الفهرس 
INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 16 |  |  |
| | | from | 16 | | |
Abstract
Listening and interpreting lung sounds by a stethoscope had been an important component of screening and diagnosing lung diseases. However this practice has always been vulnerable to poor audibility, inter-observer variations (between different physicians). Thus computerized analysis of lung sounds for objective diagnosis of lung diseases is seen as a probable aid.
In the present research different lung sounds as normal, abnormal, and adventitious lung sounds were analyzed by wheeze detection and classification to inspiratory or expiratory, sibilant or sonorous, and monophonic or polyphonic wheeze using the MATLAB.
Detection of wheeze episodes in lung sound recordings gives an important clue about the extent of underlying lung disease. The frequency range (225-2000Hz) and duration of wheeze (≥ 250 msec) used for automatic wheeze detection and the wheeze classification were adopted based on, the ATS definition of wheeze and the development over previous approaches that were used as a reference.
The present research system is very robust, computationally simple and yielded overall sensitivity 90% for wheeze episode detection. The algorithm differentiates between monophonic wheezes and polyphonic wheezes with sensitivity 91% and specificity 66%.In case of other lung sounds the proposed algorithm excluded normal sounds from being identified as a wheeze with the specificity of 90%.