM, modified attention model in OCR.
Using crowd-sourced annotations to refine medical dataset quality in segmentation problems
Aorta and thrombus semantic segmentation using U-Net, ResNet, and VBNet frameworks
Guidelines for creating medical datasets useful in machine learning
Federated learning simulations using PaSSiFLora
Configuration auto-tuning of CPU+GPU hybrid systems
Deep neural networks training using IBM Power System AC922
Fully automatic system allowing to build classifiers of book pages and deploy them as dedicated mobile application
Dataset of thousands annotated gastrointestinal endoscopy images.
In this paper we describe development of a computer vision system for accurate detection and classification of clothes for e-commerce images.
Analysis of some properties of denoising autoencoders using the problem of misspellings correction as an exemplary task
An algorithm for improving results of automatic nerve detections in confocal microscopy images of human corneal.
Exploring dependencies between execution time and power consumption of training deep neural networks using modeling and simulation.
Overview of thresholding methods for labeling objects given a list of candidate classes' scores
Improving performance of RNN training in Kaldi in collaboration with VoiceLab.ai.
Definition of a set of high-level visual features of endoscopic bleeding and evaluate their capabilities for potential use in automatic bleeding detection
Recognition of eight user activities (eg walking, lying, climbing stairs) basing on the measurements from an accelerometer embedded in a mobile device.
Distributed system for point cloud processing and transferring them via computer network regarding to effectiveness-related requirements
Review of the existing bleeding detection methods focusing their attention on the image processing techniques utilised in the algorithms.
The paper introduces KernelHive, an environment for executing, designing and tuning parallel applications in multi-level heterogeneous high performance computing systems. System architecture, sample applications and performance results are reported.
System for diagnosis recommendation and disease detection, taking into account the limited mobility of the endoscope and the doctor’s requirements
System that improves dependability of medical image recognition by integration of results from redundant components.
