Adaptive multistage nonlinear filtering and wavelet for medical image enhancement

摘要

A novel hybrid filter architecture is proposed that incorporates an adaptive multistage nonlinear filter (AMNF) and a tree-structured wavelet transform (TSWT) specifically designed for image enhancement in medical imaging using different sensors. The AMNF is used for noise suppression and image enhancement. The TSWT is used for multiresolution decomposition and reconstruction of the medical images where selective reconstruction of subimages is used to perform further enhancement of specific clinical features. The AMNF-TSWT filter produced enhanced images for different modalities namely, digital mammography, chest, skin cancer and retinal images, that provided improved visualization of specific clinical features compared to the raw digital images. The AMNF-TSWT filter should prove to be useful for the use of neural networks for feature classification using different sensors and imaging protocols.