Your address will show here +12 34 56 78
  • Text Hover
Cyber PSG

   CertiCon has developed a software framework for the semi-automatic classification of biosignals, optimised for the analysis of sleep data. The system is hardware independent and enables the implementation of an interface for the reception of various signals from various types of devices. The features essential for the automatic classification of individual segments of records in the hierarchically-arranged tree clusters are extracted from the input signals using advanced methods. The system then performs the classification of the sampled data in the tree structure of hierarchically-arranged classes (clusters) and offers experts key parts of the record recommended for evaluation.


    

  • Text Hover

The system developed by CertiCon enables a combination of the manual classification of the record by an expert with a high degree of automation, based on the evaluation of signal similarity, realised in the form of hierarchical clustering.

 

The advantages of the developed approach are primarily:  


» significantly faster classification of PSG records while maintaining the expert’s full control over classification process

 

» increased objectivity of record evaluation compared to purely manual classification

 

» increased accuracy of classification (in comparison with fully automatic classifiers)

 

» the possibility of use independently of the type of measuring (after the optimisation of the system for the given type of task) or hardware used (the system is not bound by a specific device manufacturer)

Demo application contains these parts:

 » data retrieval

» display of biosignals

» display of results of hierarchical clustering

» manual PSG classification 

» semi-automatic recurrent PSG classification 

» display of clinically interesting record parameters  

» display of classification results (hypnogram)  

» display of spectrogram


 

  • Text Hover

The analysis of PSG sleep data is a time-demanding manual process requiring extensive experience on the part of the evaluator. This process is encumbered by a great degree of subjectivity. Agreement between two experts for the classification of sleep phases is given in the range of 70 – 80%. Existing tools for the processing of biosignals usually offers the support for manual signal classification. There are also approaches promising fully automated classification. But doctors do not trust them too much. The problem with fully automated classification is the relatively low reliability of the given number of factors (a high degree of individual variability of biosignals, the dependence of classifiers on executing the records under precisely defined conditions, i.e. the arrangement of electrodes, HW filter settings, etc.)

The purpose of the demo application is to present the advantages of the chosen semiautomatic recursive classification approach realised in modern software architecture suitable both for the implementation of a separate client application, as well as for inclusion in existing biosignal analysis systems in the form of libraries.


CyberPSG was developed by CertiCon on the basis of joint research realised in the Czech Technical Univiersity in Prague.

Industry 4.0