User Tools
Sidebar
Table of Contents
Research areas
Deep neural networks in early detection of melanomas
Computer-aided dermatopathology
Research goal:
To improve skin melanoma diagnosis by providing additional means of automatic detection of important diagnostic features in histopathological images.
The research topics include:
- tissue segmentation
- epidermis segmentation
- nests of melanocytes segmentation
- epidermal morphometry measurement
Anomaly detection with the use of pre-trained CNN architectures
Research goal:
To detect anomalies in multivariate diagnostic signals of the synchrotron control system by pre-trained CNN architectures.
SOLARIS National Synchrotron Radiation Centre is a research facility that provides high quality synchrotron light. To control such a complex system it is necessary to monitor signals from various devices and subsystems. Anomaly detection prevents from financial loss, unplanned downtimes and in extreme cases cause damage. As artificial intelligence techniques including machine learning and deep neural networks have become state-of-the-art solutions for anomaly detection tasks which are one of the most challenging in data analysis, our team conducts research on the use of them for anomaly detection in multivariate diagnostic signals.
The research topics include:
- data mining and preparation
- data preprocessing
- CNN architectures building
Detection and analysis of patterns
Cell detection
To improve speed and quality of testing new drugs against Clostridium difficile infection, we developed an algorithm for automatic bacteria cytotoxicity classification. It was based on two kinds of fluorescence images - DAPI and GFP. We experimented with many different methods from classical image processing and machine learning algorithms to convolutional neural networks. This research was was conducted in cooperation with Stanford University.
The research topics include:
- fluorescence images
- image processing
- sharing information from different images
- convolutional neural networks
Bacteria response clustering
Newly opened project, in cooperation with Stanford University. The main idea is to analyse bacteria reaction in response to different serums.
The research topics include:
- data clustering
- bacteria response analysis
Reconstructing images' missing areas with generative models
The research topics include:
- reconstruction
- generative learning
- GANs
- autoencoders
