An integrated software package for model-based neuro-fuzzy classification of small airway dysfunction
Abstract
Respiratory disorders can be difficult to differentiate, as their symptoms are sometimes similar to one another. Concerning asthma which is the major interest in this research, early detection of small airway impairment could greatly help with timely diagnosis. In this research, an integrated software package (ISP) has been developed to help clinicians in classification of asthma severities using Impulse Oscillation data. The ISP has been developed based on object-oriented methodology. In this package, IOS “diagnostic” parameters such as AX and frequency-dependence of resistance as well as parameter estimates of extended RIC (eRIC) and augmented RIC (aRIC) models (simplified cases of Mead’s model) are used as elements of feature vectors to represent respiratory impedance. Two Co-active Neuro-Fuzzy Inference Systems (CANFIS) are implemented in this software to classify respiratory patterns. Both of the systems used train data sets containing 112 patterns belonging to four different classes of asthma severities including asthma, small airway disease (SAD), mild SAD (mSAD) and normal. The first CANFIS utilized IOS impedance measurements and parameter estimates of the aRIC model, while the second CANFIS utilized parameter estimates of the eRIC model instead. In order to improve on the user-friendliness and running times of the previous versions of codes for these modules, they were modified and enhanced before integration into the ISP. To encapsulate the functionalities of the different components in the developed ISP, the adapter object-oriented design pattern was used in implementing the ISP. Following the object-oriented principles led to construction of fully object-oriented software. The flexibility of the software package in enabling the developer to add new components (modules) and to upgrade current components with newly developed versions makes it a robust and reliable platform to assist researchers and clinicians in pulmonary medicine who seek more sensitive diagnostic tools to explore small airway impairments based on IOS measurements. After performing a thorough analysis of requirements of the software, the architecture of the ISP was designed and implemented in C#, one of the supported programming languages by Microsoft Visual Studio 2008. The ISP has been tested successfully on Windows XP and Vista and it has an easy-to-use user interface to interact with the user.
Subject Area
Electrical engineering
Recommended Citation
Hafezi, Nazila, "An integrated software package for model-based neuro-fuzzy classification of small airway dysfunction" (2009). ETD Collection for University of Texas, El Paso. AAI1473869.
https://scholarworks.utep.edu/dissertations/AAI1473869