M.A.-R. Abdou and M.B. Tayel
Neural network, difference fuzzy, EZW, ROI segmentation, progres- sive transmission
This paper introduces an automatic bi-channel compression tech- nique for ROI segmentation and medical image (MI) compression. A novel ROI segmentation technique is presented. This technique uses an introduced artificial neural network (ANN) and an introduced difference fuzzy model (IDFM), obtaining irregular spider hexagon ROI contours. The whole medical image is to be transmitted pro- gressively using the fast algorithm for embedded zerotree wavelet (FEZW) [1]. Different refinement levels are applied to different MI regions. High compression ratios are obtained outside ROI, and a compromise between compression ratio and image quality is to be maintained by choosing a suitable threshold level inside the ROI. The proposed work reduces complexity and storage space, saves time, and has the advantage over previous works that it is fully automatic. Several brain magnetic resonance imaging (MRI) and fluorescene ophthalmic images are analysed; results are compared with other techniques to validate the proposed work.
Important Links:
Go Back
