JavaScript is disabled for your browser. Some features of this site may not work without it.
| dc.contributor.author | Amarech, Kiflie | |
| dc.date.accessioned | 2021-10-01T10:32:54Z | |
| dc.date.available | 2021-10-01T10:32:54Z | |
| dc.date.issued | 2021-07 | |
| dc.identifier.uri | http://ir.bdu.edu.et/handle/123456789/12679 | |
| dc.description.abstract | Tuberculosis is one of the top 10 communicable and largely affected diseases in the world. To diagnose this highly crucial and rapidly transmitted disease sputum smear microscopic consider as the golden one from other different mechanisms in the aspect of maintainability, accessibility and easiness of resource challenging countries. Major and minor errors happen from day-to-day activity and to improve the performance of diagnosis mechanism external quality assessment implemented by national TB program with three metrics which are onsite evaluation, blind rechecking and panel testing. This research has been done in different Eruption‘s country like Indian, Nigeria and also Ethiopian which are Tigray, Amahira as well as Oromia region to assess the performance of TB diagnosis by looking at the root cause of Sputum smear preparation quality and another one. However, till now it has its own limitations because it is done manually. In order to fill the gap, we develop a model of sputum smear quality inspection by using ensemble feature extraction approach. The dataset recorded in regional lab after it collected from Gabi hospital, Felege Hiwot hospital, Adit clinic and also Gondar hospital as well as Kidanemihrte clinic in Gondar and labeled it by experts at regional lab here in Bahir Dar city around felegehiywot Hospital. To minimize environmental factors and elimination variation we consider controlled environments. All the data recorded using smartphone (common 15) in jpg file extension and 1728x3840 pixel dimension. Bicubic based resizing, ROI extraction using thresholding; sequential Gaussian and Gabor filters in noise reduction, augmentation and CLAHE for enhancement have been done before feature extraction; GLCM from gray label and CNN from color image selected for feature extraction. Ensemble feature fused to classifier; finally, after we have tested both CNN and GLCM features using CNN, SVM and KNN classifier, KNN scores better one which is 87%, 93% and 94% for GLCM, CNN and hybrid of CNN with GLCM respectively. | en_US |
| dc.language.iso | en_US | en_US |
| dc.subject | ELECTRICAL AND COMPUTER ENGINEERING | en_US |
| dc.title | Sputum smears quality inspection using an ensemble feature extraction approach | en_US |
| dc.type | Thesis | en_US |
