Medical Image Fusion using Lifting based Transform Techniques

Medical image fusion is the means to obtain an enhanced image with better quality and reduced redundancy by fusion of two or more images that may or may not belong to the same modalities. The imaging modalities such as Magnetic Imaging Resonance (MRI), Computed Tomography (CT), Positron Emission Tomography (PET), Single Positron Emission Tomography (SPECT) etc have found their application in diagnosis in the medical domain. The paper presents the use of these imaging modalities pairs (MRI-T1/T2; MRI-PET/ SPECT) to perform medical fusion in transform domain. The paper presents two approaches. 1) It uses the lifting step-based wavelet transform (LWT), a better algorithm that takes significantly less computational time, does distortion-free reconstruction for the Average-Max, Edge-Energy fusion rules. 2) It proposes the new fusion rule to obtain an improved fused image. The fusion rule is applied using the DWT and LWT, and comparisons are made accordingly on the image pairs. Entropy, Peak Signal to Noise Ratio (PSNR), and average gradient are used as performance metrics to assess the image quality.