MRI Image Fusion Based on Sparse Representation with Measurement of Patch-Based Multiple Salient Features

Multimodal medical image fusion is a fundamental, but challenging, problem in the fields of brain science research and disease diagnosis, as it challenging for sparse representation (SR)-based to characterize activity levels with single measurement not lose effective information. In this study, Kronecker-criterion-based SR framework was applied patch-based level, integrating salient features multiple domains. Inspired by formation process vision systems, spatial saliency characterized textural contrast (TC), composed luminance orientation contrasts, promote participation more highlighted information process. As substitute conventional l1-norm-based saliency, sum (SSSF) used metric promoting significant coefficients composition level measurement. The designed verified be conducive maintaining integrity sharpness detailed Various experiments on groups clinical images effectiveness proposed method terms both visual quality objective assessment. Furthermore, study will helpful further detection segmentation images.