Daud, Aidaayu (2009) Localization of abnormality in xray images of lungs. Masters thesis, Universiti Teknologi Malaysia, Faculty of Electrical Engineering.
PDF (1st Chapter)
PDF (Table of Contents)
An automated method is presented and proposed to detect abnormalities in frontal chest radiographs which are aggregated into an overall abnormality score. The process flow or sequence of steps are approached by using pure basic image processing techniques. The aim of this project is at finding abnormal signs of a diffuse clot of region and localized the abnormalities through the size and location from the lung image but will not determine the type of abnormalities of the disease. By using MATLAB code function and program, the scheme starts by identifying the category of the input lung image in DICOM format based on histogram area values measured and followed by the image segmentation of the lung fields with edge detection function. Edges associated with the boundaries and thresholding was used and binary images were created from the grayscale on the lung image done at histogram level corresponds to lights of region of interest on dark background. The region of interest were separated and extracted from the background by Morphology process. After getting the segmentation results for the left and right lungs of the largest size of mean area, other clot regions which were separated from the expected lung fields were identified and revealed. The abnormal clot regions were marked and labeled to differentiate the abnormalities to be seen compared with normal lung images.
|Item Type:||Thesis (Masters)|
|Additional Information:||Supervisor : Assoc. Prof. Dr. Syed Abdul Rahman Syed Abu Bakar; Thesis (Sarjana Kejuruteraan (Elektrik - Komputer dan Mikroelektronik)) - Universiti Teknologi Malaysia, 2009|
|Uncontrolled Keywords:||image processing, imaging systems in medicine|
|Subjects:||R Medicine > R Medicine (General)|
|Deposited By:||Kamariah Mohamed Jong|
|Deposited On:||11 Oct 2013 01:38|
|Last Modified:||11 Oct 2013 01:45|
Repository Staff Only: item control page