iv ABSTRACT
DEVELOPMENT OF MEDICAL IMAGE SEGMENTATION IN SEMI-AUTOMATIC TO INCREASE VISUALIZING AND ANALYSE
CT-SCAN IMAGE OF PELVIS
By Irwan Setiawan NIM : 23304009
Pelvis anatomy is conceptually difficult to comprehend and complex, hence use conventional x-ray imaging less give comprehensive information. Therefore, use of Computerized Tomography (CT) scan enabling in multi-slice imaging become one alternative. Even Though, some limitation of CT-SCAN generally is orientation from slice selecting limited to axial orientation and also the image yielded representation of all struc ture of soft tissue at determined cross-section orientation of slice. Therefore, to representation of more comprehensive pelvis anatomy by using CT-SCAN image, next analyse from image and visualizing technique is needed.
To help diagnosis process more focus to one soft tissue, image segmentation technique needed. Segmentation structure anatomy of pelvis image is relative difficult work to do in manual because complexity and variation from anatomy structure detail and also the size of data is big. Conventional method of image segmentation have been developed for image analysis in general cannot resulting good performa nce if used in medical image, especially needing analysis for multi- slice image. Deformation model or GVF active contour model ( Gradient Vector Flow) becoming one of alternative to be applied at case of medical image.
At this research, performa nce examination from active contour model base on GVF will be used to multi-slice segmentation in semi-automatic of CT-scan image of pelvis. multi-slice segmentation can be automatically if the pelvis bone structure image from one slice to next slice have relative simple structure. At the pelvis bone structure which relative more complex, back adjusment parameter every related slice still be needed to increase accuration from image segmentation. Output from multi- slice segmentation of pelvis image then be volumetric visualizing process. From volumetric visualizing pelvis image obtained, showing segmentation technique proposed to have better detail when compared by conventional segmentation method. The segmentation method proposed also earn to lessen work load from operator if done by image segmentation to every slice in manual.