Ting, Shyue Siong
(2014)
Detection And Segmentation Of Mass Region In Mammogram Image.
Masters thesis, Perpustakaan Hamzah Sendut.
Abstract
Breast cancer is the most common form of cancer and continues to be a significant
public health problem amongst women around the world. Early detection is the most
promising way to decrease the number of patient suffering from this disease.
Currently, mammography is an effective diagnostic technique for early detection of
breast cancer. Numerous studies have been carried out to develop Computer-Aided
Diagnosis (CAD) system to help radiologists. Unfortunately, none of the proposed
techniques provide good results in both mass detection and segmentation. Thus, in
this study, an automated system to detect and segment the true mass regions of any
size, shape and margin in mammogram image is proposed. Initially, the Mean
Median Intersection Point (MMIP) algorithm is proposed to obtain a point in the
intensity histogram to be set as threshold value to segment the breast region and
distinguish it from the background in the mammogram image. Then, a modified
contrast enhancement technique, namely Multi-Stage Contrast Enhancement (MSCE)
technique is proposed to increase the contrast between the mass and breast tissue
regions. Finally, an automated seed based region growing process for mass detection
and segmentation is proposed. In the proposed technique, the seed point that
produces the smallest smoothness descriptor value is set as the initial mass centre for
the region growing process. In order to produce results which mimic true mass
region, the enhanced mammogram image is multiplied with the Gaussian function
centred at a seed point location.
Actions (login required)
 |
View Item |