Medical Image Processing and Analysis
Medical images were used for non-invasively inspecting the diseases in the structures and functions of human body and brain. Based on the processing, analysis, and visualization of the medical images, we can obtain the disease information to provide proper treatment. There are many different-type medical images; such as, ultrasound images (Doppler angiography and 3D power Doppler angiography, CT (spiral CT), MRI (fMRI and diffusion MRI), PET, SPECT, etc. Different-type images have different properties; however, we always need to process these images; such as, contrast enhancement, noise remove, feature extraction, segmentation, image and model fusion, unmixed analysis, to explore more information for treatment. The planning topics contain: (1) image contrast enhancement based on the wavelet transform and Teager's operator; (2) using wavelet transform and hidden Markov model to reduce the additive noises on ultrasound, spiral CT, and MRI images, and the multiplicative noises on PET and SPECT images; (3) the detection of multiresolution edges using wavelet transform and tracking; (4) using active contour model to extract the contours of organ or tumor; (5) segmentation of homogeneous regions using wavelet transform and contextual hidden Markov tree model; (6) image and 3D model fusion of functional images and structural images using wavelet transform and principal component analysis; and (7) using linear unmixed analysis method to extract the components of tissues from CT, MRI (T1, T2, and PD), fMRI, PET, and SPECT images.
Resluts: enhancement, denoise, contour extraction, edge detection, segmentation
Members: M.-I. Shiu, I.-L. Chung, T.-F. Liu, I.-R. Lee
(a) (b) (c)
The image processing results. (a) The original VHD image. (b) segmented result. (c) extracted contours.