| Summary: | 碩士 === 國立成功大學 === 資訊工程學系 === 86 === The rapid advancement in various medical imaging modalities
provides tremendous helps in clinical disease diagnosis. On the
diagnosis of pediatric renal diseases, the physicians need to
evaluate the diseased kidney both from the structural appearance
and the functional map of the organ. Therefore, ultrasonic
imaging and single photon emission computer tomography (SPECT)
are commonly used to provide the structural and functional
images.From the two types of images, we need to develop a system
which can first segment the organ to obtain the volume and
surface data, then can integrate the structural and functional
information from the two image modalities based on some image
registration techniques. In the proposed system, we first
segment the ultrasonic images by MAP(maximum a posteriori
estimation) methods, and then the SPECT images by the moment-
preserving thresholding method to measure the volume of organs
and to visualize the 3-D kidney. Secondly, the segmented data is
registered to obtain geometric correspondence. A two-step
registration method is employed. The coarse registration is
obtained by the PAX (Principal Axis Transformation). And the
resulting data are fine-tuned through the SFIT (Surface Fitting)
method. The segmentation and registration of medical imaging
offer us an accurate and fast way in integrating multi-modality
medical information for pediatric renal disease diagnosis. It is
of great value in scientific research as well as good concern
with people's welfare.
|
|---|
