Sound-based wheeze detection techniques
Respiratory disorder is one of the common illnesses diagnosed. However, the common method used to diagnose the presence of abnormal respiratory behaviour is through the use of non-computerized instrument, stethoscope. As the modern technology advances, the use of a computer-aided diagnosis system is...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project (FYP) |
| Language: | English |
| Published: |
2016
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/67858 |
| _version_ | 1811686557943332864 |
|---|---|
| author | Soh Ruiyang |
| author2 | Ser Wee |
| author_facet | Ser Wee Soh Ruiyang |
| author_sort | Soh Ruiyang |
| collection | NTU |
| description | Respiratory disorder is one of the common illnesses diagnosed. However, the common method used to diagnose the presence of abnormal respiratory behaviour is through the use of non-computerized instrument, stethoscope. As the modern technology advances, the use of a computer-aided diagnosis system is able to greatly reduce the human error involved when using stethoscope. The respiratory sound of the patient can be recorded and input as a digital signal. In this project, characteristics of 19 normal respiratory signals and 19 wheezing respiratory signals were studied. The characteristics of both the kurtosis and F50/F90 ratio features were studied. With reference to the characteristics displayed, Fisher Discriminant Analysis was applied to design a classifier that can distinguish the difference between both classification groups; normal and wheezing. The robustness of the classifier was measured using re-substitution error technique and cross-validation technique. At the later part of the project, Neyman Pearson Hypothesis Test was introduced to determine the classification group of the test data based on the characteristics of kurtosis and F5/F90 ratio features individually. Lastly, the detection will be done using the classifier designed using FDA. The detection phase had achieved an accuracy rating of 78.95% – 90.3%. |
| first_indexed | 2024-10-01T05:02:19Z |
| format | Final Year Project (FYP) |
| id | ntu-10356/67858 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T05:02:19Z |
| publishDate | 2016 |
| record_format | dspace |
| spelling | ntu-10356/678582023-07-07T17:22:41Z Sound-based wheeze detection techniques Soh Ruiyang Ser Wee School of Electrical and Electronic Engineering DRNTU::Engineering Respiratory disorder is one of the common illnesses diagnosed. However, the common method used to diagnose the presence of abnormal respiratory behaviour is through the use of non-computerized instrument, stethoscope. As the modern technology advances, the use of a computer-aided diagnosis system is able to greatly reduce the human error involved when using stethoscope. The respiratory sound of the patient can be recorded and input as a digital signal. In this project, characteristics of 19 normal respiratory signals and 19 wheezing respiratory signals were studied. The characteristics of both the kurtosis and F50/F90 ratio features were studied. With reference to the characteristics displayed, Fisher Discriminant Analysis was applied to design a classifier that can distinguish the difference between both classification groups; normal and wheezing. The robustness of the classifier was measured using re-substitution error technique and cross-validation technique. At the later part of the project, Neyman Pearson Hypothesis Test was introduced to determine the classification group of the test data based on the characteristics of kurtosis and F5/F90 ratio features individually. Lastly, the detection will be done using the classifier designed using FDA. The detection phase had achieved an accuracy rating of 78.95% – 90.3%. Bachelor of Engineering 2016-05-23T03:26:34Z 2016-05-23T03:26:34Z 2016 Final Year Project (FYP) http://hdl.handle.net/10356/67858 en Nanyang Technological University 66 p. application/pdf |
| spellingShingle | DRNTU::Engineering Soh Ruiyang Sound-based wheeze detection techniques |
| title | Sound-based wheeze detection techniques |
| title_full | Sound-based wheeze detection techniques |
| title_fullStr | Sound-based wheeze detection techniques |
| title_full_unstemmed | Sound-based wheeze detection techniques |
| title_short | Sound-based wheeze detection techniques |
| title_sort | sound based wheeze detection techniques |
| topic | DRNTU::Engineering |
| url | http://hdl.handle.net/10356/67858 |
| work_keys_str_mv | AT sohruiyang soundbasedwheezedetectiontechniques |